This is investment research, not personal financial advice.

FAIR ISAAC CORP (NYSE:FICO) fell 16.7% to $932.26 on 4 September after the US housing regulator widened VantageScore 4.0 access from a limited rollout to every lender delivering loans to Fannie Mae and Freddie Mac. The decline erased about $4.0 billion of equity value in one session, on 5.6 times the prior 20-day average volume (Yahoo Finance 2026).

The tape treated permission as future substitution. That is directionally sensible: mortgage scoring is part of FICO's highest-margin business, and the directive attacks the standard that allowed it to raise mortgage unit prices. Yet the current evidence does not show immediate, system-wide displacement. July VantageScore use was concentrated in two lenders, formal operating guides still lag the policy statement, and FICO does not disclose the mortgage slice of Scores revenue. The 17% fall is therefore best read as a roughly proportionate reset in moat duration, not a clean estimate of next year's earnings loss.

Every lender received permission; no lender received a mandate

FHFA Director Bill Pulte instructed Fannie Mae and Freddie Mac to approve every lender for VantageScore 4.0, effective immediately. VantageScore's 4 September release said the earlier program had reached 50 lenders and presented the directive as nationwide access to its model (VantageScore September 2026). Reuters independently confirmed the order and its stated aim of lowering borrower costs by adding competition to a mortgage-scoring market dominated by FICO (Reuters 2026).

The distinction between access and use matters. Classic FICO remains approved. The interim framework still lets a lender choose Classic FICO or VantageScore 4.0 for a loan; it does not require the newer model on every loan. FICO 10T was also approved in 2022 but is not yet available on the same broad operating basis. And the reporting architecture remains tri-merge, meaning lenders still obtain data from all three national credit bureaus.

Official documentation also trails the announcement. The FHFA policy page fetched on 5 September still described the limited rollout that began on 22 April. Fannie Mae and Freddie Mac pages used similar approved-lender language rather than a completed all-lender operating update (FHFA 2026; Fannie Mae 2026; Freddie Mac 2026). That does not negate the director's order. It does mean policy intent, system acceptance and actual mortgage delivery should not be collapsed into one date.

Early adoption gives the same warning. HousingWire cited a KBW analysis that put VantageScore at 4.4% of July volume, almost all from Rocket and United Wholesale Mortgage. Lenders appeared to use it mainly to extend eligibility to borrowers who could not be scored as well under Classic FICO, while still pulling FICO on nearly all loans (HousingWire 2026). VantageScore later said it was the sole score on more than 9% of GSE-securitised mortgages from 1 May through 31 August. Those measures cover different periods and possibly different loan populations, so they cannot be reconciled cleanly from public data.

Permission removes a bottleneck. It does not remove lender model-risk work, investor acceptance tests, mortgage-insurer rules, operating-system changes or the habit of using FICO throughout underwriting and servicing. The market repriced the probability of those frictions falling away.

A 91% margin explains why mortgage choice hit so hard

FICO has two businesses that share data science but not economics. Scores licenses credit scores to the three national bureaus, which resell them to lenders and other users. Software sells decision-management, fraud and optimisation products, increasingly through the FICO Platform. The first business is a standard with toll-like economics. The second is enterprise software that still carries implementation work and migration risk.

The latest numbers show where value sits. In the 12 months to June 2026, author calculations from the FY2025 filing and the two nine-month reports put Scores revenue at $1.550 billion, 64.8% of group revenue, with $1.390 billion of segment operating income and an 89.7% segment margin. Software produced $843.6 million of revenue and $231.9 million of segment operating income, a 27.5% margin. Corporate expenses and share compensation sit outside those segment figures (FICO FY2025; FICO Q3 2026).

The June quarter made the imbalance sharper. Scores revenue rose 41% to $458.9 million and produced a 91% margin. Management attributed 49% B2B Scores growth mainly to a higher mortgage-origination score unit price. Software revenue grew only 2% to $215.3 million, and its segment margin fell to 26% from 32%. Platform annual recurring revenue rose 62%, but non-platform ARR declined 17%; total Software dollar-based net retention was 109% (FICO Results 2026).

