And their Q2 2026 report just confirms how good the business is, and how it keeps growing at unprecedented levels, even at its scale.

Just look at these numbers, the total revenues were up +24%, with a 17% increase in Search revenue, and 82% (!) increase in Google Cloud (With an insane, $514 billion backlog), and a +13% increase in YouTube Ads revenue.

The story of the quarter is the incredible growth of Google Cloud, but before we get into that, lets look at the other business segments.

Subscribe now

Google Services ($94.5 billion, +15%)

Consisting of Search that grew 17% to $63.3 billion, YouTube ads that grew 13%, to $11.1 billion, Google Network that contracted by 1%, and Subscriptions that grew 15% to $12.9 billion.

Search and Subscriptions posted a strong quarter, but we see YouTube ads and Google Network struggle more. Of course this is all overshadowed by Cloud and CapEx this quarter, but something I’m keeping an eye on moving forward.

This was said about the Subscriptions segment on the earnings call on what drove the growth:

“Due to strong growth in both YouTube subscriptions... and Google One, which was driven by demand for AI plans”

A top line growth of +15% with expanding operating margins from 40.1% to 41.8% is an excellent quarter in my book.

Google Cloud growing +82% YoY

Now, lets get into the headline story of the quarter.

Cloud is firing on all cylinders.

Not only is demand high, but this is the first quarter where Alphabet recognized revenue from TPU system sales delivered to data centres.

This is interesting, because it works as a second growth leg for Google Cloud, here is what management had to say about it:

“…we continue to expect to recognize a relatively small portion of the revenues from our existing TPU system sales agreements this year, ramping as we exit 2026... the vast majority of the revenues from these agreements will be realized in 2027.”

The TPU sales will hit the P&L in 2027, which is very interesting and could be a huge growth driver for Alphabet.

However, this is unrealized and in the future. What is true right now, is that Cloud just posted its best quarter ever.

Revenues almost doubled from $13.6 billion to $24.7 billion. Operating margins expanded significantly from 20.7% in Q2 25 to 35.5% in Q2 26 (Just insane).

When I first saw these numbers, I thought it was a blowout quarter, and that the stock price would boast upwards.

But of course, there is more to the story that concerned analysts and investors.

Anat, Alphabet’s CFO, was direct on the earnings call about Cloud margin risks:

“Given the supply-constrained environment, we plan to expand the use of third-party capacity in Q3 as a bridging strategy... it will create modest margin pressure in the near term.”

What I’m reading from this, is that the strong Q2 margins is a high mark, and not necessarily the baseline.

Alphabet will be renting outside compute to keep serving demand it can’t (yet) fill internally.

This is how Sundar looks at this strategic move:

“a short-term cost over a few months may be very high, in the lifetime of the deal... highly ROI positive.”

This is a classic short-term pain for long-term gain play, but the market will always punish a business for this, as the long-term gain is uncertain.

Subscribe now

Capital Expenditure reached $44.9 billion

Another reason why the market is a bit shaky in Alphabet (And the rest of the hyperscalers), is their insane CapEx plans.

The prevailing question is: Will these hundreds of billions yield a good return on investment?

The management team expressed a growing confidence in the investments:

“I think the dynamics look healthier than where we were about a year ago, that’s what gives us the confidence to undertake those investments.”

What I think is interesting is the added commentary on the backlog:

”we expect to recognize just over 50% of the total backlog as revenue over the next 24 months”

This basically gives us a sanity-check on the CapEx, because we can easily follow this statement up in the coming earnings report to see if Alphabet manages to do practice what they preach.

Earnings goes parabolic (+298% YoY)

Another story worth mentioning is one that many investors get wrong.

I’ve seen multiple investors claim that Alphabet is severely overvalued, and point to the earnings bloating from the most recent quarter.

It is true that the earnings is bloated this quarter, this is because Alphabet got $97.9 billion of its Net income from “Other income (expense), net” this quarter.

Why is that? Well, Alphabet owns between 4% and 6% of SpaceX. It paid $900 million for a 7.5% stake of SpaceX in 2015 (Which has since been diluted).

SpaceX went public this quarter, and much of the gains were realized in the income statement of Alphabet. Most of the $97.9 billion is from this increase in marketable securities.

So, not from Alphabets operations, but from an (incredibly good) investment they’ve made.

This is understood by serious investors, and we should not use the headline numbers to base our valuation on.

Even if you exclude the earnings from investment gains, this was a great quarter.

Income from operations grew from $31 billion to $40.7 billion (+30.3% YoY). A great result from Alphabet which reflects the business growth.

Negative Free Cash Flows

My final point will be on FCF.

Alphabet has been a cash generating machines for decades.

However, it has taken a deliberate choice to suppress FCF to invest heavily in infrastructure to support its next leg of growth.

You can disagree with the market on punishing stock for making this move, but the market does what it always do. It discounts risk.

There is a lot of uncertainty in the investments Alphabet is making, even if it makes logical sense, it is still unproven.

When (or if) Alphabet starts posting fantastic results (In terms of growth and market share), the stock price will follow. But until that proof is here, the market remains cautious.

Despite this, if we just look at the operations, Alphabet’s Net cash provided by operating activities increased 41% YoY.

However, the Free Cash Flow went negative (-$5.855 billion) in Q2, due to a heavy $44.9 billion CapEx investment.

Conclusion

A record breaking quarter for Alphabet, especially from the Google Cloud business segment.

Search and subscriptions are also holding up well and continue to grow rapidly, even at its massive scale (Remember a few years ago, when Search was deemed dead?).

The fantastic operational results were overshadowed by a more cautious market, looking at:

• Margin risk from Cloud (Due to borrowed capacity - remember, short-term pain, long-term gain).

• Capital expenditure uncertainty. Burning $44.9 billion in one quarter is no joke, and the market has not yet decided if this is a good investment.

• Negative free cash flows. It is deliberate, but it’s never a positive to see a cash generator post negative FCF numbers.

In addition, the incredible SpaceX investment has bloated the Net Income and EPS numbers. The result from this is that if we look at the current PE, it is 16.4x. This is of course without excluding the one-time effect from Alphabet’s equity investments.

The forward PE of 24.5x is more accurate to use, and it is still in line with its 10 year median forward PE:

That’s it for today! Leave your thoughts in the comments below, or reply to this email.

Subscribe now

Ready to take the next step? Here’s how I can help you grow your investing journey:

1. Go Premium — Unlock exclusive content and follow our market-beating Quality Growth portfolio. Learn more here.

2. Essentials of Quality Growth — Join over 300 investors who have built winning portfolios with this step-by-step guide to identifying top-quality compounders. Get the guide.

3. Free Valuation Cheat Sheet — Discover a simple, reliable way to value businesses and set your margin of safety. Download now.

4. Free Guide: How to Identify a Compounder — Learn the key traits of companies worth holding for the long term. Access it here.

5. Free Guide: How to Analyze Financial Statements — Master reading balance sheets, income statements, and cash flows. Start learning.

6. Get Featured — Promote yourself to over 24,000 active stock market investors with a 42% open rate. Reach out: investinassets20@gmail.com

Every software company that mentions “AI agents” on an earnings call gets sold off.

It doesn’t matter if the mention is a warning or a victory lap.

The market has decided that AI is coming for SaaS, and it’s pricing the entire sector with a shotgun instead of a scalpel.

The ETF that track tech-software (Not just SaaS), is down 21.61% from its 2025 highs, and was down as much as -40% in April.

AI is clearly disruptive to the industry - but what companies carry the actual risk?

Some of the names getting punished have a real, structural problem.

Others are being dragged down by a narrative that has nothing to do with their actual business.

And a handful are becoming stronger because of AI while trading like they’re becoming weaker.

That gap, between the story and the numbers, is where the opportunity lives.

The four questions that actually matter

Here’s a repeatable way to sort any software company by real AI exposure.

Is the moat the software, or the workflow around it? A thin layer sitting on top of a general-purpose model can be rebuilt by that model’s next version. A deep integration into a regulated process, a proprietary dataset, or a switching-cost-heavy workflow can’t.

Is pricing tied to headcount or to outcomes? Seat-based pricing is a direct bet that companies keep hiring the same number of people to do the same job. AI is explicitly designed to break that bet. Consumption-based or outcome-based pricing usually grows as AI usage grows, instead of shrinking.

Is there proprietary data AI can’t replicate? A company sitting on decades of validated, regulated, or otherwise hard-to-recreate data has something a generic model can’t just absorb. A company whose “data” is really just organized public information does not.

Is the buyer risk-averse and regulated, or a commodity back-office function? Compliance-heavy buyers move slowly and hate switching vendors. Commodity functions get automated first, because there’s no regulatory or safety reason not to.

Run any SaaS stock through those four questions and you get a much clearer picture than “AI is bad for software.”

Bucket 1: Companies at Risk

The ones getting hit hard

Chegg: the cautionary tale that already happened

Chegg is the clearest case study of AI disruption that exists, because it’s no longer a debate, it’s a finished story. The stock fell 48% in a single day back in May 2023, when then-CEO Dan Rosensweig told analysts ChatGPT was hurting new customer growth. That was the first time a public company admitted, on the record, that a generative AI tool was actively damaging its business.

It didn’t stop there. Chegg has lost over 500,000 subscribers since ChatGPT launched. Its Q1 2026 revenue fell 48% year over year to $63.3 million, with the core Academic Services segment down 57%. The company has laid off well over half its workforce across four rounds since mid-2024, and the stock has traded as low as roughly $1, down about 99% from its 2021 peak of over $113.

Run it through the framework: the moat was a searchable database of solved homework problems, exactly the kind of content a language model reproduces instantly and for free. Pricing was a flat monthly subscription with zero switching costs. There was no regulatory moat, no proprietary workflow, nothing standing between Chegg and a free alternative that got dramatically better overnight. This is what it looks like when every box in the framework is checked in the wrong direction.

Globant, and the IT-services squeeze

Globant is a harder case, because the company is executing reasonably well and the stock has still been destroyed. Shares are down roughly 60% over the past year on growing concern that AI will “disintermediate” custom software development.

Full-year 2026 revenue guidance implies growth of just 0.3% to 2.2%, essentially flat, for a company that used to compound at a much faster clip.

Management’s response has been to lean into “AI Pods,” teams that pair engineers with AI tooling, and to keep buying back stock.

Q1 2026 results actually beat guidance. But the market isn’t rewarding execution here, it’s repricing the entire IT-services model, because the mechanism is real: when an AI coding assistant does the work of several junior developers, the billable-hour model that funded this industry for two decades gets structurally smaller.

EPAM, Globant’s closest peer, tells a similar story: 2026 growth guidance was cut to 4-6.5%, and by at least one analysis its return on invested capital has fallen below its cost of capital in recent years, a sign of a business that’s having to fight harder for the same returns.

Neither company is going to zero. But the framework points to real structural pressure here, not a sentiment overreaction: seat-and-hour-based pricing, a commodifiable service (routine coding and QA work), and a buyer (enterprise IT budgets) that is actively looking for ways to spend less on exactly this.

Workday: caught in the middle, and still needing to prove it

Workday sits closer to the boundary. The stock is down 53.6% from its all time highs, triggered by conservative fiscal 2027 subscription guidance of 12-13% growth that fell short of consensus.

The mechanism is straightforward: Workday prices Human Capital Management (HCM) and payroll software per employee seat, and if AI reduces corporate headcount, the number of seats a company needs shrinks with it.

To be fair to Workday, its retention numbers are excellent: it serves roughly 65% of the Fortune 500 with a 97% gross retention rate, and displacement risk is low. But the seat-based pricing exposure is real, and unlike Salesforce below, Workday hasn’t yet shown hard evidence that its new consumption-based “Flex Credits” pricing is offsetting the risk. This is a name where the jury is still out, and the honest answer is that it belongs on a watchlist, not yet in a “buy the dip” pile.

5 More SaaS companies at risk:

• HubSpot

• DocuSign

• Fiverr

• Cognizant

• Concentrix

Bucket 2: Insulated Companies

Built to survive

Veeva Systems

Veeva sells cloud software to life sciences companies for FDA-regulated processes: clinical trials, drug safety, regulatory submissions.

That’s about as far from a “thin AI wrapper” as software gets.

Fiscal 2026 revenue came in at $3.2 billion, up 16%, with subscription revenue up 17%, and the company still put up double-digit growth while building agentic tools (Veeva Falcon, launching for early access in November 2026) directly into those regulated workflows rather than trying to replace them.

Worth noting: Veeva wasn’t immune to the sector-wide fear either. Shares were down about 16% year-to-date heading into its Q4 report before jumping more than 10% on the earnings beat.

That’s the pattern this whole article is about: good business, dragged down by a generic narrative, snapping back once the numbers forced a correction.

Run it through the framework and the reason is obvious: regulated buyer, proprietary validated data, deep workflow lock-in. None of that goes away because a model gets smarter.

Veeva is down 35.7% from its 2025 all time highs:

Constellation Software

Constellation is close to the opposite of a thin AI wrapper by design.

It’s a collection of hundreds of decentralized, deeply niche vertical market software businesses, the kind of software that runs a specific municipal utility’s billing system or a specific type of clinic’s scheduling software.

These products are often the single system of record for a small, specific customer base that has no appetite to switch, and often no real alternative to switch to.

That’s precisely the profile the framework says should survive: the moat isn’t the software’s cleverness, it’s decades of embedded, mission-critical use inside a workflow with high switching costs and a narrow, non-generic problem to solve.

A general-purpose AI model doesn’t casually rebuild hundreds of tiny, deeply specific vertical systems, and Constellation’s capital allocation discipline means it can keep buying more of them regardless of the AI narrative swirling around big-cap SaaS.

Despite the robustness of Constellation’s portfolio, the stock is down -46.4% from its all time highs:

5 More companies that are insulated:

• Roper Technologies

• Tyler Technologies

• Guidewire Software

• SS&C Technologies

• Fair Isaac (FICO)

Bucket 3: the gainers (the bucket that actually matters)

This is the group I think is most mispriced. These are companies where AI is a demonstrated tailwind, not a threat, but where sector-wide fear has still pulled the stock down.

Salesforce: same panic, opposite data

Salesforce and Workday got hit by the identical narrative, the “Seat-Count Crisis,” at the same time.

Salesforce shares fell as much as 54.1% from 2025 highs:

But look at what actually happened inside the business instead of the narrative around it.

Agentforce, Salesforce’s AI agent platform, grew annual recurring revenue from $800 million to $1.2 billion in a single quarter, up 205% year over year.

And here’s the detail that directly contradicts the bear case: seven of Salesforce’s ten largest deals in that quarter added seats.

Management is explicitly repricing around usage instead of defending the old seat model, shifting toward metered, consumption-based billing for agentic work through a platform called m3ter.

The stock was trading at 12x forward earnings with 13% revenue growth and expanding margins as of this writing. This is a valuation that assumes the bear case is already true.

