Pieter Slegers — Fair Isaac: a 114-page deep dive (Compounding Quality)

One-line take: the archive's best worked example of separating quality from price, because the answer is a no on a business the same post calls "one of the best compounding machines in history". The moat is described in the strongest terms: FICO scores are used in 90% of US lending decisions and over 95% of all mortgage-backed securities, a "tollbooth on the U.S. financial system" earning a small fee on every mortgage, auto loan and card application, protected by high switching costs, institutional inertia and decades of trust, with the highest margins in the industry and a share count reduced 30% over the past decade — a Cannibal Stock. Total Quality Score 7.8/10, at $1,230 and a $28.5bn market cap. Then the verdict: "No. We are not buying FICO at this point in time for Our Portfolio." Two reasons, both quantified where possible. Competition — "it's uncertain whether Fair Isaac can keep its monopoly going forward (AI could make it easier for VantageScore to become successful)". And dilution: "SBC as a % of Net Income equals 22% (!). FICO currently trades at a FWD PE of 25.2x. But if you take into account Stock-Based Comp, the actual valuation level is 30.7x." The pass comes with a named entry price — "We would love to own FICO at a FWD PE of 25.0x (after SBC). This implies a stock price of $901" — a 27% discount to the current price. Note the sequence: FICO was upgraded Hold→Buy on 7 May with these same two objections attached, is declined here, and is ranked Best Buy #4 on 7 June.

1. Stocks & names mentioned

One subject company; the credit bureaus and VantageScore appear as the ecosystem the thesis and the risk both run through. Research legend: QT Qualtrim · SA Seeking Alpha · STK Stock Analysis. Written post with no timestamps — the At link opens the article.

Ticker	Name	Research	View	What he said	At
FICO	Fair Isaac Corporation	QT · SA · STK · FA	Neutral	NOT BOUGHT, with a named entry price. "No. We are not buying FICO at this point in time for Our Portfolio." Two stated reasons: (1) "Fair Isaac is facing more and more competition. Especially from Vantagescore. As a result, it's uncertain whether Fair Isaac can keep its monopoly going forward (AI could make it easier for VantageScore to become successful)." (2) "SBC as a % of Net Income equals 22% (!). FICO currently trades at a FWD PE of 25.2x. But if you take into account Stock-Based Comp, the actual valuation level is 30.7x." The entry condition is explicit: "We would love to own FICO at a FWD PE of 25.0x (after SBC). This implies a stock price of $901 (current stock price: $1.230)." The quality case is not disputed — "one of the best compounding machines in history", classified as an Oligopoly, a "tollbooth on the U.S. financial system", 90% of US lending decisions and 95%+ of mortgage-backed securities, "the highest margins in the industry", share count down 30% in a decade, and a Total Quality Score of 7.8/10 at a $28.5bn market cap. Management's response to the threat is noted: FICO 10T, direct licensing, and cloud-platform investment. "We keep following up on FICO very closely. It could be added to the Portfolio one day."	read ↗
private	VantageScore Solutions	—	Neutral	The named threat, and the reason the deep dive ends in a no: "Fair Isaac is facing more and more competition. Especially from Vantagescore." The AI angle is the sharp part — "AI could make it easier for VantageScore to become successful", i.e. the barrier is model-building capability and that barrier is falling. Listed among the risks in René's own conclusion too, alongside regulatory pressure. The regulatory catalyst behind it is set out in the 7 June issue: the FHFA now permits VantageScore 4.0 for mortgage approvals.	read ↗

Three notes. (1) The SBC adjustment is applied here and not elsewhere. Restating 25.2x as 30.7x is exactly the adjustment that the 14 May Arista dive declined to make on its own 37.7x headline multiple, and that the 7 May model row for FICO also omitted. (2) The rating and the decision diverge. FICO carried a BUY on the published list from 7 May and is ranked Best Buy #4 on 7 June — while this issue says the portfolio will not own it. The two are consistent only if "Best Buy" means a ranked idea for readers rather than a portfolio candidate, which the June issue does state ("the companies in Our Portfolio are not mentioned here"). (3) The source material is guest-written. The 114 pages are by Compound with René; the Quality Score, the SBC adjustment and the no are Compounding Quality's.