That quarter contains both sides of the thesis. A 91% margin and fast pricing growth show how entrenched the score became. They also show why the regulator's competition push matters: a dollar of lost Scores revenue carries far more profit than a dollar of slower Software growth. FICO's FY2025 filing states the risk directly. Reduced GSE use of FICO Scores could materially affect revenue, results and the share price (FICO FY2025).

The distribution arrangement is another strength with a fault line inside it. Experian, TransUnion and Equifax supplied 51% of FY2025 revenue by reselling FICO scores. That concentration rose to 60% in the first nine months of FY2026 and 63% in the June quarter. Those same bureaus jointly own VantageScore. They therefore distribute FICO while funding the rival model now receiving broader GSE access. FICO retains the recognised standard and decades of lender validation; the ownership creates a channel conflict and an incentive to promote the competing model (FICO Q3 2026).

Five years of cash growth came with a shrinking accounting base

The filed history is strong. Revenue rose from $1.317 billion in FY2021 to $1.991 billion in FY2025, while operating income increased from $505.5 million to $924.9 million. Operating cash flow rose from $423.8 million to $778.8 million. The June 2026 quarter then delivered $674.2 million of revenue and $237.2 million of net income (FICO FY2021; FICO FY2022; FICO FY2023; FICO FY2024; FICO FY2025; FICO Q3 2026).

USD millions except ratios FY2021 FY2022 FY2023 FY2024 FY2025 Q3 FY2026
Revenue 1,316.5 1,377.3 1,513.6 1,717.5 1,990.9 674.2
Operating income 505.5 542.4 642.8 733.6 924.9 377.3
Net income 392.1 373.5 429.4 512.8 651.9 237.2
Operating cash flow 423.8 509.5 468.9 633.0 778.8 not comparable, quarter only
Author-computed ROIC 44.0% 46.8% 48.1% 53.5% 63.9% not shown
Gross debt / annual or trailing OCF 3.0x 3.6x 4.0x 3.5x 3.9x 5.6x

The return figures need a large warning label. Author-computed ROIC uses operating income after the reported effective tax rate divided by year-end debt plus book equity less cash. Repurchases pushed book equity below zero, so the denominator is not a stable measure of operating capital. It rose from $953 million in FY2021 to $1.176 billion in FY2025 even as retained equity became more negative. The resulting FY2025 ROIC of 63.9% confirms that little physical capital is needed, but treasury-stock accounting and debt-funded repurchases distort it. Incremental ROIC is not shown because changes in that denominator do not measure operating reinvestment cleanly.

Margins and cash conversion are cleaner evidence. FY2025 operating income was 46.5% of revenue. Operating cash flow covered the small physical and internal-use software investment burden: $778.8 million less $39.4 million of property, equipment and capitalised software produced $739.4 million of author-computed free cash flow. On a trailing basis through June 2026, the same bridge was about $1.002 billion of operating cash flow less $40.7 million of capital investment, or $960.9 million.

Share compensation complicates that bridge. Trailing share-based compensation was about $174.3 million, making conservative owner cash roughly $786.6 million if the full amount is treated as an economic cost. That adjustment is deliberately strict. FICO retires shares through repurchases, so compensation does not translate one-for-one into dilution, but the repurchases themselves are now heavily debt-financed.

The repurchase machine borrowed against moat durability

FICO spent $3.046 billion on share repurchases in the first nine months of FY2026. A June accelerated repurchase involved a $1.5 billion prepayment and an initial delivery of 1.055 million shares, with final settlement due in the September quarter. Gross debt reached $5.582 billion at 30 June, against $248.4 million of cash. Net debt was therefore about $5.334 billion, equal to 5.6 times trailing free cash flow before the share-compensation adjustment (FICO Q3 2026). Gross debt divided by trailing operating cash flow was already 5.57 times. The monitoring issue is whether that ratio remains above 5.5 times at year-end, then fails to fall below 4.5 times during FY2027.

Debt does not threaten near-term survival. The franchise throws off cash, capex is low, and FY2026 guidance called for $850 million of GAAP net income. The capital structure still changes the equity maths. At $932.26 and the latest filed 21.598 million shares, equity value was about $20.135 billion. Adding net debt produces enterprise value near $25.47 billion. A change in enterprise value therefore lands on a smaller equity base than it did before the borrowing.

