A perfect quality score does not make a stock resilient

2026-09-14 · By Lubin Danilo, founder of Lubin Investment

Analyze a stock on Lubin Investment

I compute two separate scores for every stock: a financial quality score out of 10, and a structural resilience score out of 5 stars against disruption by artificial intelligence. Comparing both across a hundred stocks my model rates perfect, I find only one in six keeps strong resilience: the two scores measure different things.

The first time I saw a resilience score of 5 out of 5 next to a quality score of only 8 out of 10, I assumed it was a fluke. That was while writing my piece on Motorola Solutions: the company passes only eight of my ten financial criteria, yet earns the maximum score on a completely different measure I compute separately, structural resilience to disruption by artificial intelligence. A strong financial score and strong AI resilience, I figured, should broadly move together.

That stuck with me. If the two scores really measure different things, the reverse should exist too: stocks that look perfect on paper financially, but fragile against AI. To settle it, I pulled the hundred highest scoring stocks from my screener, the ones sitting at the top of the public table, and fetched each one's resilience score.

The result surprised me twice. First because the gap between the two scores is far wider than I expected: barely one stock in six in this group keeps strong resilience. Second because when I rechecked the scores live while writing this piece, several of these supposedly perfect scores had already moved. Neither score is as fixed as it looks.

Two different questions, not one score twice

My quality score out of 10 looks entirely backward: five year revenue growth, net margin, free cash flow per share growth, discipline on shares outstanding, return on capital employed. It answers one factual, verifiable question: has this company already proven, with real numbers, that it turns its business into growing cash? Nothing forward looking about it, only results already published.

The resilience score asks a completely different and far less certain question: will this same business model still work by 2033, once artificial intelligence and automation have reshuffled its industry? I compute it on five independent criteria: will the need this company serves persist, does it keep control of its market, do its strengths hold up against the AI wave or benefit from it, can it expand toward adjacent needs, and above all, does it actually capture the value it creates rather than letting it leak to customers, suppliers, or future AI agents that could route around it. Each criterion is worth half a star or a full star, up to five.

The two scores do not necessarily line up, and that is exactly what makes them complementary: one says whether the company has already won, the other bets on whether it keeps winning once the rules of the game change. Confusing the two is the trap I wanted to avoid by separating them from the start.

Among a hundred perfectly rated stocks, the gap is enormous

My public screener caps out at a hundred rows per screen, full pagination being reserved for paid accounts so nobody can scrape all seven thousand scores in one click. That limit has an unintended but useful side effect here: the first hundred rows, sorted by descending score, form exactly the population I wanted to study, the absolute elite of my universe on financial quality. I pulled the resilience score for each of these hundred stocks.

Result: the average falls to 2.56 out of 5, far from perfection. Sixteen stocks out of a hundred reach 4 stars or more, and the exact same number, sixteen, sit at 1 star or below. Not a single one reaches the full 5 stars in this group: the highest I observed was 4.5.

Motorola Solutions, the stock that triggered this whole investigation, does not even show up in this group: its quality score of 8 out of 10 falls just short of the top hundred cutoff. Yet it holds the maximum 5 star resilience score, a level I found nowhere else in this top 100. If you want to understand why a proprietary radio standard can be worth more, long term, than slightly faster financial growth, I explain it in the dedicated article. The lesson cuts both ways: perfect on one axis says nothing about the other.

Same perfect score, worlds apart: ASML versus On the Beach

ASML sits near the top of both scales at once. On the financial measure, it passes nine of ten criteria as I write this, it still showed ten out of ten a few weeks ago in my last check, the score moves, more on that below. On resilience, it earns 4.5 out of 5 stars: the company holds a near monopoly on EUV lithography, the only technology capable of etching the most advanced chips, the very ones AI data centers demand. Its second quarter 2026 results show 9.3 billion euros in net sales and a backlog ending the year around 38.8 billion euros, more than its entire 2025 revenue: demand is not slowing down. You can read my full thesis on ASML for the mechanics of that monopoly.

But quality and resilience say nothing about price. My model values the stock at 52.6 times its annual free cash flow, at the 67th percentile of its own history: ASML is trading more expensively today than over two thirds of its own recent history. Fair buy price under my model: below 1,297.93 euros. Current price: 1,491.20 euros, a premium of 13%. A remarkable company, at a price that has already priced in much of that quality.