Same fear, same sector, and different underlying data than Workday. That’s the kind of distinction the four-question framework is built to catch, and exactly why treating “seat-based SaaS” as one undifferentiated bucket is a mistake.

Alphabet: the “AI kills Google” thesis, disproven by Google’s own numbers

Most investors have most likely already forgotten that stock was trading sub $150 in April of 2025 (Now trading at $354).

The bear case and narrative on Alphabet was: “ChatGPT kills Search”.

This narrative had slowly built since November 2022, but in 2025 the narrative hit the stock price hard.

This is maybe the purest example of sector-wide fear versus company-specific reality.

The Q1 2026 numbers say the opposite happened: Google Search revenue hit $60.4 billion, up 19% year over year, with management citing search queries at an all-time high.

Google Cloud grew 63% to $20 billion, with a contracted backlog that’s nearly doubled to over $460 billion.

Alphabet still has real risks worth mentioning: capital expenditure is enormous (2026 guidance of $180-190 billion), free cash flow has compressed sharply during the build-out, and the AI model race with OpenAI and Anthropic is genuinely competitive, not a foregone conclusion.

But the specific mechanism the market feared, generative AI directly cannibalizing search traffic and ad revenue, has not shown up in the data.

Google folded AI Overviews directly into the product instead of losing users to a separate one, and it’s monetizing at a similar rate to classic search.

Synopsys: benefiting from the boom, priced like a casualty

Here’s a genuinely odd one in my opinion.

Synopsys, half of the EDA software duopoly (alongside Cadence) that every advanced chip design runs through before reaching a fab, reported Q2 fiscal 2026 revenue of $2.28 billion, up 42% year over year, directly riding the AI chip design boom.

And the stock was still down roughly 20.8% year-to-date and 41% from its all time highs.

At the same time Cadence, its closest peer with a similar growth story, was up around 19-21%.

Some of that gap is company-specific, Synopsys is digesting a roughly $35 billion Ansys acquisition, which creates real integration risk worth taking seriously rather than dismissing.

But a chunk of the underperformance looks like collateral damage from the broader “AI kills software” narrative bleeding into a company whose entire business exists because of the AI chip buildout.

Switching costs in EDA are about as extreme as software gets: engineers spend years learning a specific tool flow, and Synopsys and Cadence together dominate the market almost by default.

The data infrastructure “Fab Five”

Bank of America has grouped Snowflake, Datadog, MongoDB, JFrog, and Twilio as a basket that’s up roughly 30% on average in 2026, while the broader software sector ETF (IGV) was down about 12% over the same stretch.

The logic is simple: every AI model in production needs somewhere to store data, something to monitor performance and cost, and infrastructure to run on, and none of that shrinks when AI gets more capable, it grows.

Datadog is the clearest example: Q1 2026 revenue crossed $1 billion for the first time, up 32% year over year, with AI-linked revenue now over 10% of the total and growing faster than 100% annually.

The stock jumped over 30% in a single session on that report and pulled Snowflake and MongoDB up alongside it.

Snowflake’s product revenue grew 30% in its most recent quarter, directly tied to the volume of data enterprises are processing for AI workloads.

Worth being honest about the other side: this group isn’t cheap. Snowflake in particular has traded at 120-140 times forward earnings at various points this year, which leaves very little room for a slowdown in enterprise data spending.

But the mechanism: more AI usage requiring more data infrastructure, is the opposite of the seat-cannibalization risk facing Chegg or Globant.

5 More companies that are AI gainers:

• Palo Alto Networks

• Arista Networks

• CrowdStrike

• ServiceNow

• Oracle

How to run this yourself

Before buying (or avoiding) any SaaS name on an AI narrative, ask:

1. Does the moat live in the software itself, or in the data, workflow, and switching costs around it?

2. Is revenue tied to headcount, or to usage and outcomes?

3. Would a generic AI model actually replace what this company sells, or does it need this company’s specific data and workflow to be useful at all?

4. Is the stock down because the business is actually impaired, or because it shares a sector label with companies that are?

That last question is the one the market keeps ignoring.

The takeaway

AI disruption in software is real, Chegg proved that beyond argument, and the IT-services squeeze on Globant and EPAM is a live, structural story happening now.

But the market is currently pricing entire categories, seat-based software, anything labeled “SaaS,” as if they share Chegg’s problem. But many don’t.

The actual opportunity isn’t avoiding AI risk, it’s owning the companies where AI is making the business stronger while the stock still trades like it’s a victim.

Salesforce with Agentforce actually adding seats, Alphabet with search revenue accelerating, Synopsys riding a chip design boom while priced like a laggard, a data infrastructure basket growing because AI needs somewhere to run.

The panic is still real, but identifying the quality SaaS businesses that have sold off is where the opportunity lies in the current market.

I hope you enjoyed the article.

Ready to take the next step? Here’s how I can help you grow your investing journey:

1. Go Premium — Unlock exclusive content and follow our market-beating Quality Growth portfolio. Learn more here.

2. Essentials of Quality Growth — Join over 300 investors who have built winning portfolios with this step-by-step guide to identifying top-quality compounders. Get the guide.

3. Free Valuation Cheat Sheet — Discover a simple, reliable way to value businesses and set your margin of safety. Download now.

4. Free Guide: How to Identify a Compounder — Learn the key traits of companies worth holding for the long term. Access it here.

5. Free Guide: How to Analyze Financial Statements — Master reading balance sheets, income statements, and cash flows. Start learning.

6. Get Featured — Promote yourself to over 24,000 active stock market investors with a 42% open rate. Reach out: investinassets20@gmail.com

Some of the capital freed up from trimming Alphabet and exiting ASML and LVMH is going here first. I’m adding to Taiwan Semiconductor, and unlike ASML, this is a bet on business momentum, not on a multiple that needs to keep expanding.

The numbers underneath the stock price

TSMC trades at 18.8x forward earnings against a long-term expected…

Read more

I recently freed up ~15% of my portfolio’s capital to allocate to better opportunities.

I’m opening a ~5% position in Uber, on the view that the market’s biggest fear about this business, autonomous vehicles cutting it out entirely, is overblown and will eventually benefit Uber.

Read our recent analysis of Uber:

Why the AV disruption thesis…

Read more

As I’ve communicated recently, I’m making some changes to the portfolio. It’s time to free up capital and reallocate it to emerging compounders that will accelerate my compounding rate.

This is my current allocation:

Let’s get into the details.

Trimming Alphabet by roughly a third

Let’s start with the one that isn’t a conviction call. I’m n…

Read more

I used to spend most of my investing time on work that didn’t require me.

Reading through earnings releases line by line. Scrolling through 300-page annual reports hunting for the three paragraphs that actually mattered. Manually comparing numbers across quarters to spot a trend I could have seen in 30 seconds with the right tool. Summarizing earnings call transcripts just to arrive at a conclusion I could have reached faster.

That was the job. And most of it was grunt work.

I still do all of those things. I just don’t spend 10 hours a week doing them anymore.

This article is about exactly what changed, the specific workflows, the exact prompts, and the places where AI will mislead you if you’re not paying attention.

Read my previous article on how to set up Claude as your personal AI investment analyst here:

Let’s get into it 👇

First: What those 10 hours actually look like

If you run a concentrated portfolio of quality compounders seriously, meaning you actually read filings, track your thesis per holding, and think carefully before you act, your weekly research time probably breaks down something like this:

• Earnings processing (releases, transcripts, filings): 3–4 hours

• Annual report deep dives on new or existing positions: 2–3 hours

• Monitoring existing holdings across quarters: 1–2 hours

• Research on new ideas: 2–4 hours

• Cutting through macro and sector noise: 1 hour

That’s 8–14 hours per week for someone doing this properly. Most of it is preparation: gathering, formatting, summarizing, comparing. Almost none of it is the actual thinking that drives investment returns.

The goal isn’t to do less. The goal is to spend those hours on judgment instead of formatting.

The mindset shift that makes this work

Before we get into the workflows, this needs to be said clearly.

Claude is not an analyst. It has no skin in the game, no track record, and no judgment built from watching businesses succeed and fail over decades. It has read everything and experienced nothing. It will give you a confident-sounding answer whether it knows what it’s talking about or not.

What it is (when used correctly) is the fastest research preparation tool that has ever existed for individual investors.

Think of it this way. The 10 hours you’re about to reclaim are hours that never required your judgment in the first place. What remains is the conviction calls, the thesis judgments, the decision to hold through a painful quarter or sell when something has broken in the business, those still take exactly as long as they should. AI should not touch those decisions.

The job of AI in your investment process is to handle the preparation so you can spend more time on the thinking. Keep that distinction sharp at all times, and this becomes enormously useful. Blur it, and you’ll make worse decisions than you did before.

Caveat: There is a value to reading all the documents yourself, you might find something that AI misses. So let me be clear, the best possible scenario is that you do everything yourself. But let’s be real, very few have the time, the discipline, and the fortitude to do that consistently over time. I prefer to spend my energy and cognition where it matters most: Judgement around specific investment decisions.

Workflow 1: Processing earnings in 20 minutes instead of 2 hours

Saves approximately 2–3 hours per week during earnings season

Earnings season is the most time-intensive period in any fundamental investor’s calendar. You own 12-20 positions. Each one reports. Each report has a press release, a set of financials, and a +60-minute earnings call transcript. Doing this properly, the old way, takes hours.

Here’s the new way.

What you do: Paste the documents of the earnings press release and the full earnings call transcript into Claude. Before you do, paste your investment thesis for that holding, the reasons you own it and the specific metrics you’re watching. Then run this prompt:

What this looks like in practice

When Constellation Software reported Q1 2026, I pasted my thesis, which centers on their ability to deploy capital at high returns through vertical market software acquisitions, sustain mid-single-digit organic growth, and grow free cash flow per share over time by +15% annually alongside the release and transcript.

Claude flagged immediately that organic recurring revenue growth had decelerated to approximately 4%, that management used notably more cautious language around M&A multiples than in prior calls, but that free cash flow available to shareholders had jumped 44% and the acquisition pipeline commentary remained constructive. It surfaced exactly the tension I needed to think about, in about four minutes.

What you still do yourself

Decide what the verdict means for your position. Claude told me organic growth was decelerating. It cannot tell me whether 4% organic growth at Constellation Software is a problem or a feature of their model at scale. That judgment is mine, built from years of following the business.

Looking at the data since 2017, a 4% organic growth is actually in the higher range of what Constellation has delivered:

Where this goes wrong

Claude will sometimes misread numbers from a pasted document, particularly when tables are involved. Always verify the specific figures it pulls against the actual press release. Never trust a number Claude gives you without checking the source.

Subscribe now

Workflow 2: Annual report deep dives in under an hour

Saves approximately 2 hours per company

The annual report is the most important document a public company produces. It is also, for most companies, 80% repetitive. Risk factor sections that haven’t changed in four years. Legal disclosures. Compensation tables. Pages of accounting policy notes that apply to almost no investment decision you will ever make.

The skill in reading an annual report is knowing what to look for. Claude doesn’t develop that skill for you, but once you have it, Claude can find the relevant sections faster than any human.

What you do: Upload the annual report PDF directly into Claude, or paste the sections most relevant to your analysis. Then run this prompt:

What this looks like in practice

When I was going deeper on MercadoLibre ahead of their Q1 2026 results, I used this prompt against their most recent 20-F. Claude pulled out the specific language management used around their fintech flywheel, the way Mercado Pago and the credit business reinforce the commerce platform, and flagged that the risk factor language around credit portfolio quality had become meaningfully more specific than in the prior year’s filing. That was a signal worth investigating further. It would have taken me two hours to find it manually. It took Claude four minutes.

What you still do yourself

Evaluate whether what Claude found is good or bad. It can tell you that credit risk language became more specific. It cannot tell you whether that reflects prudent disclosure or a genuine deterioration in the underlying portfolio. That requires your own judgment, your own reading of the fintech lending environment, and your own view of management’s track record.

Where this goes wrong

Two places. First, Claude sometimes misreads PDFs, particularly tables, footnotes, and anything in a non-standard format. Always check the numbers it extracts against the source document.

Second, and more importantly: Claude has a bias toward sounding thorough. It will give you a confident summary even when the source material is ambiguous or the answer isn’t clear. If you ask it whether the competitive position is strong, it will give you a structured answer. That answer may or may not reflect reality. The confidence of the output is not evidence of its accuracy. Read the sections that matter yourself. Use Claude to find them faster.

Workflow 3: Monitoring your existing holdings without drowning

Saves approximately 1–2 hours per week

This is the part of investing that almost no one covers, and the part where most individual investors are genuinely weakest. Not finding stocks. Holding them.

You own a portfolio of stocks. Each one has a thesis. Each one reports every 90 days. The question every single quarter is the same: did anything change? And it’s hard to answer consistently, because your memory of why you bought something three years ago is imperfect, and the quarterly noise makes it easy to drift from your original reasoning without noticing.

The solution is a thesis document for each holding that you update quarterly. Claude makes this both easier to build and faster to use.

Step one: Build the thesis document. For each holding, run this prompt:

Save that checklist. Then every quarter, after earnings, run this:

What this looks like in practice

My thesis on Kinsale Capital rests on three things: their ability to underwrite non-standard E&S risks at a combined ratio well below the industry, their cost advantage relative to larger competitors, and premium growth that reflects both market share gains and pricing power. When Q1 2026 showed commercial property premiums falling 28% on heightened competition, Claude flagged it immediately as a contradiction of the market share assumption, while also noting the combined ratio had improved to 77.4% and reserve development remained favorable, both confirmations of the underwriting discipline assumption. The verdict: THESIS WEAKENED — WATCH. That’s exactly right. Not broken. But worth monitoring closely.

Where this goes wrong

The thesis checklist is only as good as the thesis you give Claude to start with. If your original investment notes are vague, like “Kinsale is a great compounder with a good management team”, the checklist will be vague too. Garbage in, garbage out. The discipline of writing a specific, falsifiable thesis for each holding is the work. Claude just helps you track it.

Workflow 4: First-pass screening on a new idea in 30 minutes

Saves approximately 1–2 hours per new idea

Every serious investor has the same problem with new ideas: there are more interesting businesses than there is time to investigate them properly. The first-pass filter, is this worth two weeks of serious research? Is itself time-consuming.

Claude helps with the workload.

Where this goes wrong, and this is important

This is the workflow where AI is most dangerous. Claude will give a confident assessment of a business’s competitive advantages based entirely on the information you give it and its training data. It has no ability to assess whether a moat is real in practice, whether customers actually stay, whether pricing power holds under pressure, whether the culture that drove the returns is still intact. A confident “COMPELLING” verdict from Claude on a new idea means almost nothing without your own independent judgment. Use this as a filter, not a conclusion. The serious work still starts after this prompt.

