The Score Wars: FICO, VantageScore, and the Price of the Number

For decades, one company's algorithm was legally required at the gate of the largest consumer credit market on earth — every conforming mortgage, three FICO scores, no substitutes. Monopolies at mandatory gates price like monopolies at mandatory gates: the number's wholesale cost compounded at rates that eventually summoned Senate letters, and then the gate itself swung open — two new models admitted, bureau counterattack pricing landing within seventy-two hours, and nine figures of projected savings. But the war is stranger than the headlines: the challenger is owned by the bureaus, the incumbent blames the middlemen, and the report wrapped around the cheapening score is getting more expensive. This is the full anatomy of the fight over the number.

The core thesis

The score wars are the purest case study in this research series' oldest theme: whoever owns a mandatory checkpoint owns an annuity. Interchange, the bureaus' data resale, the tri-merge — and above them all, the score itself: a number computed from data the scorer doesn't even own, required by regulation on every conforming loan, priced accordingly. What makes this war analytically rich is that it isn't competition arriving by market forces — no lender could unilaterally switch — but by regulatory decree, two decades after the data itself commoditized. And the combatants scramble every clean narrative: the challenger, VantageScore, is owned by the three bureaus — an oligopoly attacking a monopoly from below, as we foreshadowed in the bureau economics report; the incumbent's defense is that the bureaus' reseller markups, not its royalties, drove costs; and the referee is a housing regulator explicitly wielding score approval as a price-negotiation lever.

Our thesis: the war's first-order effects — score-component prices collapsing toward zero, tens of millions of newly scoreable borrowers at the mortgage gate — are real and consumer-positive. But the second-order effects deserve equal billing: costs don't die at checkpoints, they migrate (the tri-merge bundle is already repricing upward as the score reprices down), and dual-model lending introduces a genuinely novel risk — model arbitrage, lenders routing each borrower to whichever score flatters them, with separate pricing grids formalizing the choice. Competition at the gate is better than monopoly at the gate; it is not the same thing as a cheaper or safer gate. The next credit cycle will grade the models; the next pricing cycle will reveal where the toll actually went.

The number cost what it cost because the law said you had to buy it. The law just stopped saying that — and every party at the gate is now repricing at once.

The gate: how one score became mandatory

The mechanics of the monopoly were regulatory, not technical. Fannie Mae and Freddie Mac — the conforming market's buyers of last resort, standing behind more than $8 trillion in mortgage funding — required "Classic FICO" scores from all three bureaus (the tri-merge) on essentially every loan they purchased. Lenders originating to sell had no choice; a mandated input at the industry's chokepoint. Congress noticed first: the Credit Score Competition Act of 2018 ordered a validation process for alternative models, and in October 2022 the regulator validated two — FICO 10T and VantageScore 4.0, both trended-data models — kicking off an implementation saga that consumed years, floated and shelved a "bi-merge" (two bureaus instead of three, fiercely resisted), and finally cracked open in mid-2025 when the housing regulator announced lenders could deliver VantageScore 4.0-scored loans. The formalization came in April 2026: a joint announcement extending both new models to FHA lending, with the GSEs accepting VantageScore 4.0 immediately from approved lenders in a limited rollout, FICO 10T's historical data publishing for evaluation, separate pricing grids per model, and — crucially — lender choice, one model per loan. The single-score era, after roughly three decades, is formally over; what replaces it is still being negotiated in public.

The price of the number

The provocation was arithmetic. FICO's wholesale mortgage royalty escalated from a few dollars to $4.95, then to $10 per score in the most recent jump — a doubling in a single year for, as one senator's office put it, the identical product, with compound growth over five years that the same office pegged near 100% annually and a single-year industry cost estimate around $500 million. Multiply by three scores per tri-merge, per applicant, across every application including the majority that never fund, and the toll's scale explains the politics. FICO's defense deserves its steelman, because it contains a real insight: the company argues bureau resellers historically marked scores up ~100% on the way through — the monopolist pointing at the oligopoly's distribution margin — and responded with a direct licensing program and a "performance pricing" option ($4.95 per score plus $33 per funded loan, shifting cost from shopping to closing). Note what both framings concede: the number's price bore no meaningful relation to its marginal cost, which rounds to zero. That's not an accusation; it's the economics of mandatory checkpoints, and it's precisely why the fight was always going to be resolved by whoever controls the mandate rather than by either combatant.