At the other end of the ranking, On the Beach, the British online travel agency focused on beach holidays, also earns the maximum score on my financial grid. Its resilience, though, caps out at 0.5 out of 5 stars, the second lowest in the entire group. The mechanism is simple to understand: unlike ASML, On the Beach owns no proprietary hotel inventory and no airline infrastructure, only a booking interface between the traveler and suppliers, exactly the kind of interface an AI agent can rebuild or bypass easily. The irony is almost perfect: the company confirmed this year that it has integrated with major AI platforms and submitted its own app to ChatGPT to open a new distribution channel, a rational short term response that illustrates exactly the long term risk my resilience score tries to capture: if booking through an AI agent becomes the norm, what stops that same agent from booking directly with the hotel? Its bookings still grew 7% in the first half of 2026, and the stock trades almost exactly at its fair buy price, a discount of just 0.8%: the market here seems to agree more with my resilience score than with my quality score.

Other names in this group approach the same two dimensional peak: Progressive, the auto insurer, thanks to decades of proprietary telematics data; Curtiss-Wright, a supplier of sole source qualified parts for long life military platforms, locked in by certifications that are hard to replicate; and Mastercard, protected by its payment duopoly with Visa, but already back down to 9 out of 10 like ASML since the last update of the public table. The table below widens the sample to the most contrasted profiles in the group, from the most resilient to the most fragile.

Neither score is carved in stone

While preparing this piece, I individually rechecked fifteen stocks from this group on publication day instead of copying my screener table as is. Result: six of them, 40% of the rechecked sample, including ASML, Mastercard, Progressive, ServiceNow, Curtiss-Wright and Bentley Systems, had already slipped from 10 to 9 out of 10 since the public table was last updated. Nothing alarming: my scores recompute continuously as new quarters replace old ones in the calculation window, and a single criterion tipping the wrong way is enough to lose a point. But it means a frozen table, even mine, is already slightly out of date by the time you read it.

The resilience score moves for a different but equally real reason: it is a bet on 2033, not a verifiable fact today. It depends on my judgment about the speed and direction artificial intelligence will take in each industry, a judgment I revise as new signals arrive. One outlier in this group: MercadoLibre shows a 154% discount under my valuation model, a number so extreme it needs to be read with caution, more a sign that the model is pricing in a growth acceleration that has yet to be confirmed than an established fact, the same caution that applies to any discount well above 100%.

That is exactly why I recompute both scores live on every page of the site instead of freezing a ranking once and for all: today's quality does not guarantee tomorrow's resilience, and tomorrow's resilience says nothing about the price you pay today. All three read separately, never as a substitute for one another. You can compare both scores for any stock on my analysis tool, or read my full methodology to see how each criterion is computed. This five star resilience score remains distinct from the A to E resilience scale I introduced earlier on a smaller sample: the five criteria version covers a far wider universe.

FAQ

What is the 5 star resilience score, and how does it differ from the quality score out of 10?

The quality score out of 10 measures financial results already published: growth, margins, cash generated. The resilience score, out of 5 stars, measures something else: the likelihood that a business model still holds up against disruption by artificial intelligence by 2033, across five criteria (the need served, control of the market, strengths against AI, expansion into adjacent needs, and real capture of the value created). These are two independent measures, not one score recalculated differently.

Why can a stock rated 10 out of 10 have very low resilience, like On the Beach?

Because the quality score only looks at past numbers (margins, growth, cash), without asking whether the business model itself is under threat. On the Beach shows solid financial results today, but its role as a booking intermediary, with no proprietary inventory, is exactly the kind of interface an AI agent can rebuild or bypass. Good numbers yesterday do not guarantee the model survives tomorrow.

Can a stock's quality score change from one day to the next?

Yes, and this article verifies it: of fifteen stocks in this group individually rechecked on publication day, six already showed a different score than the public table. Criteria recompute continuously as new quarters replace old ones in the calculation window; a single criterion tipping over can be enough to change the overall score.

Where can I see the quality score and resilience score for a specific stock?

Both scores appear together on every individual stock page, accessible from the full screener or by searching directly for the name or ticker of the company you are interested in.

Related reading

• European quality stocks: which ones are actually cheap?
• Semiconductors: did our stocks crack in July?
• Cash ROCE: does reinvested capital create value?

Analyze a stock on Lubin Investment

About the author

Written by Lubin Danilo, founder of Lubin Investment. A self-taught individual investor, I find fundamental analysis fascinating, and it has delivered excellent results. For three years now, my performance has beaten the S&P 500. But analyzing every stock took too much time: sites with incomplete data, calculation methods and criteria never aligned with mine. And spotting the best stocks was just as time-consuming, even with my own well-defined checklist. So I put my software development background to work to build this software, base my investment strategy on its results, and share it with people who share the same passion as me. It judges a company's quality and its price separately, using criteria drawn from the financial literature (Warren Buffett, Michael Mauboussin, Aswath Damodaran).