Workflow 5: Cutting through macro and sector noise

Saves approximately 1 hour per week

The financial internet produces infinite content. Most of it is on the interest rate takes, the tariff commentary, the recession probability estimates. This will often never matter for the long-term value of the businesses you own. The trap is reading all of it anyway.

When something from the macro might matter for a specific holding, use Claude to assess the actual business-level impact quickly, rather than reading twelve articles that all say the same thing.

What this looks like in practice

When tariff concerns were dominating financial media in early 2026, I ran this prompt for LVMH. Claude quickly surfaced that Fashion & Leather Goods had direct exposure to transatlantic goods flows, that China recovery remained the bigger driver of the thesis, and that management’s Q1 commentary had been notably vague on the North America impact. That was a useful 10-minute framing that let me stop reading tariff articles and focus on the actual questions that mattered for the position.

What the reclaimed time is actually for

Saving 10 hours a week sounds like the goal, but its not.

The goal is to spend those 10 hours on the work that actually drives investment returns over decades. The reading that builds pattern recognition, the thinking that develops conviction, the uncomfortable questions about what you might be wrong about in your highest-conviction positions.

The investors who compound wealth seriously over a long period aren’t faster at grunt work. They spend more time thinking clearly about fewer, better questions. AI just gets you there faster.

What you should never use AI for

This section matters as much as everything above it.

Never use AI to decide whether to buy or sell. Claude will give you a structured answer to any question you ask, including “should I add to this position?” That answer is worthless. It has no skin in the game, no understanding of your portfolio construction, no memory of what you paid, and no ability to weigh qualitative factors that only come from years of following a business. The buy and sell decisions are yours. Permanently.

Never trust a number Claude gives you without checking the source. Claude fabricates figures with complete confidence. It will cite a ROIC of 23% for a company whose actual ROIC is 14%. It will pull a revenue figure from the wrong quarter. It will confuse operating income with net income. Every single number that matters (every number you might act on) needs to be verified against the primary source.

Never use AI as a substitute for reading the filings yourself. Claude’s summaries flatten nuance. An earnings call transcript has texture, the pause before an answer, the question that gets deflected, the topic that the CEO rushes past. You only catch that if you’ve read enough transcripts to know what normal sounds like. Claude has never read a transcript with skin in the game. You have. Use Claude to find the sections worth reading carefully. Then read them carefully.

Never use AI on questions where the answer requires judgment about the future. “Will Novo Nordisk maintain its competitive position in GLP-1s?” is not a question Claude can answer. It will give you a confident framework and a balanced conclusion. That conclusion is constructed from patterns in its training data, not from any genuine understanding of pharmaceutical competitive dynamics, patent cliffs, or the likelihood that oral Wegovy changes the category. On questions of qualitative judgment about the future, Claude is confidently mediocre. Your own reading, thinking, and pattern recognition (Even if its not perfect) is more valuable.

Never let AI make you feel like you’ve done the work when you haven’t. This is the most subtle and dangerous failure mode. Running five prompts and getting five structured outputs feels like research, but its not. It’s “organized preparation”. The research is what happens when you sit with the questions those outputs raise and think hard about what you actually believe.

Final Thoughts

Investing well is a craft. It rewards patience, judgment, and the ability to think clearly about businesses over long periods of time. None of that can be automated. None of it should be.

But the hours you spend reformatting tables, skimming risk factors, and manually comparing numbers across eight quarters? Those hours were never doing the work. They were just the cost of accessing the work.

AI eliminates that cost. What you do with the time it gives back is still entirely up to you.

The prompts above are a starting point, not a system. Your system will develop as you use them, refine them, and figure out where they work for your process and where they don’t. Start with the earnings processing prompt this quarter. Build from there.

Ten hours is a lot of thinking time to give back to yourself every week. Use it well.

The prompts from this article, ready to copy:

• Prompt 1: Earnings Processing

• Prompt 2: Annual Report Deep Dive

• Prompt 3A: Extract My Core Thesis Assumptions

• Prompt 3B: Quarterly Thesis Check

• Prompt 4: Cold-Start Business Assessment

• Prompt 5: Business-Level Macro Impact Assessment

(Save these to a text file and customise them to your own portfolio and investing philosophy before your first use.)

Ready to take the next step? Here’s how I can help you grow your investing journey:

1. Go Premium — Unlock exclusive content and follow our market-beating Quality Growth portfolio. Learn more here.

2. Essentials of Quality Growth — Join over 300 investors who have built winning portfolios with this step-by-step guide to identifying top-quality compounders. Get the guide.

3. Free Valuation Cheat Sheet — Discover a simple, reliable way to value businesses and set your margin of safety. Download now.

4. Free Guide: How to Identify a Compounder — Learn the key traits of companies worth holding for the long term. Access it here.

5. Free Guide: How to Analyze Financial Statements — Master reading balance sheets, income statements, and cash flows. Start learning.

6. Get Featured — Promote yourself to over 24,000 active stock market investors with a 42% open rate. Reach out: investinassets20@gmail.com

June was a solid month for the Quality Growth Portfolio. Positive development in 3 of the portfolio’s ‘laggards’ and FX tailwinds sent the portfolio up 5.21%, marking a solid return since the March sell off.

Kinsale Capital, Novo Nordisk, and Mastercard have been a drag on the portfolio for most of the past year. Not because the underly…

Read more

Welcome to the June edition of Top 5 Buys ✅

You can access our Top 25 Buys for 2026 list as a premium member here.

Top 25 Buys

In this article, we will discuss our top stock picks for June 2026.

Let’s get into it 👇

The Market Sentiment: Fear

A month ago, the market was flirting with Greed, then the Strait of Hormuz situation escalated, the oil price spiked, and the Fear & Greed Index dropped to Fear territory at 35:

This is the overarching trend of 2026 so far. The fear and greed cycle on steroids. Tariff scares in April, recovery in May, geopolitical shock in June, and whats next?

Every few weeks the market finds a new reason to panic, sells off, then mostly forgets about it a few weeks later.

I don’t trade the index. But I do pay attention to what it tells me: when fear shows up, prices move before fundamentals do. That gap is where buying opportunities exist.

This month’s five picks all share something in common. Each one reported strong fundamental momentum recently, and each one has seen its share price disconnect from that momentum, for reasons that have little to do with the actual business.

Here are this month’s Top 5 Buys 👇

This is not investment advice. Always conduct your own due diligence and make your own investment decisions.

Top 5 Quality Buys June 2026 🚀

Meta Platforms 📱

The preferred digital advertisements platform is building its own AI infrastrucutre.

Social Media & AI Infrastructure

Meta just delivered one of its strongest quarters in years, and the stock is still down roughly 17% over the past 52 weeks. Q1 2026 revenue grew 33% year-over-year to $56.3 billion, beating estimates. Net income reached $26.7 billion. And the market’s response has been to focus almost entirely on the capex number.

Meta guided 2026 capital expenditure up to a range of $125 billion to $145 billion to fund AI infrastructure and data centers. That’s a lot of money.

But despite the result from the AI investment, the fundamental business of Meta is throwing off enormous free cash flow from an advertising engine that is getting more efficient (And more effective with the use of AI).

The market is pricing Meta like an AI cost center. The Q1 numbers say it’s still primarily an advertising compounder that happens to be investing heavily in its next potential leg of growth.

The Model: Distribution First, Monetization Second

Meta’s core business has always followed the same playbook: build the platform people spend the most time on, then monetize that attention better than anyone else.

Facebook, Instagram, WhatsApp, and Threads collectively reach close to half the world’s population. That distribution is the moat, and it creates a lot of optionality and new potential revenue streams for Meta.

AI is simply the newest lever for squeezing more value out of it; better ad targeting, better ranking algorithms, and now entirely new product surfaces like AI-powered Business Agents inside WhatsApp, Instagram, and Messenger.

Growth Drivers

Ad efficiency from AI: Better targeting and creative generation tools are lifting advertiser ROI, which supports continued pricing power even as impression growth moderates.

According to Meta, “For every $1 advertisers spent using Advantage+ shopping campaigns, they saw 17% more purchases compared to advertisers who used manual shopping campaigns.” (Source: Meta for Business)

Business Agents: A global AI-powered agent now operates across WhatsApp, Instagram, and Messenger, aimed at small and medium businesses that previously couldn’t afford a dedicated marketing or customer service function. This is a new revenue stream for Meta.

Instagram Plus and AI subscriptions: Meta is testing premium, ad-light tiers, an early step toward diversifying revenue beyond pure advertising. Meta expects a 1-2% conversion rate on Instagram and Facebook plus. The two platforms have 2.35 billion and 3.07 billion monthly active users.

This growth comes at very low additional operational costs, as Meta already has the infrastructure in place, they just monetize the distribution they’ve already built.

The Numbers

Meta trades at a forward P/E of 16.76x, a meaningful discount to where it traded for most of the past two years, despite revenue growth accelerating.

Return on invested capital sits at 24.3%, exceptional for a company spending this aggressively on infrastructure.

The PEG ratio is 0.8, which tells you the market isn’t pricing in much credit for growth at all right now. That’s unusual for a business compounding earnings the way Meta currently is.

Bottom line: Why now?

Meta is a high-quality compounder trading at a discount because the market can’t decide whether to treat the AI buildout as an opportunity or a threat to margins. The Q1 numbers already answered that question: revenue accelerated, margins held up, and entirely new monetization surfaces are opening up. Patient investors who can tolerate capex headlines are being offered a re-rating opportunity here.

I’m not sure about the AI investment for Meta, I believe it’s the wrong strategic bet for the business. But, as we’ve seen in the past, even poor capital allocation decisions (Like the Metaverse, buybacks at aggressive all time highs and so on) the digital advertisement engine of Meta can’t be held down. It is a pristine business, and all marketers I know always go back to spending more on Meta ads as opposed to other digital ad services.

S&P Global SPGI 🏛️

S&P Global finally trading at a reasonable price.

Financial Data, Ratings & Indices

S&P Global just posted a 10.4% revenue increase and a 28% jump in net income, and the stock is trading at one of its lowest valuations in over a decade. If you’ve ever wondered what a wide-moat business mispriced by macro noise looks like, this is a prime example.

The Model: Three Toll Booths

S&P Global sells judgment, data, and benchmarks that the entire financial system relies on to function.

The Ratings division prices the creditworthiness of nearly every major bond issued globally; you cannot easily route around a credit rating that institutional investors require.

Market Intelligence sells data, analytics, and workflow tools on subscription, recurring revenue that barely notices fluctuations in the market.

Indices earns asset-linked fees on trillions of dollars sitting in S&P-benchmarked ETFs and mutual funds, growing automatically as markets rise over time.

This is a duopoly business (alongside Moody’s) operating in a market with almost no real substitutes.

Growth Drivers

Bond issuance reacceleration: Billed issuance jumped 28% in the most recent quarter, driving double-digit Ratings revenue growth. Corporate borrowers refinancing into a more stable rate environment is a direct tailwind for SPGI.

Indices growing fastest of all five segments: Indices revenue grew 17% last quarter, the fastest of any division, powered by rising ETF and mutual fund AUM. This segment scales almost for free.

Capital return acceleration: Management now expects to return 100% or more of adjusted free cash flow to shareholders in 2026 through dividends and buybacks, up from prior guidance.

Mobility spin-off: S&P Global is on track to separate its Mobility division in 2026, a move that should sharpen the remaining business’s growth and margin profile and may unlock value the market isn’t currently crediting.

The Numbers

S&P Global trades at a forward P/E of 20.11x, well below its 3-year and 5-year average multiples, which have run closer to 30-35x.

The free cash flow yield is also at its highest level only seen a few brief times over the past decade:

Full-year guidance calls for 6-8% organic revenue growth, with adjusted EPS guided to $19.40-$19.65.

Bottom line

In my opinion, S&P Global isn’t cheap because the business got worse, but because the stock had a difficult year on sector rotation and general financial-sector pressure. Looking at the numbers, the underlying engine kept compounding. A toll-booth business with this kind of pricing power rarely trades at a meaningful discount to its own history. Right now, it does.

SPGI is set to return all free cash flows to investors. This is a double edged sword - on one hand it is great to get a dividend and buyback yield, but I ideally want to invest in businesses that can reinvest the proceeds at a high return.

Despite this, SPGI is a great defensive quality compounder to bring stability to ones portfolio.

The rest of this article, including our next three picks is for Premium subscribers only.

Read more

Sea’s stock is 53.6% below its recent high. The bear case is that TikTok Shop is dismantling Shopee, gaming peaked, and margins are collapsing. The bull case is that three businesses running simultaneously at 40%+ revenue growth, for the first time in company history, suggests something very different is happening.

Revenue grew 46.6% in Q1 2026. Net income grew 6.7%. The market is looking at that gap and concluding the business model is broken. I think it’s doing something else entirely, confusing deliberate reinvestment with structural deterioration.

Sea Limited is made up of 3 primary businesses:

Business 1: Shopee

Southeast Asia’s Amazon (+Brazil)

Shopee is the dominant e-commerce marketplace across Southeast Asia and the #2 platform in Brazil. In Q1 2026 it generated $37.3 billion in GMV (up 30.2%), $5.1 billion in revenue (up 45.1%), and processed 4 billion orders. The most important number: revenue is growing faster than GMV. Shopee is taking a larger cut of every transaction, not just doing more transactions.

Shopee Q1 2026 key metrics vs prior year:

How is the take rate rising?

Advertising. Shopee’s ad revenue grew 80% year-over-year in Q1 2026, sellers paying to appear first in search results. This is structurally identical to how Google monetises search: high-margin, defensible, and nearly pure profit at the margin.

The VIP program is the detail most analysts skip. Launched in Indonesia in early 2025, it now has 3.5 million subscribers across Indonesia, Thailand, and Vietnam. Those members are 2.5% of all buyers but generate roughly 20% of all GMV. In Indonesia, VIP members spend 40% more after subscribing than they did before. That is not a loyalty program. That is a flywheel lock-in.

The TikTok Threat

Southeast Asia e-commerce market share Q1 2026:

TikTok Shop is a real competitor. It holds 18% of Southeast Asian e-commerce GMV, is growing at 40–55% annually, and its Tokopedia partnership gives it serious infrastructure in Indonesia, the region’s largest market. By 2027–2028, it could reach 25%+. This is a real threat for SE.

What TikTok Shop wins: impulse purchases, fashion, beauty, discovery-driven, low-ticket categories. What Shopee dominates: electronics, home goods, FMCG. These are categories where customers care about delivery guarantees, returns, and warranties. These are meaningfully different shopping occasions.

The scenario that should concern investors is not TikTok Shop at 18%, but TikTok Shop at 28% with a fintech stack that competes with SPayLater. That is the product Shopee’s moat actually depends on preventing.

Lazada, Alibaba’s bet on Southeast Asia, has essentially conceded. After spending over $7.4 billion, it has repositioned as a premium brand platform. One less serious competitor for mass-market GMV.