The counterattack — and the recapture

The bureaus' response to the April opening was a coordinated pricing strike: within seventy-two hours, VantageScore 4.0 component pricing landed at $0.99 (TransUnion), free indefinitely (Experian, with a public pledge to stay at least 50% below FICO if it ever charges), and $4.50 with free bundling through 2026 (Equifax). Industry analysis projects $900 million-plus in annual savings from the score-component repricing alone — and the regulator has said publicly it's pushing both scorers toward 99 cents. Textbook competition. Now the recapture, which the celebratory coverage mostly skipped: the tri-merge report bundle — the three full credit reports the scores ride on — jumped from roughly $33.50 to $47.05 year over year, with some reseller pricing far higher. FICO predicted exactly this move: bureaus losing score-distribution margin would recover it through data fees, because the reports remain as mandatory as the score used to be (the bi-merge that would have weakened that leg was shelved). The system-level read: the toll didn't disappear — it migrated one layer down the stack, from the algorithm to the data, where the oligopoly's position is strongest and least contested. Total credit costs per mortgage may still fall on net; but anyone modeling this fight as monopoly-defeated-prices-fall is reading one layer of a three-layer repricing.

Who the new models see

Beneath the pricing war sits the substantive difference: what the models can see. VantageScore 4.0 uses trended data plus expanded inputs — including rent and other non-traditional payment history where furnished — and scores tens of millions of people classic models cannot: the thin-file, new-to-credit, and recently inactive populations we mapped in the credit invisibility report. FICO 10T reads twenty-four months of trended behavior with different construction. The mortgage-market translation is profound: applicants who were unscoreable at the gate — disproportionately young, immigrant, and minority households with perfect rent histories and empty files — become underwritable, which is why the political framing on all sides has centered on renters. For the credit-building world this is the strategic tailwind we've tracked all series: every furnished tradeline, every reported rent payment, every month of clean Metro 2 data is worth more when the models reading it are hungrier and the gates accepting those models are taller. The caveat belongs in the same paragraph: being seen cuts both ways — trended models read your balance trajectories, not just snapshots, so the newly visible are also the newly legible, in both directions.

The risks nobody's pricing yet

• Model arbitrage. Lender choice plus separate pricing grids means the same borrower can carry different scores — and different loan prices — under each model. Rational lenders will route borrowers to the flattering model; rational risk officers should ask what systematic selection that creates in each model's book. Adverse selection between scoring systems at mortgage scale is genuinely novel territory.
• Cycle-untested at the gate. Both models passed validation on historical data, but neither has priced a full national housing downturn as the score of record. The 2008 lesson — models perform until the regime changes — applies squarely.
• The governance tangle. The challenger is owned by the three companies that own the data both scorers depend on and that price both scores' distribution. That's not an alleged conspiracy; it's a structure — one where the data layer wins every scenario, which the tri-merge repricing already demonstrates.
• Operational drag. Dual models, dual grids, historical-data evaluation, limited rollouts: the transition costs land on lenders now, in exchange for competition benefits that accrue over years — a timing mismatch that will tempt some to wait it out in Classic FICO, slowing the very competition the opening was meant to create.

Scenarios and what we're watching

What we're watching: VantageScore 4.0's actual share of delivered loans (announcements aren't adoption); FICO 10T's move from historical data to live delivery; the tri-merge bundle price (the recapture gauge); early delinquency performance by model — the arbitrage detector; whether score competition escapes the mortgage silo; and the 99-cent question — whether regulatory jawboning gets both numbers to a dollar, completing the strangest journey in credit: the most valuable number in American finance, priced at last like what it computationally is. The score was never the moat; the mandate was. Now that the mandate is plural, everyone in the stack — scorer, bureau, lender, borrower — is about to find out what the number was actually worth.

Frequently asked questions

This report is for general information only and does not constitute financial or investment advice. Pricing, rollout details, and regulatory positions in this market are changing rapidly; figures are drawn from publicly reported sources including regulator announcements and congressional correspondence.