The same premise as the stock side, tested where the evidence is far stronger. Sports and prediction markets resolve in days, so a few months buy thousands of scored, out-of-sample predictions — real statistical power, instead of 113 monthly rebalances. The finding so far is the same one: liquid markets are priced well enough that a public-data model does not reliably beat them.
The actual upcoming fixtures across five sports, with their live order books — price, spread and depth right now, keyless from Polymarket. No model and no picks: the data the investigation is about, so the “priced too well to beat” claim isn't something you take on faith.
Real win-probability models for the NFL and MLB, tested out-of-sample and held to the market's price. Football rewards passing over the line; baseball is nearly a coin flip. Neither model beats the market — the same verdict, two fresh directions.
Real football data, not markets — every game from 2019 to 2025 with its full box score, plus team-season stats (record, yards, turnover margin). Keyless from ESPN's public data.
Standings and games for 2024–2026 — record, runs for and against, run differential, and every game's starting-pitcher matchup. Keyless from ESPN's public data.
Every venue's moneyline for the same game on one axis, over time — Polymarket, Kalshi, and a sportsbook's de-vigged line. Gaps surfaced, executable locks computed net of fees, and the price history charted per venue.
Paper bets, hedges and parlays the strategy engine issues when it predicts a net-of-cost edge — entry frozen at issue, graded against the closing line and the outcome, in public. No real money, and no strategy has earned belief yet; the settled record is the test.
Can a keyless model out-price Polymarket (no), and is the market itself mispriced enough to arbitrage (no). A Poisson forecaster scored against thousands of resolved markets, favorite-longshot and arbitrage tests broken out by sport, and the calibration to show it — the honest null result, in full.
A tool for comparing lines across books and reading prop markets the way the stock screener reads filings — descriptive, math shown, no picks sold as certainties. It is being built; it is not here yet, and this card will become a link when it is, not before.