Business 2: Garena

The cash engine nobody models correctly

Garena operates Free Fire, one of the most downloaded mobile games in history, plus licensed titles including EA Sports FC and Call of Duty: Mobile. It exists inside a company known for e-commerce and fintech, which means most analysts look at it wrong.

A 61.6% EBITDA margin. On $931 million in quarterly bookings. Garena generated over $500 million in operating profit in a single quarter, and that cash is flowing directly into Shopee’s logistics build-out and Monee’s loan book. This is the engine that makes Sea’s reinvestment cycle possible without raising external capital or diluting shareholders.

The user base has been roughly flat for two years (620–670 million quarterly). What’s changing is how those users spend. The paying user ratio has climbed from 8% to 10.9%. Average spend per paying user has risen from $0.83 to $1.40. This is a fundamentally healthier growth trajectory than buying cheap installs.

However, the concentration risk is real. If Free Fire faded sharply, it would hurt badly. The evidence so far points toward an evergreen franchise, IP collaborations with Jujutsu Kaisen, Naruto, and Squid Game have kept the game current, and Arena of Valor just hit record quarterly bookings in its tenth year of operation. But this risk is on the table.

Business 3: Monee

The fintech stack that makes Shopee defensible

Formerly SeaMoney, Monee is Sea’s financial services arm. It lends to Shopee shoppers via SPayLater and to small businesses on the platform, and operates digital banks: SeaBank in Indonesia, Philippines, and Brazil, plus MariBank in Singapore.

Growing a loan book at 71% annually while improving credit quality is not something you see often. The NPL of 1.1% declining year-on-year tells you the underwriting model works at scale.

The competitive advantage for Monee is data. Every Shopee purchase, every payment, every return generates a transaction trail. Monee uses that to assess creditworthiness with far more precision than a bank that only sees salary and credit score. A regional bank cannot replicate this without first building a marketplace with hundreds of millions of buyers. A fintech startup cannot replicate it either, they don’t have the data.

SeaBank Indonesia ended 2025 with 28 million customers. It is the only Indonesian digital bank generating returns on assets comparable to conventional retail banks. Most digital banks are still burning cash acquiring customers.

Brazil is the next major growth vector. Monee obtained a new financial licence in Brazil in Q1 2026. The Brazilian loan book crossed $1 billion, up 250% year-over-year. SPayLater penetration in Brazil is roughly 10% of GMV versus significantly higher in mature Southeast Asian markets.

The Flywheel

Each business makes the others stronger. Shopee generates buyers and transaction data. Monee uses that data to lend profitably, and its credit products, SPayLater, digital banking, make Shopee harder to leave. Garena generates cash that funds Shopee’s logistics build and Monee’s loan book growth. No external capital required, resulting in no dilution.

What Is Sea Worth?

Sum-of-the-parts

Sea’s balance sheet holds $11.1 billion in cash and investments against less than $800 million in debt, roughly $10.3 billion in net cash, nearly 20% of the entire market cap. In November 2025, the company announced its first-ever $1 billion share buyback, a signal that management believes the stock is undervalued.

Sum-of-parts valuation bridge vs current market cap.

At $86/share, the market implies Shopee and Monee are almost free. That is the opportunity, or the warning sign, depending on your view of TikTok Shop.

I love to look at the sum of all parts, but we almost never see a business being fully valued from this perspective, there is always a ‘conglomorate discount’. Look at Mercadolibre and Amazon as an example.

Despite this, Sea Limited is in a strong position in a very interesting market. If SE can continue to prove its profitability while keeping their Shopee market share, the result for investors can be lucrative.

Here is what needs to be true for our bear, base, and bull case to play out:

The Risks You Need to Own

Shopee margin compression

Management guided Shopee’s profit margin at ~0.6% of GMV for all of 2026, versus a long-term target of 2–3%. This is deliberate investment, but patience is required. The re-rating catalyst is the first quarter where this metric inflects upward year-over-year.

Garena concentration

Free Fire is extraordinary. It is also a single game. A better replacement could erode it faster than IP collaborations can offset. The franchise looks evergreen, but that assessment has to be revisited every quarter.

Untested credit cycle

Monee’s 1.1% NPL looks pristine today. The loan book has only existed at scale during a period of relative economic stability. A sharp consumer downturn in Southeast Asia or Brazil is the real test, and that test has not yet arrived.

Governance concentration

Forrest Li holds ~59% of voting rights. Minority shareholders have limited ability to influence governance. You are backing the founder, and the founder alone. That is a feature for some investors and a dealbreaker for others.

Currency risk is also real and underappreciated. Sea earns in Indonesian Rupiah, Brazilian Reals, Thai Baht, and Vietnamese Dong, and reports in USD. FX moves add volatility that has nothing to do with operational performance and can distort quarter-to-quarter comparisons significantly.

Concluding Thoughts

Sea is a company where three distinct businesses compound each other, all running at 40%+ revenue growth simultaneously, on a balance sheet with $10+ billion in net cash, at a valuation that implies the flywheel stops spinning.

The market’s central fear, that TikTok Shop structurally impairs Shopee’s profitability, is a legitimate risk, not a dismissible one. TikTok Shop at 25%+ with a fintech layer would be a materially different competitive environment. What the market is getting wrong in my opinion is treating that as the current reality rather than a tail risk to monitor.

The single metric to watch: Shopee EBITDA as a percentage of GMV. The moment that starts inflecting upward year-over-year, the re-rating begins. Until then, you are being paid in growth while you wait, in a business where the underlying flywheel is turning faster than the stock price reflects.

Ready to take the next step? Here’s how I can help you grow your investing journey:

1. Go Premium — Unlock exclusive content and follow our market-beating Quality Growth portfolio. Learn more here.

2. Essentials of Quality Growth — Join over 300 investors who have built winning portfolios with this step-by-step guide to identifying top-quality compounders. Get the guide.

3. Free Valuation Cheat Sheet — Discover a simple, reliable way to value businesses and set your margin of safety. Download now.

4. Free Guide: How to Identify a Compounder — Learn the key traits of companies worth holding for the long term. Access it here.

5. Free Guide: How to Analyze Financial Statements — Master reading balance sheets, income statements, and cash flows. Start learning.

6. Get Featured — Promote yourself to over 24,000 active stock market investors with a 42% open rate. Reach out: investinassets20@gmail.com

Disclaimer:

This newsletter is for informational purposes only and does not constitute financial, investment, or other professional advice. The views expressed are solely the author’s opinions and may change without notice.

Investing in securities involves risk, including the potential loss of capital. Past performance is not indicative of future results.

The author may hold positions in securities mentioned. Readers should do their own research and consult a licensed financial advisor before making investment decisions.

The Quality Growth Portfolio had a strong first quarter.

Capital being deployed at unprecedented scale, free cash flow compounding, acquisition machines running.

But one position is sitting differently from the rest. And that tension is worth addressing.

Let me start there👇

Read more

Today we’re looking at what the true value of a portfolio is.

Many view the price as the true value of the portfolio, but is that fair?

Mr. Market is a manic depressive that changes his mind constantly.

We’ve seen it hundreds of times, a high quality business sells off due to a prevailing narrative.

Meta Platforms traded at a single digit multiple in 2022 after a series of bad news:

The stock sold off big time. So what did Meta do to change the trend? Nothing special, it continued to deliver solid growth, high margins, and produce high levels of free cash flows.

The price went up +700% in less than 3 years… Same business, but a radical different price.

This begs the question: Is price really a good measure of the true value of a portfolio?

If we look at the business, it most often hasn’t changed at all, even if the stock price has changed significantly.

After all, we’re trying to buy great businesses and be long-term business owners, not stock traders.

So, today, I want to show you how I look at the true value of the portfolio, and more importantly, how I evaluate if a purchase was successful or not.

A better way to look at the true value of a stock portfolio

The true value of the portfolio: earnings and free cash flows per share

I look at both, because free cash flows can be temporarily depressed due to large capex investments. And earnings can be more easily manipulated by accounting shenanigans.

Why focus on earnings and free cash flow per share?

The short answer: Earnings & free cash flows is the weighing machine, stock price is the voting machine. And in the long-term, the weighing machine wins.

In the long-run, the stock price will follow the growth in earnings. In the short-term, you have no control over how the market will price that growth.

Of course, we have to look at a broader picture, to consider how our portfolio is rated compared to other indexes like the S&P 500, MSCI world index and the ‘risk free rate’ of the 10 year treasury bond.

Factors worth tracking for your portfolio:

• The PE and FCF yield

• Compare PE and FCF to the S&P 500 and the MSCI world index

• Compare FCF yield to the risk free rate (10 year treasury bond yield)

• The expected future growth

• Compare it to the expected growth of the S&P 500, MSCI world index, and expected GDP growth globally or for the US.

Here is a overview how I compare my portfolio to the S&P 500:

Successful investments

So, we can’t control how the market rates our businesses, but we can control our entry price, or what multiple we purchase the business at.

Buying a great business at a high multiple is a high risk endeavors, we have the classic example of Microsoft, that used 16 years to regain its dot com bubble highs:

But buying a business at a low multiple can be just as unprofitable as buying Microsoft in 2000.

There are plenty of examples of former darlings that the market used to love, but just keep rating lower and lower.

A well known examples is Evolution AB. The market used to look at the business like an unstoppable force in a growing global market, rating the business at ~90x earnings at its peak.

90x, became 30x, 30x became 10x.

So, buying a business at low (or lower) multiples is not enough, we have to be right about the business, and the discount the market applies to the stock has to be overestimated.

So, entry price is important, and buying a quality business that will continue to grow its earnings and free cash flows per share over time is important.

In periods of multiple contraction, or when the stock price is selling heavily off, like we see in the software industry right now, how should we think about tracking the real performance of our portfolio?

Look-through earnings: The real scorecard

This brings us to a powerful tool for evaluating your portfolio: look-through earnings.

Warren Buffett introduced this idea in his 1989 annual report, specifically to help Berkshire Hathaway shareholders understand the true performance of their equity portfolio. The logic is simple: Rather than looking at what the market says your stocks are worth today, you add up your proportional share of the earnings of every business you own.

François Rochon, founder of Giverny Capital, has applied this framework for decades. As he explains:

This means that if you own 50 shares of a company, you are not holding a ticker symbol. You are a part-owner of a real, cash generating business.

These businesses have real customers, employees, and make real money.

The stock price only tells you what the market is willing to pay for one slice of this business on this day.

Look through earnings tells you what your slice of the business actually produces.

Rochon illustrates this with a simple example.

Imagine a $100,000 portfolio made up of:

• 2,000 shares of ABC (trading at $25 a share). Total value: $50.000

• 500 shares of XYZ (trading at $100). Total value: $50.000

ABC earned $2 per share, XYZ earned $8.

Multiply each by the shares you own, and your total owner’s earnings were $8,000.

The following year, ABC earned $2.40 and XYZ earned $9. Your owner’s earnings grew to $9,300.

This is a 16% growth rate, regardless of what either stock did in the market that year.

That 16% is the real scorecard. Not what Mr. Market said your portfolio was worth on any given Tuesday.

Why price alone is a terrible measure of performance

Buffett put it plainly:

The voting machine can be wildly irrational.

Rochon points to Coca-Cola as a perfect case study. In 2000, Coke traded at a PE of 41. 10 years later, the stock had barely moved:

Over that same period, earnings nearly tripled, a nine percent annual growth rate. The business did its job. The stock didn’t move, because investors in 2000 had massively overpaid.

As Charlie Munger said:

Two factors that determine intrinsic value

Rochon is direct about what actually drives intrinsic value, and he identifies two factors that the market frequently gets wrong.

The first is the future prospects of the business.

Earnings can grow 20% in a year, but if the underlying business model does not look sound over the next five years, that growth will prove temporary and should not be considered a true increase in intrinsic value.

This is why great investors spend most of their time on qualitative analysis: the moat, the management, the competitive dynamics, not just the spreadsheet or discounted cash flow analysis.

Determining future prospects is inherently subjective and imprecise, but it is the work.

The second factor is the price you pay.

If you pay too high a price relative to intrinsic value, even if intrinsic value grows at a good rate, you will not be rewarded by the market.

The Coca-Cola example above is the proof. Overpay at 41x earnings, wait twelve years as earnings triple, and still earn almost nothing. The valuation contraction absorbed all the compounding.

This is why entry multiple is something you can and must control. You cannot control how Mr. Market rates your businesses next year. You can control what you pay today.

How to use this in practice

Each year, calculate your portfolio’s total owner’s earnings. Add up your share of earnings across every position you hold.

Owner earnings can be calculated as follows:

• Earnings or FCF per share + dividend per share x # of shares

This can (And should) be done every quarter to keep track of your holdings.

Then compare that number to the prior year. That growth rate is your real return on the underlying businesses, independent of market noise.

Rochon does exactly this at Giverny Capital. As he writes:

For your own portfolio, you want to see:

• Owner’s earnings growing faster than your benchmark index (E.g. S&P 500 or MSCI World Index) underlying earnings growth

• A PE and FCF yield that is reasonable relative to the index and the risk-free rate

• Expected future growth that justifies the multiple you paid at entry

If those three things are true, you are in good shape, even if the market hasn’t recognized it yet.

Rationality in the face of volatility

The deepest benefit of looking through earnings is psychological. When you know your businesses are compounding earnings at 12-15% per year, a 30% market drawdown stops being terrifying and starts being an opportunity. You are no longer at the mercy of Mr. Market’s mood swings, because you have an independent measure of whether your investments are working.

Rochon makes this point directly:

Buffett has said the same thing for fifty years. The stock market exists to serve you, not to instruct you. If it offers you a great business at a silly price, buy more. If it prices your existing holdings at a premium to intrinsic value, enjoy the quotation and do nothing.

Concluding Thoughts

The true value of your portfolio is not the number on your brokerage screen this morning. It is the sum of what your businesses are earning on your behalf, growing year after year, compounding quietly regardless of what any market does on any given day.

Get the entry multiple right. Own businesses with durable earnings power and strong future prospects. Track your look-through earnings annually, not your portfolio price daily.

Do that consistently, and the market will eventually do the weighing, and the price will follow the earnings.

Ready to take the next step? Here’s how I can help you grow your investing journey:

1. Go Premium — Unlock exclusive content and follow our market-beating Quality Growth portfolio. Learn more here.

2. Essentials of Quality Growth — Join over 300 investors who have built winning portfolios with this step-by-step guide to identifying top-quality compounders. Get the guide.

3. Free Valuation Cheat Sheet — Discover a simple, reliable way to value businesses and set your margin of safety. Download now.

4. Free Guide: How to Identify a Compounder — Learn the key traits of companies worth holding for the long term. Access it here.

5. Free Guide: How to Analyze Financial Statements — Master reading balance sheets, income statements, and cash flows. Start learning.