This is where capital allocation meets regulation. Repurchasing shares while the scoring standard remained dominant increased per-share exposure to continuing Scores pricing. It also left less room for error if competition changes that pricing. The 2026 proxy ties annual incentives to adjusted EBITDA, adjusted revenue and ARR; long-term awards include adjusted EBITDA, total shareholder return and environmental and social measures. The disclosed scorecard rewards growth and market performance, but it does not make mortgage concentration separately visible (FICO Proxy 2026).

The repurchases have succeeded mechanically in reducing the share count. They have not reduced enterprise risk. June debt issuance and the accelerated repurchase occurred weeks before every GSE lender received VantageScore access. Timing alone does not prove a poor decision, since management could not control the directive. It does show why a capital-allocation audit belongs inside the event analysis rather than after it.

Public filings cannot isolate the mortgage toll

FICO does not disclose mortgage Scores revenue. That is the most important missing number in the article. B2B Scores includes mortgage, auto, card and other lending uses, while B2C covers consumer score products. Public results reveal the segment margin and quarterly growth drivers, not mortgage revenue, unit volumes or the split between bureau price and FICO's own economics.

A sensitivity can still show the scale. Starting with trailing Scores revenue of $1.550 billion, an 89.7% segment margin, a 23.5% tax rate and 21.598 million shares, every 10 percentage points of total Scores revenue lost would reduce annualised EPS by about $4.92 before cost action or offsetting growth. This is an author stress test, not a forecast.

Assumed mortgage share of Scores Displacement within that pool Total Scores revenue loss Approximate EPS effect
40% 10% 4.0% $1.97
50% 25% 12.5% $6.15
60% 50% 30.0% $14.77

The middle row would reduce FY2026 guided GAAP EPS of $36.86 to roughly $30.71 before considering Software growth, non-mortgage Scores growth, pricing responses or cost reduction. The severe row would take it near $22.09. Conversely, no immediate displacement would leave current-year guidance largely intact while the risk moves into the multiple.

Competition can hurt in two ways. Unit share may move to VantageScore, or FICO may reduce price to preserve workflow share. The second path could produce a smaller model-share change but a wider profit effect because Q3 growth came mainly from higher mortgage score pricing. A lender might also use VantageScore for delivery while continuing to pull FICO elsewhere in its process, slowing revenue attrition. Monthly model share alone will not answer the full question.

The rival's claims deserve scrutiny too. VantageScore says its 4.0 model uses trended data, can score 33 million more adults and performed better than benchmark models in its 2025 assessment. Those are VantageScore-sponsored findings rather than independent loan-loss evidence (VantageScore Model 2026). Their policy force comes from access and cost arguments. Their economic force will come only when lender behaviour and mortgage performance confirm them.

Four adoption paths produce a wide valuation fan

FICO's value depends more on cash-growth duration than on the next quarter. FY2026 guidance of $36.86 GAAP EPS puts the post-fall price at 25.3 times guided earnings. Adjusted guidance of $42.43 puts it at 22.0 times, though that measure excludes $188 million of share compensation and related tax items. On trailing author-computed free cash flow, equity value is 21.0 times FCF; enterprise value is 26.5 times FCF (FICO Results 2026).

The primary valuation is a five-year equity-cash DCF because FICO is asset-light and operating cash flow is already after cash interest. It therefore discounts free cash flow to equity at a cost of equity and does not subtract net debt a second time. The model assumes existing debt can be refinanced while cash interest remains in operating cash flow; a forced principal repayment would reduce equity cash and belongs in the severe cases.

The scenario starting points come from operating drivers rather than the share price. Severe downside begins at $468 million to $642 million of equity cash: trailing FCF after the roughly $319 million after-tax effect of a 30% Scores revenue loss, with the low end also charging most share compensation. Bear begins at $654 million to $828 million after the approximately $133 million after-tax effect of a 12.5% Scores revenue loss, again varying the share-compensation treatment. Base uses $787 million to $961 million, spanning trailing FCF after all share compensation to reported trailing FCF. Bull starts at $961 million to $1.025 billion, requiring continued Scores pricing and initial Software conversion. All cases exclude new borrowing.

