INTERNAL FORMULA RESEARCH · generated and settled by our systems
INTERNAL FORMULA RESEARCH · generated and settled by our systems
| Formula | W–L | Priced | ROI | Units | CLV | vs break-even | Last pick | Picks |
|---|---|---|---|---|---|---|---|---|
| pinn_devig | 241–257 | 498 | -5.1%-13.8% to +3.6% | -25.5u | +8.32¢ | not yet separable | Oct 30 | Tonight → |
| consensus_devig CORE | 161–143 | 304 | +11.2%-0.8% to +23.2% | +34.1u | +5.57¢ | not yet separable | Oct 30 | Tonight → |
| priced_lineedges_floored | 256–241 | 497 | -3.7%-12.0% to +4.5% | -18.6u | +1.53¢ | not yet separable | Oct 30 | Tonight → |
| consensus_wnba | 1739–1292 | 3031 | +0.3%-2.8% to +3.4% | +9.5u | +1.42¢ | not yet separable | Sep 30 | Tonight → |
| model_matrix | 15242–11353 | 26595 | +1.3%+0.2% to +2.4% | +342.7u | +1.35¢ | beats break-even | Sep 30 | Tonight → |
| role_gated_board | 264–227 | 491 | -4.9%-12.8% to +3.0% | -24.0u | +1.35¢ | not yet separable | Oct 30 | Tonight → |
| form_wnba | 1626–1244 | 2870 | +3.1%-0.3% to +6.4% | +88.2u | +1.25¢ | not yet separable | Sep 30 | Tonight → |
| disciplined_board LIST | 846–643 | 1489 | +1.5%-3.1% to +6.0% | +21.8u | +1.22¢ | not yet separable | Oct 30 | Tonight → |
| system_lineedges | 1238–775 | 2013 | -1.8%-5.4% to +1.7% | -36.9u | +1.19¢ | not yet separable | Oct 30 | Tonight → |
| scout_ml | 1160–936 | 2096 | +0.8%-3.2% to +4.7% | +16.4u | +1.12¢ | not yet separable | Oct 18 | Tonight → |
| claude_combined | 230–198 | 428 | -1.7%-10.4% to +7.0% | -7.4u | +0.12¢ | not yet separable | Jul 18 | Tonight → |
| disciplined_llm | 118–159 | 277 | -15.3%-27.0% to -3.6% | -42.4u | +4.23¢ | losing | Aug 2 | Tonight → |
| pe_pretip | 6986–6986 | 13972 | -5.5%-7.2% to -3.9% | -771.9u | +1.21¢ | losing | Oct 24 | Tonight → |
| role_gated_lineedges | 852–436 | 1288 | -8.7%-12.6% to -4.9% | -112.2u | +1.13¢ | losing | Oct 30 | Tonight → |
| priced_lineedges | 1464–857 | 2321 | -3.2%-6.4% to -0.0% | -75.2u | +1.09¢ | losing | Oct 30 | Tonight → |
| both_agree_open · small sample | 10–3 | 13 | +43.5%-1.4% to +88.4% | +5.7u | +6.94¢ | not yet separable | Oct 30 | Tonight → |
| pinn_devig_open · small sample | 40–32 | 72 | +7.0%-15.5% to +29.5% | +5.0u | +5.50¢ | not yet separable | Oct 30 | Tonight → |
| both_agree · small sample | 22–30 | 52 | -22.6%-47.5% to +2.4% | -11.7u | +5.09¢ | not yet separable | Oct 30 | Tonight → |
| disciplined · small sample | 22–30 | 52 | -22.6%-47.5% to +2.4% | -11.7u | +5.09¢ | not yet separable | Oct 30 | Tonight → |
| pinn_devig_open_fade · small sample | 32–40 | 72 | -22.2%-42.6% to -1.8% | -16.0u | -5.79¢ | losing | Oct 30 | Tonight → |
| both_agree_fade · small sample | 21–4 | 25 | +53.1%+24.4% to +81.7% | +13.3u | -6.12¢ | beats break-even | Oct 30 | Tonight → |
Full-season records, flat 1 unit per pick. ROI is over priced bets only; CLV is the average closing-line value on close-captured picks — the leading indicator (an ROI far ahead of its CLV is running hot). Ranked by CLV first — beating the closing line is the more reliable signal at these sample sizes — with any formula whose ROI range sits entirely below zero moved to the bottom regardless of its CLV. Every number here is over PRICED bets only: picks posted without odds earn nothing whether they win or lose, so counting them inflates a record without ever moving the money. The range beside each ROI is a 95% interval; where it spans zero, the formula has not yet shown an edge distinguishable from break-even, however good the headline looks. With this many formulas compared, expect a couple to clear the bar by chance alone — which is why we do not claim one formula beats another.† marks a number inflated by unpriced signal rows: those picks earn nothing whether they win or lose, so they pad the win rate, and because an unpriced row carries a bet price of zero its recorded CLV is enormous — on the worst offenders most of the published CLV comes from a handful of such rows. Where you see †, read the ROI and the priced count, not the win rate or the CLV. CORE/LIST mark what the house card serves tonight. Records are shown for internal research comparison — including the losing rows.