6. Get Featured — Promote yourself to over 24,000 active stock market investors with a 42% open rate. Reach out: investinassets20@gmail.com

May was a quiet month for the portfolio, up +2.05%.

After April’s strong recovery, May felt like the market taking a breather. The businesses we own kept doing what they do: compounding earnings, taking market share, and building their moats. The stock prices mostly moved sideways to slightly up, and that’s fine.

No positions were adde…

Read more

Welcome to the May edition of Top 5 Buys ✅

You can access our Top 25 Buys for 2026 list as a premium member here.

Top 25 Buys 2026

In this article, we will discuss our top stock picks for May 2026.

Let’s get into it 👇

The Market Sentiment: Greed

After the chaos of April, tariff announcements, geopolitical noise, and a brief detour into Extreme Fear, the market has bounced back with surprising speed. Investor sentiment has settled around Greed, as the S&P 500 reclaims ground near all-time highs.

The investors who acted in April have already been rewarded. Now the question is whether this recovery holds, or whether the underlying uncertainty reasserts itself.

Honestly? It doesn’t matter that much. What matters is finding businesses that compound regardless of what the macro throws at them.

I can’t predict the future, but I can point to 5 quality growth businesses that continue to compound regardless of macro economics.

Here are this month’s Top 5 Buys 👇

This is not investment advice. Always conduct your own due diligence and make your own investment decisions.

Top 5 Quality Buys May 2026 🚀

#1 of 5🏛️ Kinsale Capital

Technology-Driven Insurance Compounder · $KNSL · Richmond, Virginia

E&S Insurance

Kinsale Capital ($KNSL) is the best business model in insurance. A pure-play Excess & Surplus (E&S) lines specialist that has turned underwriting into a competitive advantage. Since IPO in 2016, Kinsale has compounded shareholder returns 33.6% annually by doing something deceptively simple: insuring risks that standard carriers won’t touch, faster and cheaper than anyone else.

The E&S market exists because some risks are too unusual, complex, or volatile for standard insurance markets. Think cannabis businesses, cyber liability, construction projects in difficult environments, and emerging industries. It’s a great business to be in, if you can underwrite like Kinsale Capital does.

The Model: Tech Advantage in a Legacy Industry

The insurance industry runs on legacy IT. Old systems, slow quoting, manual underwriting. Kinsale built its entire platform from scratch, proprietary technology enabling faster quoting, better data capture, and more granular risk selection than any competitor. The result is a structural cost advantage. Kinsale’s expense ratio sits around 20–21%, this is one of the lowest in the E&S market. Combined with disciplined underwriting producing a combined ratio in the mid-to-high 70s, Kinsale is generating ROE consistently above 25–30%. Legacy competitors can’t close this gap. Their infrastructure is too expensive to replace.

Growth Drivers

• E&S market expansion: The E&S market has been taking share from standard lines for years as risk complexity grows. Kinsale is perfectly positioned as a pure-play operator.

• Technology flywheel: As Kinsale writes more policies, its data advantage improves. Better data → better pricing → better loss ratios → more capital to grow.

• Small-to-mid market focus: Kinsale targets smaller, more fragmented accounts where broker relationships are stickier and competition from large carriers is weaker.

• Investment income: A growing float invested at improving rates adds a compounding tailwind to earnings.

The Numbers

Kinsale is trading at its lowest forward PE in over a decade, of 14.77x.

Despite the historical low levels, it still trades at a premium to other insurance businesses, as it should, with much higher combined ratio, return on equity and growth.

The unit economics for Kinsale Capital is one of the most durable in financial services.

Bottom line

Kinsale is a rare investment case, a financial company with a genuine technological moat. Insurance is typically a commodity. Kinsale has made it a compounding machine by building technology that incumbents can’t replicate without tearing down everything they have. The E&S market is growing, the advantage is durable, and management has been disciplined throughout. This is the kind of business you want to own for a decade.

#2 of 5 🌸 Interparfums

Asset-Light Fragrance Compounder · $IPAR · New York, USA

Luxury Fragrance

Interparfums ($IPAR) is one of the best-kept secrets in consumer goods. It’s not a luxury house. It doesn’t own the brands. It licenses them, and that distinction is crucial.

The model is elegant: Interparfums signs long-term exclusive licensing agreements with prestigious fashion brands (Coach, Jimmy Choo, Montblanc, Kate Spade, Karl Lagerfeld, and more), then handles the creation, manufacturing, and global distribution of fragrances under those brands. The brand owner gets royalties without operational headaches. Interparfums captures the economics of luxury fragrance with a fraction of the capital intensity.

The Model: Licensing as Leverage

The fragrance business is incredibly attractive. Gross margins are thick, products have long shelf lives, gifting occasions are recurring, and prestige brands command premium pricing. The Coach license was just renewed through June 2031. A new license with Longchamp (through 2036) was signed in July 2025. The portfolio is diversifying and lengthening.

Growth Drivers

• Jimmy Choo franchise strength: The “I Want Choo” franchise is a genuine hit. Jimmy Choo is growing around 16% annually.

• New license pipeline: Longchamp launches in 2027. New brands entering the portfolio add future growth optionality.

• Global prestige fragrance tailwind: Consumer demand for prestige and luxury fragrance remains robust even as consumers become more selective elsewhere.

• Geographic expansion: North America, Western Europe, and Asia/Pacific all strengthened in 2024–2025.

The Numbers

2026 is a consolidation year by management’s own guidance, they’re laying the groundwork for a strong 2027 as new brands ramp. For long-term investors, that creates a window.

Combine this with the drop in valuation over the past years. Interparfums is now trading at a 7.92% forward free cash flow yield, and a 18.7x fwd. PE ratio, it’s cheapest valuation in over a decade:

Bottom line

Interparfums is a textbook asset-light compounder. It has captured the economics of luxury fragrance without the capital intensity of building a luxury brand. The licensing model is misunderstood by investors who worry about renewal risk, but the track record of renewals and an expanding portfolio makes this business far more durable than it appears. Patient investors who look through the 2026 consolidation year will likely be rewarded in 2027 and beyond.

Read more

How FICO Makes Money

FICO is best understood as a toll booth on the American credit system.

Every time a bank, mortgage company, or auto dealer pulls a credit score to make a lending decision, there is a very high probability they are paying FICO a royalty. FICO invented credit scoring in 1989. The score became the industry standard because it works. It has been validated against hundreds of millions of loans over 35 years. And 90% of top US lenders still use it today.

The business has two parts:

Scores (about 55% of revenue)

FICO licenses its algorithm to the three credit bureaus (Equifax, Experian, TransUnion), who pass it on to lenders. FICO charges a royalty per score. The Scores business runs an operating margin of roughly 89%. Almost every dollar of price increase falls straight to the bottom line.

Software (about 45% of revenue).

The FICO Platform is an AI-powered cloud system used by banks, insurers, and telecoms to automate lending decisions, fraud detection, and collections. Platform ARR is growing fast and is the company’s reinvestment engine.

FICO has compounded by 21.9% annually since its inception in 1990, despite its recent -48% sell off:

The Moat: Toll Booth

The core of the FICO moat is what investors call standard-setting lock-in. When a product becomes so embedded in an industry’s processes, regulations, and infrastructure that switching becomes expensive, even when a cheaper alternative exists.

Layer 1: Woven into regulation

The FICO Score is not just widely used. It is built into federal lending guidelines, Fair Housing Act compliance frameworks, and the investor documentation behind the $10+ trillion US mortgage-backed securities market.

When a fund manager in Tokyo buys a US mortgage security, credit quality is expressed in FICO scores. When a bank’s compliance officer certifies lending standards, FICO is the measuring rod. Replacing it requires re-underwriting historical loan performance, retraining models, renegotiating investor documentation, and convincing regulators. None of this happens quickly.

Layer 2: 70 years of proprietary data

FICO has been in credit analytics since the 1950s. Its models have been trained and refined against more loan outcomes than any competitor can replicate. An independent study published in 2025 found FICO Score 10T is the most predictive score for first-time homebuyer mortgages. In finance, that edge matters. A slight improvement in default prediction saves large lenders hundreds of millions per year.

Layer 3: Extraordinary Pricing power

This is the clearest proof that the moat is real. FICO has raised its mortgage royalty repeatedly and aggressively, with lenders complaining the entire time. And yet the price keeps going up.

Why can FICO keep raising prices? Because the math is simple. On a $400,000 mortgage, a $15 credit fee is not the variable that drives the decision. As FICO’s own EVP put it: their royalty represents approximately two-tenths of one percent of total mortgage closing costs.

The lobbyists complain, but lenders pay.

The Risk You Cannot Ignore

Here is the real bear case, and it deserves a serious hearing.

On July 8, 2025, the FHFA officially allowed mortgage lenders to use VantageScore 4.0 alongside Classic FICO when originating loans for Fannie Mae and Freddie Mac. FICO stock fell nearly 19% in a single session. In March 2026, another 9% drop followed as the credit bureaus (who own VantageScore) began offering it at $1 per score versus FICO’s $4.95.

VantageScore is a joint venture of Equifax, Experian, and TransUnion. The same bureaus that distribute FICO scores own a direct competitor. Their economic incentive to steer lenders toward VantageScore is obvious.

This is a legitimate, structural risk. Not a temporary headwind.

What makes it more complicated:

The FHFA mandate allows VantageScore, but it does not require it. Re-tooling mortgage underwriting systems, retraining credit analysts, and updating compliance frameworks takes years. CEO Will Lansing has faced this question publicly for 15 years:

The numbers still back him up. B2B Scores revenue grew 42% in Q3 FY2025 even after the VantageScore announcement.

The concern is trajectory, not today’s share. The bureaus pricing VantageScore at $1 is predatory. If lenders start accepting it for non-GSE lending first, and then GSE lending, the long-term price ceiling for FICO scores comes down. This is a multi-year risk to watch, not an immediate catastrophe, but a real challenge to the pricing power that has driven so much of FICO’s earnings growth.

The Software Business: What the Bears Miss

When people debate FICO, they focus almost entirely on Scores. The Software segment is becoming very interesting.

FICO Platform is an enterprise AI decisioning system. Banks use it to automate loan origination, fraud detection, and collections. Insurers use it for underwriting. Telecoms use it to manage credit exposure. T-Mobile recently deployed a FICO Platform solution it says transformed its onboarding capabilities for connected device financing.

Platform Dollar-Based Net Retention Rate is 136%. That means existing Platform customers spend 36% more each year, on average. Enterprise software companies with NRR above 120% are considered elite. FICO Platform is delivering at a level that suggests real product-market fit, not just price increases.

The migration story matters too. Non-platform software ARR is declining as legacy customers move to Platform. This creates a short-term ARR headwind but a long-term margin expansion story, as SaaS recurring revenue is more predictable and higher lifetime value than old-school licenses.

FICO’s financial profile is unusual. The company carries net debt, largely from aggressive share buybacks. But the Scores business generates so much free cash flow that this is manageable.

In fiscal year 2025:

• Revenue: $1.99 billion, up 15.9%

• Free cash flow: $769.9 million, up 23.4%

• Diluted EPS: $26.54, up 29.78%

• Shares bought back: $1.58 billion worth

Note the share count. FICO has been buying back stock aggressively for years. The per-share economics compound faster than the headline revenue numbers suggest. Aggressive buybacks works out great when the stock price is fair, but when multiples get extreme (like in 2024-2025) the capital allocation is questionable.

In Q2 FY2026, FICO deployed $606 million in buybacks in a single quarter, the largest in company history, plus $170 million more after quarter-end. At a stock price around $1,100 at the time, that is a significant signal from management.

The Scores segment runs at roughly 89% operating margins. The non-GAAP operating margin for the company overall reached 58% in recent quarters. Trailing four-quarter free cash flow as of Q2 FY2026 was $867 million, up 28%.

Growth: Where Does It Come From Here?

FICO’s revenue growth has three drivers.

Volume

More credit pulls as the economy grows, housing activity recovers, and auto financing continues. This is modestly positive and relatively predictable.

Price

FICO has raised mortgage royalties four times in its history, three of them in the last three years. The VantageScore threat limits how aggressively this continues in mortgages, but there is real runway in auto, card, and personal loan markets, where pricing has been flat for years. B2C direct consumer pricing is also under monetized.

Platform growth

With Platform ARR growing 49% and NRR at 136%, this segment is in the early stages of its potential. FICO’s installed base of enterprise clients is enormous. Attaching more Platform use cases to existing relationships is a multi-year growth driver that doesn’t depend on credit cycle dynamics at all.

Consensus estimates point to roughly 15 to 18% EPS growth over the next two years. Given the buyback program reducing the share count by about 2.5% annually, free cash flow per share growth will run above headline revenue growth.

Subscribe now

Valuation

At roughly $1,236 per share (May 2026), FICO trades at:

• 25x forward earnings

• 32% below its 10-year average multiple of 37x

• 69% below the September 2024 peak

Here are my scenarios for FICO:

The base case: continued double-digit growth from scoring volume and Platform expansion, with a modest re-rating toward (but not back to) historical averages, generates attractive returns from today’s price.

The bear case requires a collapse in Scores pricing power AND the Software segment stalling AND the market assigning a sub-25x multiple to a still-cash-generative business. That’s a lot going wrong simultaneously.

Three Assumptions for a Fantastic Outcome (+15-20% CAGR) 🚀

For FICO to compound at 15 to 20% per year from here, three things need to go right.

Assumption 1: VantageScore adoption stays slow for 5+ years.

The regulatory door is open, but building a new standard takes decades. If mortgage lenders move slowly (which history strongly suggests), FICO’s B2B Scores revenue continues to grow even at flat pricing. Price increases in auto and card scoring (where VantageScore has no regulatory mandate) provide an additional tailwind. This alone could add 8 to 10 percentage points to annual EPS growth.

Assumption 2: FICO Platform becomes a $1B+ ARR business by 2028.

Platform ARR is $349 million today, growing at 49% with 136% net retention. If growth slows to a still-excellent 30% per year, Platform hits $1 billion ARR by late 2027 or early 2028. At Software margins that are expanding as the mix shifts to SaaS, this would add meaningful earnings power and justify a higher quality multiple across the whole business.

Assumption 3: Management keeps returning capital aggressively at depressed prices.

The Q2 FY2026 buyback of $605 million in a single quarter (when the stock was near multi-year lows) is the kind of capital allocation that creates enormous per-share value. If management continues deploying $1+ billion per year in buybacks at depressed prices and the business keeps growing, the per-share earnings math becomes very compelling very fast. EPS in FY2028 in this scenario could reach $55 to $60, versus ~$32 today.

None of these assumptions require anything extraordinary. They require the business to keep doing what it has done for 30 years, while management continues to be disciplined with the cash it generates.

