Valuo.
A soccer player-value tracker across eleven leagues, men's and women's: its own valuation model, backtested against real transfer fees, flagging who the market has wrong.
PublicValuo treats player value like a market. Around 5,000 players across nine men's leagues (the Big 5, the Eredivisie, Liga Portugal, the Championship and MLS) plus the WSL and NWSL each get two independent reads: a Valuo Index computed from underlying performance, and a Market Index from what the market prices them at. When the two disagree by enough, the player is flagged as undervalued or overvalued.
The model scores per-90 production, availability, and an age curve for resale value. Each rate is shrunk toward a below-average prior until a player has around 900 minutes, so a single good cameo cannot top the board. Both indices are percentiles within league and position, which keeps every comparison like-for-like. An earlier version on mismatched scales made every star look overvalued, and fixing that was the most important correction in the build.
The two data sources needed two models. The Premier League runs on the public Fantasy Premier League API, which has expected goals and live crowd-set prices. The other leagues come from API-Football, which has neither, so that model leans on per-90 output and match ratings, and its market side is real transfer fees, compared only within the group of players clubs have actually paid for. Players without a recent fee get a value but no market reading, rather than an invented one.
It is backtested in public. Premier League players flagged undervalued at season end saw their market value rise by 16.5 percentile points the following season, against an 8.7-point fall for those flagged overvalued. Across 1,000 real transfers, the rating before a move ranks the fee actually paid with a correlation of 0.40, against 0.25 for a goals-and-assists baseline. A regression trained on those fees gives every player a euro estimate (R² 0.58 on transfers it never saw; European-competition minutes earned their place by cross-validation, injury history did not), and a league-strength adjustment makes ratings comparable between leagues. It was rebuilt after the first version, fitted on players who switched leagues, ranked MLS stars among Europe's best: movers are a hand-picked sample, so they understate the gaps. The current version measures league strength from European competition, where players from different leagues face the same opponents, and fills the leagues that never play in Europe from transfer fees corrected for the market's measured wealth bias. On held-out transfers it ranks fees across leagues at 0.67, up from 0.59.
Nightly GitHub Actions sync both sources and commit snapshots, which redeploys about 3,900 prerendered pages. Accounts (Better Auth on Neon Postgres, Google or email sign-in) sync watchlists and squads across devices, and a daily job sends each watchlist alert once, by email through Resend and by web push.
On top of that sit model-versus-market trend charts, a similar-player search that finds cheaper alternatives in any league, a wonderkid radar, a Director of Football view of each club's gaps and targets, a budget squad builder that solves the selection exactly as a knapsack problem, head-to-head comparisons, a live FPL price-change watch, an FPL transfer advisor, a transfer-window tracker, a weekly report and newsletter, and a public, documented JSON API.
Women's football is kept deliberately separate: there is no shared transfer market and the WSL has no recorded fees, so women's players get ratings, trends and projections but no euro estimates, and cross-league rankings never mix the two pools. An hourly monitor on GitHub Actions checks the sites and Vercel deploys independently of Vercel itself, and every page passes an automated WCAG 2.1 AA audit in both themes.