2. Talking points

The business, in one sentence

"Fair Isaac makes money mainly by charging lenders, banks, and credit bureaus fees every time you pull a FICO credit score to evaluate someone for a loan, credit card, or mortgage."
Plus a B2B software and analytics segment for fraud detection, lending and customer-management decisions.
Classified in the house taxonomy as an Oligopoly — one of the three buckets set out in the 26 May manifesto.

The moat, stated in market-share terms

90% of US lending decisions and over 95% of all mortgage-backed securities use FICO scores.
Three sources of durability named: high switching costs, institutional inertia, and decades of trust — "FICO scores are deeply embedded in underwriting systems and the U.S. financial system."
The economics that follow: "the highest margins in the industry", high returns on capital, and a capital-light structure.

The cannibal characteristic

"Because FICO is so profitable, they return quite some capital to shareholders through buybacks" — share count down 30% over the past decade.
That is the third of the portfolio's three buckets (owner-operator, monopoly/oligopoly, cannibal), so on paper FICO qualifies on two of the three.
The tension: a 22%-of-net-income SBC bill means a meaningful share of that buyback is offsetting dilution rather than shrinking the company.

The three concerns, as René states them

Competition from VantageScore; regulatory pressure; and "advances in AI could disrupt their business model".
Management's answer is recorded rather than assessed: FICO 10T, direct licensing initiatives, and investment in the cloud platform.
Note the AI risk is framed as competitive rather than substitutive — AI does not replace credit scoring, it lowers the cost of building a rival score.

The SBC arithmetic that decides it

SBC at 22% of net income — more than double Arista's 12.6%, which the same author had called "a red flag" a week earlier.
The restatement: 25.2x forward becomes 30.7x after SBC. That single adjustment moves the stock from "cheapest valuation level of the past 10 years" to expensive.
This is the archive's cleanest demonstration that a cheapness claim depends entirely on which earnings number you divide by.

The no, with a price attached

"We would love to own FICO at a FWD PE of 25.0x (after SBC). This implies a stock price of $901 (current stock price: $1.230)."
That is a 27% fall from here — a real condition, not a placeholder, and materially stronger than the Arista pass, which named no price at all.
The follow-up commitment: "We keep following up on FICO very closely. It could be added to the Portfolio one day."

Publishing someone else's research

The 114 pages are credited up front and again at the end to Compound with René, "he was so kind to share the investment case with us".
What is retained in-house is the part that matters: the Quality Score, the SBC adjustment, and the decision.
Worth registering as a structural point about the product — a distribution channel for outside research, with the house screen applied on top.

3. In plain English

A jargon-free summary of the thesis behind each argued name. (Renders on each name's consolidated page.)

FICO — Fair Isaac Corporation Neutral

Fair Isaac owns the credit score that American lending runs on. Every time a bank checks whether to give someone a mortgage, a car loan or a credit card, it pays Fair Isaac a small fee to pull a FICO score. That happens in about 90% of US lending decisions, and in more than 95% of the mortgage bundles sold to investors. It is described here as a tollbooth on the American financial system — and the description is fair.

The business is extremely profitable as a result. It has the highest margins in its industry, needs almost no capital to run, and has bought back so much of its own stock that the share count has shrunk by 30% in ten years, which quietly increases everyone else's ownership.

So why is it not being bought? Two reasons, and the second is the decisive one.

First, the monopoly may be cracking. A rival score called VantageScore is gaining ground, and artificial intelligence makes building a credible competing score much cheaper than it used to be. The write-up is careful not to overstate this — it says the outcome is uncertain, not that FICO loses.

Second, the cheapness is an illusion once you adjust for how staff are paid. Fair Isaac hands out shares worth 22% of its annual profit as compensation. That is a real cost to existing owners even though no cash leaves the business. Counting it, the shares cost 30.7 times next year's profits rather than the 25.2 times the headline suggests — so what looks like the cheapest the company has been in a decade is not cheap at all.

The refusal comes with a number, which makes it testable: at 25 times profits after the share-based pay is deducted, the shares would be worth $901. They cost $1,230. That is the price at which this becomes a purchase.

Summary derived from the archived Compounding Quality post (text in transcript.txt) for personal study. The underlying 114-page investment case is the work of Compound with René. Not investment advice. © Compounding Quality / Pieter Slegers for source material.