Case Five-year owner-cash path Discount / terminal assumptions Value per share
Severe downside down 2% to up 2%; fast substitution and broad mortgage price cuts 12%-13% / 1%-2% $160-$305
Bear up 5%-8%; adoption spreads and Software cannot offset Scores pressure 11%-12% / 2%-2.5% $350-$580
Base up 10%-13%; migration is gradual and FICO 10T retains a place 10%-11% / 2.5%-3% $595-$990
Bull up 15%-18%; VantageScore stays selective and Platform ARR converts 9%-10% / 3%-3.5% $1,070-$1,630

A multiple cross-check gives similar signposts. Eighteen times guided GAAP EPS is about $663; 22 times is $811; 26 times is $958. Applying those same multiples to adjusted EPS gives about $764, $933 and $1,103. The post-event price almost exactly matches 22 times adjusted guidance, but adjusted EPS removes a cost that matters to per-share economics.

The reverse equity-cash DCF is more revealing. At the $20.135 billion equity value, $960.9 million of trailing FCF must grow about 13.3% annually for five years under a 10.5% cost of equity and 3% terminal growth. Starting from $786.6 million after share compensation raises the required rate to about 18.4%. The market therefore continues to price double-digit cash growth after the fall.

The $932.26 close sits near the top of the broad base range, below the bull range and far above the bear ranges. That placement supports a roughly proportionate reaction verdict. The decline removed a premium that assumed longer mortgage pricing power, but it did not price a swift collapse of Scores. A clean overreaction claim would require better evidence that all-lender permission will remain concentrated. A clean underreaction claim would require evidence that lenders can replace FICO rapidly across workflow and investor use. Neither has arrived.

Three clocks now measure the moat

The first clock is monthly mortgage delivery. VantageScore share above 15% for two consecutive months, with adoption spreading beyond Rocket and UWM, would show that approval has become broad substitution. A flat share with continuing lender concentration would show that system and investor frictions still protect FICO.

The second is FICO's FY2027 reporting cycle. B2B Scores growth below 10% after mortgage unit-price effects normalise would be the first reported sign that volume or pricing pressure has reached the segment. A Scores margin below 85%, against 91% in the June quarter, would show that the profit pool itself is changing. Management's disclosure of price versus volume would matter more than the headline growth rate.

The third is formal model implementation. Fannie Mae and Freddie Mac must update their operating material, and FICO 10T needs a broad availability timetable. A genuine two-modern-model framework would be different from today's asymmetry, where VantageScore has immediate all-lender approval while FICO 10T has been validated and approved but is not yet broadly available for loan delivery. That policy sequence will determine whether FICO competes with a modern product or defends Classic FICO against one.

The housing cycle can blur all three. The US 30-year fixed mortgage rate was 6.71% on 3 September, keeping affordability strained and mortgage volumes sensitive to rates (FRED 2026). Rising originations could allow FICO revenue to grow even while unit share falls; weak originations could make stable share look like revenue pressure. Model share, score pulls per loan, unit price and absolute mortgage volume need to be read together.

Source notes: the missing disclosure is now the most valuable one

Verification is partial for one substantive reason, not because the filings were thin. The FY2021-FY2025 10-Ks, June 2026 10-Q, earnings release, proxy, regulator pages, market data and event coverage were retrieved and checked. The Finance API sidecar correctly resolved FAIR ISAAC CORP but was stale at 21 August for prices and 31 July for filings, so it could not verify the 4 September move or the later directive. Those facts were reconciled to Yahoo Finance, VantageScore, Reuters and the official policy pages (SEC 2026).

What remains unknowable from public reporting is the mortgage share of Scores revenue, the price-versus-volume split beneath the quarterly B2B result, and the number of lenders that will stop pulling FICO rather than add VantageScore selectively. Formal GSE operating pages also had not caught up with the director's announcement when fetched. These gaps widen the scenario ranges; they do not erase the event.

FICO entered 4 September with an 89.7% trailing Scores margin, fast mortgage price growth and $5.3 billion of net debt after aggressive repurchases. It left with every GSE lender allowed to use a competing model. The share price now assumes the franchise can still compound owner cash at a double-digit rate while its most profitable standard faces an open adoption test. Monthly delivery share, FY2027 B2B Scores economics and the FICO 10T timetable will show whether that assumption was reset enough.

References