What Must Be True for the Base Case

Even for the base case to work, you need to believe:

FICO’s VantageScore headwind plays out slowly, not suddenly. The operating evidence today: 42% B2B Scores growth in Q3 FY2025, strong Q1 FY2026 results, supports the slow-transition view.

Pricing power survives in non-mortgage markets. Auto, card, and personal loan scoring have seen almost no price increases. These are large markets where FICO’s position is equally dominant.

The FICO Platform compounds. With 136% NRR and 49% ARR growth, this is the kind of business that can grow for years inside an existing customer base.

Management keeps allocating capital well. Will Lansing has run this company since 2007. The buyback program has been consistent and value-accretive for years.

The Bear Case Simply Put

The bureaus own VantageScore. They will subsidize it at $1 per score because every FICO score they displace is economics that stays inside their ecosystem. Over 5 to 10 years, lenders adopt VantageScore for more and more use cases. FICO’s pricing power erodes. Revenue growth slows from mid-teens to mid-single digits. The stock, previously priced as a quality compounder at 40 to 50x earnings, re-rates to 20 to 25x. You get a decade of flat returns.

This scenario is what the stock is partially pricing today. It is not crazy. It requires close monitoring.

Concluding Thoughts

FICO invented credit scoring, became the universal language of creditworthiness in the world’s largest economy, and spent three decades building a pricing moat so strong it can charge $4.95 for something that costs almost nothing to produce.

The VantageScore threat is real. The regulatory environment is hostile. The stock got expensive at the peak.

But at 25x forward PE, 48% below its peak in 2024, with an $867 million free cash flow machine, management buying back stock at the fastest pace in the company’s history, and a Platform business growing at nearly 50% per year, the risk-reward looks more interesting than the narrative suggests.

This is not a set-and-forget holding. The VantageScore situation requires active monitoring. If adoption accelerates and Scores pricing starts to compress meaningfully, the thesis weakens. But the bears are pricing in a level of structural decline that the current operating numbers do not support.

FICO is not a broken business, yet. Currently it is discounted. Whether that discount is temporary or permanent is a question that will take 3 to 5 years to answer fully.

Whether I decide to start a position in FICO remains to be decided, you can follow the Quality Growth portfolio and get premium content by becoming a member today:

Subscribe now

Ready to take the next step? Here’s how I can help you grow your investing journey:

1. Go Premium — Unlock exclusive content and follow our market-beating Quality Growth portfolio. Learn more here.

2. Essentials of Quality Growth — Join over 300 investors who have built winning portfolios with this step-by-step guide to identifying top-quality compounders. Get the guide.

3. Free Valuation Cheat Sheet — Discover a simple, reliable way to value businesses and set your margin of safety. Download now.

4. Free Guide: How to Identify a Compounder — Learn the key traits of companies worth holding for the long term. Access it here.

5. Free Guide: How to Analyze Financial Statements — Master reading balance sheets, income statements, and cash flows. Start learning.

6. Get Featured — Promote yourself to over 24,000 active stock market investors with a 42% open rate. Reach out: investinassets20@gmail.com
   

Disclaimer:

This newsletter is for informational purposes only and does not constitute financial, investment, or other professional advice. The views expressed are solely the author’s opinions and may change without notice.

Investing in securities involves risk, including the potential loss of capital. Past performance is not indicative of future results.

The author may hold positions in securities mentioned. Readers should do their own research and consult a licensed financial advisor before making investment decisions.

I don’t buy stocks I’m not ready to own for a decade. But that doesn’t mean I stop looking.

There’s a list I keep of companies that have passed my initial quality filter, that I’ve spent real time understanding, but that I haven’t pulled the trigger on yet. Either the price isn’t right, or I’m still working through a question I can’t fully answer, or I’m simply being patient.

Today I want to share five names sitting on that list right now. These aren’t speculative ideas. Every one of them has a durable competitive advantage I can articulate clearly, a management team that allocates capital like an owner, and a business model I’d be comfortable holding through a downturn.

What they don’t all have yet is the right entry point.

1. FICO: The toll road nobody talks about

If you’ve ever applied for a mortgage, a car loan, or a credit card in the United States, FICO scored you. You probably didn’t think about it. Neither did the lender. That’s exactly the point.

Fair Isaac Corporation sits in one of the most enviable competitive positions in American finance. Its FICO Score is embedded so deeply into the credit decisioning infrastructure of US lenders that replacing it isn’t a technology problem, it’s a coordination problem. Every major lender uses it. Every regulator references it. Every consumer knows their number. The network is the moat.

What makes FICO genuinely interesting right now isn’t the scoring business alone. It’s the transition happening underneath it. The company has been transforming its software division into a subscription-based SaaS platform. Revenue from this segment is becoming stickier, more recurring, and higher margin. And then there’s the pricing story: FICO charges mortgage lenders a fee each time a score is pulled, and that fee has been going up, meaningfully, year after year. The lenders complain. They always complain. But they keep paying.

Put all three together, the scoring monopoly, the software platform transition, and demonstrated pricing power, and you get a business that has compounded both earnings and free cash flow at a remarkable rate for nearly a decade.

The SaaS platform ARR tells an equally compelling story. Platform ARR has grown from $47.7M in 2021 to $348.8M LTM, a near 7x increase in four years.

Meanwhile, Dollar-Based Net Retention on the platform sits at 136% LTM, meaning existing customers are spending significantly more each year.

FICO has traded at astronomic multiples in the last few years, but now FCF yields are at multi year highs, giving investors a opportunity to both get earnings/fcf growth and multiple expansion if the investment case turns out bullish.

Read more

Here’s something I used to do every time I started researching a new stock.

I’d open the annual report. Read 40 pages of reporting jargon. Open a competitor’s filing. Try to hold both in my head simultaneously. Build a rough DCF in a spreadsheet. Go back and re-read the sections I’d already forgotten.

It worked. But it took forever. And the result was only great if I spent significant time and effort to make it happen.

Now I do most of that groundwork in Claude. Not because AI replaces my thinking (it definitely doesn’t), but because it dramatically accelerates the legwork so I can spend more time on what actually matters: forming a strategic long-term view of the business and if it fits into my portfolio.

This guide shows you exactly how I set it up. You can steal every prompt if you want to. I’ll use Constellation Software (CSU) as the running example throughout, since it’s a stock many of you know well and it’s a great test case for quality compounder analysis.

The setup in Steps 1–2 takes about 10 minutes once. After that, each new stock research session takes 5 minutes to get to a point that used to take 3-5+ hours.

Why Most Investors Use AI Wrong

If you’ve opened Claude or ChatGPT and typed “Tell me about Constellation Software,” you’ve experienced the problem.

You get a fluent, confident, surface-level summary. It reads well. It tells you nothing you couldn’t find on the Wikipedia page. It’s almost useless for serious investing.

Claude out of the box is a general-purpose assistant. It doesn’t know you’re a quality-focused investor who cares about ROIC, capital allocation, and durable competitive moats. It doesn’t know what you already know, what your investment horizon is, or what you’re trying to decide.

The difference between a generic AI response and a genuinely useful research session is context and structure. That’s what this guide is about.

The 5-Step Setup

Step 1: Choose the Right Tool and Plan

I use Claude’s pro plan to get access to the most powerful AI model. This guide will revolve around Claude. I’ve used multiple AI Models, but nothing comes close to Claude in my opinion.

Here’s why this matters for investors:

• Claude handles very long documents better than most alternatives. You can paste or upload an entire 100+ page annual report and have a coherent conversation about it.

• The Pro plan gives you access to Claude’s most capable models with a significantly larger context window. This is critical when you’re working with 10-Ks, earnings transcripts and competitor filings simultaneously.

• Claude tends to be more careful about expressing uncertainty than alternatives, which matters when you’re making real financial decisions.

Important caveat

Claude’s knowledge has a training cutoff. It will not know last quarter’s earnings. Always provide recent documents yourself rather than relying on Claude’s own knowledge of current financials.

Step 2: Write Your Investor Identity Prompt

This is the most important step in the entire guide. If done right, it transforms Claude from a generic assistant into something that thinks like your analyst and provides tremendous value.

At the start of every research session, paste this prompt (customised to you) before you do anything else. It primes Claude with your investing philosophy, your framework, and the level of analysis you expect.

PROMPT 1: Your Investor Identity (customise and save this)

Save this as a text file and paste it at the start of every new conversation. It takes 10 seconds and it changes everything.

Step 3: The Annual Report Deep-Dive

Now the real work begins. Upload or paste the company’s most recent annual report (or 10-K for US companies). For CSU, this is their annual letter to shareholders plus the full financial statements.

Don’t ask Claude to “summarise” it. That’s the generic approach that gives you generic results that are low value. Instead, use a structured prompt that mirrors how a good analyst actually reads a filing.

PROMPT 2: Annual Report Deep-Dive

This prompt alone will get you 80% of the way through a first-pass analysis. The output won’t be perfect, but it will be structured, consistent to the framework you want, and far faster than reading the full document yourself from scratch.

Step 4: The Red Flag Check (Steelman the Bear Case)

This is a crucial step, and will often be extremely valuable.

After you’ve built a positive view of a company, confirmation bias kicks in hard. You start reading everything through a bullish lens (I’ve fallen into this trap many times). Claude can help you fight this by being instructed to argue against you.

PROMPT 3: Steelman the Bear Case

The quality of Claude’s bear case will depend heavily on the quality of the document you gave it. If you also upload a short-seller report or a critical analyst note, add that to the conversation before running this prompt.

Step 5: Build Your Valuation Scenarios

I don’t use Claude to build my actual discounted cash flow. I do that in a spreadsheet where I control the inputs. But I use Claude to stress-test my assumptions before I build it, this catches lazy thinking before it manifests into a valuation model.

PROMPT 4: DCF Assumption Stress Test

Fill in your own numbers before running this. The output gives you a structured framework you can take directly into your spreadsheet model.

The result from my Constellation Software examples 👇

Claude acts as my sparring partner for my assumptions, giving me pointers to consider for each of the assumption I have given it.

This is how I would proceed:

Claude then does the calculations based on my inputs, and creates visuals and graphs to make it more professional looking.

I would follow this up with a sensitivity analysis to better understand how movements in FCF per share growth and FCF yield will affect the return.

Note: Always give Claude current numbers like today’s stock price, or for this case, today’s FCF yield.

My prompt (Claude now has all the context, and you can ask it to do what you want and get accurate results):

How awesome is that?

I now have a well put sensitivity analysis on Constellation Software that is basically ready for presentations or a report.

If you want it in excel or Power Point format, you can just ask Claude to create the document for you with the proper formatting.

This used to take an analyst hours to do. Claude can now do this for you in a few minutes, with the right prompts and correct reporting documents.

After you’ve done this one time, this process will take you less than 5 minutes, and you can refine the prompts over time to fit your style or needs.

The Prompt Playbook: Save These

Here are three additional prompts I use regularly that didn’t fit into the five steps above. These are worth saving to a text file alongside your identity prompt.

Competitive Comparison

PROMPT 5: Head-to-Head Competitor Analysis

Earnings Call Analysis

PROMPT 6: Earnings Call Deep-Read

Initial Screening

PROMPT 7: 5-min analysis (before deep research)

What Claude Can’t Do (Very important)

This section matters as much as the prompts.

• Claude can hallucinate specific numbers. Always verify financial figures against the source document. Never use a number Claude states without checking the original filing.

• Claude doesn’t know what happened last quarter. Its training has a cutoff date. For anything recent, you must provide the documents yourself (I prefer to always provide documentation).

• Claude can’t tell you what a stock is worth. It can help you stress-test assumptions, but the judgment call on valuation is yours.

• Claude is susceptible to the quality of what you give it. A well-written, honest annual report produces better analysis than a PR-heavy one. Garbage in, garbage out.

• Claude doesn’t replace pattern recognition built over years of reading businesses. It accelerates research. The decision maker and analyst is still you.

• Claude and AI is a tool that can enhance a great thinker into an exceptional one. By taking the grunt work for you, it frees up time to focus on the truly important questions.

The right mental model

Think of Claude as a very well-read intern who has read everything but experienced nothing. They can synthesise documents faster than any human, structure frameworks clearly, and challenge your thinking. But they have no skin in the game, no track record, and no judgment built from watching businesses succeed and fail over decades. That judgment is yours.

Where to Start Today

Don’t try to implement all five steps at once. Here’s the minimum viable version:

1. Copy Prompt 1 and customise it to your own investing philosophy. Save it as a text file.

2. Pick a stock you already know well and upload its latest annual report.

3. Run Prompt 2 and Prompt 3 back to back. See whether the analysis surprises you, and whether it catches anything you missed.

That’s it. The whole thing takes 25 minutes for a stock you’re already familiar with. Once you’ve done it once, the workflow becomes second nature.

If you build something useful on top of this, a prompt I haven’t included, a workflow that works better for a specific type of business, leave it in the comments below or reply to this email.

Subscribe now

Marius

Whenever you are ready, this is how I can help you:

1. Go Premium to access exclusive content & follow our market-beating Quality Growth portfolio. Read more here.

2. Essentials of Quality Growth — Join more than 300 investors who have bought the guide. Essentials of Quality Growth Investing is a multi-step guide for building a stock market portfolio of 10-20 high-performing quality compounders.

3. (Free) Valuation Cheat Sheet — Learn an easy and reliable method of valuing a business. Learn how to set a margin of safety for your investments.

4. (Free) How to identify a compounder — Learn how to effectively look for great companies that you can buy and hold for the long term.

5. (Free) How to analyze the financial statements — Learn how you read & analyze the balance sheet, income statement, and cash flow statement.

6. Promote yourself to +15.000 stock market investors (42% open rate) — Contact us via: investinassets20@gmail.com

Warren Buffett has spent decades looking for one thing: businesses that can stay profitable for years, even when competitors try to steal their customers.

He calls this advantage a “moat.”

Think of medieval castles. The wider and deeper the moat around the castle, the harder it was for invaders to get in. Same with businesses. The stronger their competitive advantage, the harder it is for rivals to chip away at their profits.

Here’s the thing most investors miss: not every successful company has a moat. Some businesses are great today but vulnerable tomorrow. Others look boring but are practically impossible to displace.

The difference? Five questions.

Question 1: Can Competitors Easily Copy This?

This is where most companies fail the moat test.

If a rival can replicate your business model in six months, you don’t have a moat. You have a head start. And head starts evaporate fast.

Take restaurants. You can copy almost anything—the menu, the decor, the service style. That’s why most restaurants struggle to maintain high margins. There’s always another spot opening down the street.

Compare that to Visa. Good luck building a competing payment network. You’d need to convince millions of merchants and billions of consumers to switch. Even with unlimited money, it would take decades.

The pattern: Patents, proprietary technology, massive scale, or network effects create moats. Execution alone doesn’t.

Question 2: Would Customers Care if This Company Disappeared?

Brutal question. But it cuts right to the core.

If your business vanished overnight, would customers scramble to find you? Or would they shrug and buy from whoever’s convenient?

