HUNCH LAB.
The outcome is unforecastable — but what's profitable isn't. The games are parimutuel with no house, so positive EV lives wherever the crowd prices an outcome below its true odds. Find that gap here, then deploy it for real.
Pick an edge
Tune it
The honest verdict
Pulling the Tape…
Found an edge? The template you tuned is the strategy — the same pure code runs live. Take it to the Cup wheel and play it with paper $pUSDC; top 50 on the board split $5,000.
Run this in the Cup →Prefer to bring your own brain? Pull the Tape from /api/lab/v1/tape and score with scoreBetsAgainstRound — see the Lab docs.
How the Lab works
- The Tape is every settled round's honest record — the winning pocket, the full pool distribution, and the wheel's true k/37 odds. Outcomes alone only fuel streak-hunting; the pools reveal where the crowd mispriced.
- A strategy reads the pools + true-odds and emits bets. The templates above are named real edges — and each compiles to the exact same code the deploy path runs, so there's no rewrite between backtest and live.
- The Scorer injects your bets into each round's real historical pool and settles with the production math, so a backtest pays exactly what the live wheel would — dilution and refunds included.
- Watch the market-impact flag: past ~10% of a pool, the backtest is an upper bound because your real size would have moved the crowd.