Live simulation
Paper Account — what if you'd followed every pick?
Every account starts with $10,000, and each model's curve begins on the day its first pick was time-stamped — models launched on different dates, so start dates vary. From that point they compound forever — never a rolling window, so winning bets stay in the curve and the bankroll only ever grows for a profitable model. Stakes are 1% of current bankroll (flat), ¼-Kelly (sized by model edge) or ½-Kelly (aggressive). Every price gets a 2% slippage haircut for what a real user actually gets after the line shortens, and a 20% live-exposure capstops a single slate eating the bankroll. Settled from real results, updates every minute.
Starting bankroll: $10,000 per modelTool accounts: 1% bankroll stake¼-Kelly · 2% max2% slippage · 20% exposure cap
Loading paper accounts…
How to read this page
- Flat 1% stakes 1% of the current (compounding) bankroll on every pick. ¼-Kelly sizes each bet by model edge (f* / 4, capped at 2% of bankroll) and skips bets where edge vanishes after slippage. Both use the same slippage-adjusted price (published × 0.98).
- ROI% = net profit ÷ total dollars staked. A healthy long-run number is +2% to +5%.
- Avg CLV is the headline sharpness metric — positive means we beat the closing line on average. Stabilises in ~50 picks; ROI takes 500+.
- Max DD is the worst peak-to-trough drop. Anything under 15% is healthy for a flat-staked book; over 25% means the staking is too aggressive. Sharpe / Sortino are risk-adjusted return — 1.0+ is good, 2.0+ is excellent (Sortino only penalises downside vol).
- Exposure cap rejects a pick if live (unsettled within 6h) stakes would exceed 20% of the bankroll. Stops a Saturday AFL slate from concentrating risk.
- Pending picks (kickoff still ahead) appear in the open-bets list and don't move the bankroll yet.
Research output only — not bet advice. See responsible gambling.