Coca-Cola passes this test easily. People have emotional connections to the brand. They’ll pay more for Coke than generic cola, even when the taste difference is minimal. That’s pricing power—the ultimate sign of a moat.

Airlines? They fail hard. Nobody cares which airline they fly. They just want the cheapest ticket. That’s why airlines have such terrible economics despite moving millions of people every day.

The test: If customers view your product as interchangeable with competitors, you’re a commodity. Commodities don’t have moats.

Question 3: Does the Company Have Pricing Power?

Here’s the simplest moat test: If they raised prices 10% tomorrow, would customers keep buying?

Apple can do this. They’ve done it repeatedly with iPhones, and people keep lining up. That’s a moat.

Most retailers can’t. Raise prices and shoppers just go next door. That’s the absence of a moat.

Pricing power shows up in the numbers. Look at gross margins over time. Companies with moats maintain or grow their margins. Companies without moats see margins slowly erode as competition forces prices down.

Fico is a great example of pricing power and increasing gross margins:

Costco is also interesting. They deliberately limit their pricing power—they cap markups at 14% on branded goods and 15% on their Kirkland private label. But this constraint is actually their moat. By committing to low prices, they’ve built customer loyalty so strong that their membership renewal rate exceeds 90%. Members know Costco won’t gouge them, so they keep coming back.

The insight: Real pricing power means you can raise prices without losing customers. If you can’t, your moat is weak or nonexistent.

Question 4: Are Returns on Capital High and Staying High?

This is where finance meets reality.

Return on Invested Capital (ROIC) measures how much profit a company generates from every dollar it invests in the business. High ROIC companies turn money into more money efficiently.

We know that high ROIC businesses perform much better than their low-ROIC counterparts:

But here’s what matters for moats: consistency.

Companies with real moats maintain high returns on capital year after year. Why? Because their competitive advantages protect them from the usual pattern where high returns attract competition and drive returns back down.

Look at Microsoft. For decades, they’ve generated returns on capital well above 20%. Competitors try to chip away at Office, Windows, and now Azure, but Microsoft’s ecosystem moat keeps the profits flowing.

Compare that to most retailers or manufacturers. They might have one great year with high returns, but then competition catches up and returns fall. That’s capitalism working—but it also signals no moat.

The data: According to research from Missouri State University, companies with wide economic moats consistently outperform companies with no moat. The difference isn’t subtle. It compounds over decades.

Here are the highest ROIC industries for US companies:

Question 5: Is the Moat Getting Wider or Narrower?

Moats aren’t static. They’re always changing.

Buffett himself says moats are either widening or shrinking, even when the changes aren’t obvious in the short run.

Here’s a breakdown from Morningstar on different moat types:

From the data, wide and narrow moat businesses outperform the no-moat companies significantly.

Amazon’s moat has widened dramatically. Every new Prime member makes the service more valuable. Every third-party seller on the platform gives customers more options. Every AWS customer makes Amazon’s cloud business harder to displace. The feedback loops compound.

Traditional retail banks? Their moats are shrinking. Fintech companies can now do most of what banks do—and they’re doing it cheaper and faster. The regulatory moat banks once enjoyed is eroding.

The question: When you look at a company, ask yourself: will their competitive advantage be stronger or weaker in five years?

If you want to get deeper into this subject, I highly recommend Measuring the moat by Mauboussin.

Real World Examples

Let’s apply this framework:

Costco - Wide Moat:

• Can’t copy the scale advantages and buying power (Question 1)

• Customers love the treasure hunt experience and low prices (Question 2)

• Membership model locks in recurring revenue, 90%+ renewal rates (Question 3)

• Consistent returns year after year (Question 4)

• Global expansion and e-commerce growth widening the moat (Question 5)

Apple - Wide Moat:

• Brand built over decades, impossible to replicate (Question 1)

• Emotional connection drives customer loyalty worldwide (Question 2)

• Can charge premium prices vs. competitors (Question 3)

• Maintains high margins despite intense competition (Question 4)

• Brand strength endures across generations (Question 5)

Most Airlines - No Moat:

• Easy to replicate business model (Question 1)

• Customers just want cheap tickets (Question 2)

• Zero pricing power, constant fare wars (Question 3)

• Returns on capital barely cover cost of capital (Question 4)

• Competition keeps intensifying, any advantage is short lived (Question 5)

Why This Matters For Your Money

Companies with wide moats compound returns over decades.

Think about it: if a company can maintain high returns on capital while reinvesting profits back into the business, shareholders win. The returns compound.

Research from Morgan Stanley shows that companies sustaining ROIC above their cost of capital for longer than the market expects generate significantly higher returns for shareholders. These are the “compounders” that create generational wealth.

But companies without moats are traps. They might look cheap on a P/E ratio, but cheap isn’t enough if profits are about to get competed away.

This is why we’re very careful about buying bargain stocks. The price often reflects the quality (But not always).

Conclusion

Five questions. That’s all you need.

Can competitors copy this? Would customers care if it disappeared? Can they raise prices? Are returns staying high? Is the moat widening or shrinking?

Get these right, and you’ll avoid most bad investments.

Because in the end, capitalism is brutal. High profits attract competition like blood attracts sharks. The only defense is a moat wide enough to keep the sharks out.

Most companies don’t have one. The ones that do are worth finding—and holding onto.

Whenever you are ready, this is how I can help you:

1. Go Premium to access exclusive content & follow our market-beating Quality Growth portfolio. Read more here.

2. Essentials of Quality Growth — Join more than 300 investors who have bought the guide. Essentials of Quality Growth Investing is a multi-step guide for building a stock market portfolio of 10-20 high-performing quality compounders.

3. (Free) Valuation Cheat Sheet — Learn an easy and reliable method of valuing a business. Learn how to set a margin of safety for your investments.

4. (Free) How to identify a compounder — Learn how to effectively look for great companies that you can buy and hold for the long term.

5. (Free) How to analyze the financial statements — Learn how you read & analyze the balance sheet, income statement, and cash flow statement.

6. Promote yourself to +25.000 stock market investors (42% open rate) — Contact us via: investinassets20@gmail.com

April was a good month for the portfolio, up +12.34%.

After a bruising March, where tariff uncertainty and macro repricing hit almost every position, April delivered a meaningful recovery. The businesses we own kept compounding. The market started paying attention again.

No positions were added or removed. The portfolio remains concen…

Read more

Today we’re breaking down a quality darling trading at its cheapest valuation in a decade.

Most investors don’t pay attention to Copart.

It’s a company that picks up totalled cars, stores them on land it owns near every major city in America, and sells them at online auction to a global network of 300,000 registered buyers.

They don’t have any exciting AI products, they don’t go viral, and there is nothing fancy about the business. They make money on crashed cars, owned real estate, and a marketplace that has compounded for four decades, run by a family with deep expertise in the industry.

And right now, for the first time in roughly ten years, it’s trading at a 30–35% discount to its long-term average valuation multiples.

The stock is down ~40% from its all-time highs. Two real headwinds have arrived simultaneously. And management just deployed $1.12 billion buying back its own shares in under 90 days.

This is the full breakdown. Let’s get into it. 👇

How Copart makes money

When your car is totalled in an accident and your insurance company decides it isn’t worth fixing, they need to get rid of it. They call Copart.

Copart sends a truck to pick it up, stores it at one of their 250+ yard locations across 11 countries, lists it on their online auction platform, and sells it to the highest bidder. This could be a dismantler in Texas, a rebuilder in Poland, a parts dealer in West Africa, or a mechanic in the Middle East looking for a repairable car.

For this service, Copart charges fees to both the seller (the insurance company) and the buyer. Crucially, they don’t own most of the vehicles, they operate the marketplace on a consignment basis. The seller keeps the proceeds minus fees. Copart keeps the fees.

This model produces economics that would be extraordinary in any industry. In a business that involves physically moving crashed cars around, Copart runs:

• 42% EBITDA margins

• ~32% return on invested capital

• $1.4B of free cash flow on $4.6B of revenue

• $5.1B of cash on the balance sheet with zero meaningful debt

For context, Mastercard, widely considered one of the best business models in the world, runs about 57% EBITDA margins. Copart runs 42% picking up cars from people’s driveways. It is one of the most capital-efficient physical businesses in public markets.

The Moat: A tale of Five Layers

Copart’s moat consists of five layers that create a resilient and sustainable competitive advantage. The five layers compound and build on each other.

1. Physical land ownership

Copart owns the land its salvage yards sit on. More than 250 locations, the majority owned outright rather than leased. Getting a salvage yard permitted near a major metropolitan area takes years of regulatory work, community relations, and capital. You simply cannot replicate this network by writing a cheque. It took Copart four decades to build.

This also explains one of the most interesting strategic moves in recent memory: Copart holds hundreds of acres of otherwise idle land specifically reserved for hurricane-season vehicle storage. That land sits empty, generating zero revenue, until a major storm hits and suddenly becomes the most valuable real estate in the country. No competitor can do this because no competitor has been building this long enough.

2. The digital auction platform, years ahead

Copart moved to an exclusively online auction platform in 2003. Their only real competitor, IAA (now owned by RB Global), was still running physical auctions until COVID forced their hand in 2020. That is a 17-year head start on building a global digital buyer base, an online-first culture, and decades of auction data.

That data advantage matters more than it looks. Every auction outcome: what vehicle, what condition, what buyer, at what price, feeds back into search results, recommendations, and AI-powered tools. A competitor starting today would need years just to replicate the data, let alone the buyer habits.

3. +300,000 global buyers

Copart has 300,000 paying registered members from virtually every non-sanctioned country in the world. International buyers now account for roughly 40% of US vehicles sold and nearly 50% of auction value.

Why does this matter? International buyers pay more. On average, the vehicles purchased by international buyers are 38% more valuable than those purchased by domestic buyers, because they’re shipping them across an ocean, which only makes economic sense for higher-quality vehicles. These buyers pay freight cost, import duties, and logistics, and they do it because Copart’s auction depth gives them confidence they’re getting fair market value.

CEO Jeff Liaw puts it simply: “Liquidity begets liquidity.” More buyers produce higher prices, which attract more sellers, which attract more buyers. Every Copart auction you participate in as a buyer increases the platform’s value for every other buyer. This is a textbook network effect, running inside a business that buys crashed cars.

4. Proprietary data

Copart employs approximately 1,000 engineers, a number that would be notable for a tech company, let alone a salvage auction business. They’ve deployed LLM-powered total-loss decision tools that help insurance carriers make instant total-loss calls from a small sample of vehicle photos. These tools are trained on millions of historical auction outcomes that nobody else has. Two full years into deployment, they’re still expanding.

5. Title Express

When a car is totalled, the insurance company needs to obtain the vehicle title from the bank (if there’s a loan) or the owner. This sounds mundane. It is, in practice, the critical bottleneck that determines how quickly the whole claim resolves.

Copart’s Title Express platform is 5x larger than any competitor’s and delivers cycle times 10 days faster than insurance carriers can achieve on their own. Ten days sounds small. On a ~$15,000 vehicle with daily storage costs, ten days faster means real money saved per claim, multiplied across millions of claims per year. This is the operational advantage that keeps carriers coming back. The one that’s hardest for IAA to replicate because it’s built on scale and purpose-built technology.

Structural Tailwind: Why Cars Keep (And will continue to keep) Getting Totalled

Here is the most important insight about Copart’s business, and the reason this company has compounded for 40+ years despite declining accident rates.

Total loss frequency: the percentage of accident-involved cars that get written off rather than repaired has been rising relentlessly for four decades:

Year Total Loss Frequency according to CCC Crash Course 2026 report:

• 1980 ~4%

• 1990 ~5%

• 2000 ~10%

• 2015 15.6%

• 2025 23.1%

That’s a nearly 6x increase since 1980, and the trend shows no sign of reversing. The CCC 2026 Crash Course report documented record 23.1% total loss frequency for full-year 2025, with Copart’s management citing 24.2% for Q4 2025 specifically.

Why does this keep happening?

Modern cars are extraordinarily sensor-dense. Every bumper has radar. Every A-pillar has a camera. Every headlight assembly has adaptive motors and alignment sensors. CCC data shows that 28.3% of all repairable estimates in 2025 required at least one sensor calibration, up from 21.8% just a year prior. On insurance-company-approved repair estimates specifically, the calibration rate was 35.6%.

The math is simple: if your front bumper gets hit at 15 mph, the repair used to cost $1,200. Now it costs $2,800 because the radar module in that bumper needs recalibration, reprogramming, and verification. At some point, that repair cost exceeds the car’s value, and the carrier writes it off as a total loss. More complexity = higher repair costs = more total losses. And cars only get more complex.

The average US light vehicle was 12.8 years old in 2025. Older cars have lower actual cash values, meaning even modest repair estimates push them over the total-loss threshold.

Electric vehicles make this worse. EVs are sensor-heavy at the perimeter. Any damage that involves bumper-mounted sensors, battery integrity concerns, or charging systems can quickly make repair uneconomical. Jeff Liaw said on the Q4 2025 call: “They total, if anything, more easily.”

The crucial distinction: accident frequency has declined every year for 40+ years thanks to ADAS and safer road design. But total loss frequency has risen 5x over the same period. The total-loss trend has overwhelmed the accident-frequency trend every single year of Copart’s existence as a public company. Liaw’s long-term forecast:

“We’ll reach 25% and we’ll reach 30%.” The current reading is 24.2%. He’s not projecting something extraordinary, he’s extrapolating a 40-year trend.

Subscribe now

So Why Is the Stock Down 40%?

Two headwinds have arrived at the same time, and they’ve been brutal to the reported numbers.

Headwind 1: The underinsurance cycle

Auto insurance premiums rose 17% in 2024 and 7.6% in 2025 as carriers recovered from a brutal claims-inflation period. Consumers responded rationally: they dropped coverage.

The Insurance Research Council found that 33.4% of US drivers are either uninsured or underinsured. This is a 10-point increase since 2017. More starkly, earned car years (active collision policies) fell 4.1% YoY in Q2 2025, even as the actual vehicle fleet grew 1.4%.

That 5.5-point wedge between active policies and vehicles on the road is the underinsurance cycle. When a driver without collision coverage has an accident, the car doesn’t go to Copart, it ends up scrapped, retained, or sold privately. Copart’s US insurance unit volumes fell 9.5% in Q1 2026 and 10.7% in Q2 2026.

This is a cycle, not a structural shift.

Carrier profitability is now at a multi-decade peak. Progressive earned $11.3B in 2025. Allstate earned $10.2B. State Farm earned $12.9B (more than double the prior year) and announced a $5B policyholder dividend alongside rate cuts in 40 states averaging roughly 10%. When carriers cut prices, coverage becomes affordable again, earned car years recover, and salvage volume follows with a 6–12 month lag. We appear to be at the beginning of that cycle today.

Headwind 2: The Progressive/IAA volume shift

Progressive is the fastest-growing major US auto insurer of the past decade. It has been shifting a larger portion of its salvage volume to IAA (Copart’s competitor). Sell-side estimates suggest IAA’s share of Progressive’s salvage business moved from ~75% to ~90% during 2025. Since Progressive represents roughly 15% of the US auto insurance market, this shift amounts to an estimated 50,000–70,000 annualised unit headwind for Copart.

Management won’t say “Progressive” on earnings calls, careful not to escalate a dispute with a customer they presumably want back. But Liaw’s language in Q2 2026 earnings call gave it away: “We may be in a uniquely or unusually Copart adverse moment in time in that respect… but over the long haul we view those trends as often more cyclical than they are secular.”

The net result of both headwinds together: the headline numbers look terrible. Revenue down 3.6% in Q2 2026. EPS below consensus. Unit volumes down 8–10%. The stock has been punished.

The market is always forward looking, but the last 17 quarters have seen OK growth despite the recent headwind to unit volumes.

The Fundamentals

Despite the volume headwinds, the underlying financial quality has been remarkable.

Revenue, earnings and free cash flow per share since 2019:

Gross, operating and free cash flow margins remain robust, FCF margins have expanded significantly since 2023:

Return on invested capital is consistently above 25%, but the decline since 2022 is something investors should monitor over time:

The balance sheet is extraordinary:

• Cash: $5.1B

• Debt: effectively zero

• Total liquidity: $6.35B (including a new $1.25B undrawn revolver signed January 2026)

• Stockholders’ equity: $9.79B

This is one of the cleanest balance sheets in the S&P 500 for a company of this size.

Low debt levels create more room for flexibility and making good capital allocation decisions in down cycles. And the management team has proven its prudence for capital allocation in the past.

The Competitive Landscape

Copart operates in a genuine duopoly for US insurance salvage auctions. Its only real competitor is IAA, now owned by RB Global following a $7.3B acquisition that closed in March 2023.

That acquisition was messy. Leverage spiked to 3.0x. There was CEO turnover. Integration costs were significant. But RB Global has stabilised. Leverage is still high, but manageable (Although not something we would invest in):

Here’s where the valuation comparison gets interesting — and somewhat baffling:

• RB Global trades at a forward PE of 23.6x

• Copart trades at a forward PE of 20.8x

• RB Globals 5 year average ROIC is 7.1%

• Copart’s 5 year average ROIC is 31.6%

• RB Globals long term EPS growth estimate is 10%

• Copart’s long term EPS growth estimate is 15%

• RB Globals interest coverage is 4.1x (Not good)

• Copart’s interest coverage is infinite (No debt, no interest payments)

Cheaper valuation, higher return on invested capital, higher future growth estimates, and no debt compared to high debt levels…

In plain English: Copart is trading at a lower multiple than its more levered, lower-quality, lower growth competitor.

The only explanation is momentum. RBA is improving from a messy base. CPRT is declining from a pristine one. The market is paying for direction-of-change, not quality of business. That mispricing tends not to last.

Management: What The Buyback Actually Tells You

Jeff Liaw, who became sole CEO in April 2024, is worth paying close attention to. His earnings call language is intellectually rigorous, unusually specific, and direct about uncertainty in a way that most public-company management teams are not.

But what matters most right now isn’t what Liaw says, but what Copart does.

For nearly three full years (2023, 2024, 2025, and Q1 2026) Copart repurchased zero shares while cash built from $1.8B to $5.1B. The inaction looked strange for a company sitting on a mountain of cash.

Then, in under 90 days:

• November 2025 – January 2026: 5.48M shares repurchased for $218M at ~$39.82 average

• February 1 – March 2, 2026: 24.26M shares repurchased for $899M at ~$37.11 average

• Total: 29.7M shares, $1.12B deployed, at a blended average of $37.60

Liaw’s characterisation of the timing on the Q2 2026 call: “There’s no particular witchcraft or anything magical to it. I think it’s a function of what general valuation multiples are and where interest rates are, our own views of Copart’s relative valuation.”

No witchcraft. Just management deploying 22% of its entire cash balance to buy back stock at the cheapest valuation in a decade, after doing nothing for three years when the stock was at $50–$60.

This is exactly how great capital allocators behave. Copart did massive buybacks in 2011 (29% of shares outstanding) and 2016 (12% of shares outstanding):

Both times, the stock subsequently compounded dramatically.

Willis Johnson, the founder who still owns ~74 million shares worth ~$2.5B, built a culture of treating each dollar as precious, and buying back stock only when the math is clearly in shareholders’ favour.

The message from this buyback is unambiguous: management believes the stock at $34–38 is materially undervalued relative to their internal estimate of intrinsic value.

Valuation: Multi-Year Lows

At $33.87, here’s where Copart trades against its own history:

Copart currently trades at 21.22x forward PE, not seen since 2017:

Free cash flow yields are also trading at the highest since 2017:

This is a 30–35% discount to long-term averages. Copart last traded at these multiples in 2017. Before that, briefly during the 2008–2009 financial crisis. In 32 years as a public company, this is the deepest non-crisis multiple compression Copart has ever experienced.

The market is implying, through this multiple, that Copart will grow revenues at roughly 2.5–3.5% annually with flat margins. Given a business that has compounded at 10%+ every rolling three-year period since 2003, a 300,000-member global buyer network still expanding, and a structural total-loss-frequency trend that has beaten expectations every year, the market’s implied growth rate looks about half of what this business demonstrably delivers over time.

Three 18-month valuation scenarios:

Bull case $55–65:

Unit volumes inflect positive in Q3–Q4 2026 as carrier rate cuts restore coverage. Progressive allocation partially reverses. Buybacks reduce share count 7–10%. 2027 EPS of ~$2.10–2.20 at 28x = $58–62.

Base case $44–50

Modest volume recovery. Progressive shift stabilises. Margins hold at 41–42%. Buybacks provide EPS floor. 2027 EPS of ~$1.85–1.95 at 24–26x = $44–50.

Bear case $25–30

Progressive loss proves structural. Underinsurance persists. DOJ settlement costs $300–500M. Margins compress. 2027 EPS $1.50–1.65 at 18–20x = $26–30.

That is the short term scenarios, here are the three long-term scenarios for Copart:

• Long-term value estimate: $47.52

• Current price: $33.87

• Upside: +40.3%

• CAGR potential: +17.5%

The buyback math alone is meaningful

At a cumulative $2.5B deployed at ~$38, the share count falls roughly 6.7%. Applied to base-case 2027 EPS, post-buyback earnings per share would be approximately 7% higher than pre-buyback. That turns a flat-EPS year into a growth year and puts a hard floor under the stock.

Analyst consensus price targets range from $32 (Barclays, Underweight) to $62–65 (HSBC and CFRA, Buy/Strong Buy). Consensus sits around $44–48. Everyone agrees the stock is cheap relative to the business quality, the debate is purely about whether volumes recover.

What Could Go Wrong?

No breakdown is complete without an honest risk assessment.

The underinsurance cycle could be slower to reverse than expected. Carrier rate cuts in 40 states are happening now, but the lag from rate filing → policy in force → accident → salvage consignment is 6–12 months. If macro conditions worsen and drivers prioritise other expenses over reinstating coverage, the salvage pool recovery could be delayed by another 12–18 months.

The Progressive/IAA shift may be partly structural. IAA has genuinely improved its service metrics post-acquisition. Part of this volume shift likely reflects real carrier-level preference for two-vendor redundancy — not just IAA’s improved execution. It probably doesn’t fully reverse. We size the sustained headwind at 1.5–2.5% of Copart revenue over the next 24 months.

The DOJ investigation is an unresolved overhang. Since October 2023, Copart has disclosed a DOJ letter related to potential money-laundering control violations in its buyer-onboarding processes. As of March 2026, there’s no update beyond boilerplate. Based on comparable enforcement cases, we estimate a likely settlement in the tens-to-hundreds of millions. This is real but not balance-sheet-threatening. But “we don’t know” is the accurate answer.

Long-duration ADAS risk. Over a 20-year horizon, if autonomous vehicles genuinely achieve mainstream adoption, accident frequency could decline sharply enough to offset rising total-loss frequency. Liaw’s response to this on Q4 2025 was honest: “At this point, a de minimis effect.” The fleet-turnover math (16M new vehicles into 289M VIO) suggests this wouldn’t show up materially for decades even in a scenario where fully autonomous vehicles succeed. But it’s a real tail risk.

The Investment Case Summarized

Copart is a duopoly infrastructure business with 41.2% EBITDA margins, 31.6% ROIC, $5.1B of net cash, and a structural tailwind (rising total loss frequency) that has compounded for 40 straight years.

Today it trades at 20.8 forward PE, this is a 30% discount to its 10-year average and a discount to its lower-quality competitor (RB Global). Two cyclical headwinds (consumer underinsurance and one large carrier’s volume shift to IAA) are real, but most likely temporary.

The company’s management (with deep insider ownership and a 40-year track record of disciplined capital allocation) just put $1.12 billion of the company’s own money into buybacks at $37.60 after not buying a single share for three years.

The one question that matters: is the unit-volume weakness a cycle or a structural break? The 40-year total-loss-frequency trend, the all-time-high auction quality metrics (bidder count, pure sale rate, gross returns), and the insurance carrier profitability setup all say it’s a cycle.

At $33.87, you’re being offered one of the highest-quality business models in the US market at a price the company’s own management believes is cheap.

Subscribe now

Whenever you are ready, this is how I can help you:

1. Go Premium to access exclusive content & follow our market-beating Quality Growth portfolio. Read more here.

2. Essentials of Quality Growth — Join more than 300 investors who have bought the guide. Essentials of Quality Growth Investing is a multi-step guide for building a stock market portfolio of 10-20 high-performing quality compounders.

3. (Free) Valuation Cheat Sheet — Learn an easy and reliable method of valuing a business. Learn how to set a margin of safety for your investments.

4. (Free) How to identify a compounder — Learn how to effectively look for great companies that you can buy and hold for the long term.

5. (Free) How to analyze the financial statements — Learn how you read & analyze the balance sheet, income statement, and cash flow statement.

6. Promote yourself to +25.000 stock market investors (42% open rate) — Contact us via: investinassets20@gmail.com

Markets have been brutal lately, especially for quality investors. Tech has sold off. Software has completely collapsed. And a handful of businesses with real moats, cash flows and compounding opportunities, are trading at multi-year low levels.

That’s what we’re digging into today.

These are not businesses that are deteriorating, it is a list of quality names where the fundamentals are still intact, facing sentiment-driven price reductions. The valuation levels has reached levels where long-term investors are likely to be rewarded.

Let’s take a look at the 5 undervalued companies in question 👇

Subscribe now

1. 🧩 Topicus.com $TOI.V

Price: ~CAD $99.60 | Down ~48.8% since July 2025

The business

Topicus is the European arm of the Constellation Software empire. A serial acquirer of niche vertical market software businesses in the Netherlands and broader Europe.

It operates the same playbook as its parent: find small, mission-critical software companies with sticky recurring revenue, buy them at fair prices, never sell, and let the cash flows compound.

Constellation Software still owns 48% of shares outstanding, which tells you everything you need to know about who’s watching this closely.

Why it’s down

Two things have conspired against Topicus in the past year:

• A brutal sector-wide selloff in vertical market software driven by AI fears

• A period of slower organic growth.

The AI risk is real, to some extent. Yes, it is much easier to create the code for services that can replace many of Topicus’ VMS businesses. However, Topicus’ advantage is:

1. Deep relationships with organizations and government. The name of the game here is trust, reliability, and assurances. And Topicus has built a network of loyal customers that won’t trust anybody to process their data or to create workflows deeply embedded in the corporation.

2. Data, data, data. It’s all about the data, and Topicus has this in groves. It’s hard to create industry defining services without customer data to improve the product over time.

3. Switching costs are real. Software is often a really low cost for a corporation. If the migration is a hassle (Which it always is), a company will dread switching software unless they absolutely have to.

AI lowers the barrier to build. It does not lower the barrier to replace. A better product is not enough when the incumbent owns the workflow, the data, and the relationship.

Is Topicus a quality business?

Let’s first look at the growth in the most recent quarters:

• Revenue per share CAGR: +20.4%

• Free cash flow per share CAGR: 26.9%

• Book value per share CAGR: 21.7%

The business is growing rapidly ✅

Return on capital has consistently been in the 15-25% range since 2018:

Gross, operating, and free cash flow margins have been stable:

The company is in solid financial shape:

Valuation

Topicus is is trading at all time low valuation multiple levels.

Right now, the FCF yield is 7.61%, which is really high for a quality business.

Especially with the back drop that the business is not broken, but its cheap because the market doesn’t want to touch software right now.

High quality business, attractive valuation.

2. 🚗 Copart

Price: ~$33.36 | Down ~47.8% since May 2025

The business

Copart is the dominant global marketplace for salvage and used vehicles. It operates a two-sided platform connecting insurance companies (who need to dispose of totaled cars) with buyers (dismantlers, rebuilders, dealers, and exporters).

The business model is extraordinarily capital-efficient once yards are built: network effects, marketplace dynamics, and significant switching costs for insurers who have integrated Copart deeply into their claims workflow.

Why it’s down

Copart’s volumes at key insurance customers declined slightly as the number of uninsured drivers increased following several years of massive insurance premium inflation.

When premiums rise too fast, some drivers drop coverage. Fewer insured drivers means fewer total loss events, which means fewer cars flowing through the auction. Q2 FY2026 revenue fell 3.6% year-over-year to $1.12 billion, with EPS dropping from $0.41 to $0.36.

The market reacted harshly to a business it expects to compound consistently.

The numbers that matter

Copart’s growth has been impressive:

• Revenue per share CAGR: +12.9%

• Earnings per share CAGR: 19.1%

• Free cash flow per share CAGR: 27.4%

Although revenue have slowed down, free cash flow keeps compounding at attractive rates ✅

Copart’s return on capital is unmatched, usually staying well above 25%:

Margins are not only stable, but also expanding over time. This is a sign of a business with a sustainable competitive advantage.

Copart’s moat hasn’t moved an inch, and there is no sign of deterioration.

They have over 200 yards, their own auction technology, and are deeply embedded into the insurance industry claim process (It will take a long time to replicate).

Copart’s balance sheet remains flawless, with a $5.1 billion of net cash position.

While revenue growth has slowed in the last few years, earnings per share and free cash flow continue to grow at attractive rates.

Valuation verdict

At 21x forward earnings for a business with no debt, 34% net margins, $5.1 billion in cash, and a near-monopoly position in vehicle auctions, this is a genuinely attractive entry point.

If you subtract the $5.1 billion from the market cap, you get a forward PE of ~17x.

The FCF yield is also at a multi-year high of 4.79%, not seen since 2017:

Subscribe now

Read more