Every resolved signal is counted. Wins and losses. The numbers below update automatically when games resolve. There is no manual entry.
| Sport | Tier | Resolved | Wins | Win Rate | Avg CLV | Note |
|---|---|---|---|---|---|---|
| ⚾ MLB | SNIPER | 8 | 5 | 62.5% | +20.0pp | Highest edge, lowest volume |
| ⚾ MLB | STRONG | 1,250 | 747 | 59.7% | +8.4pp | Primary signal tier |
| ⚾ MLB | ELITE | 167 | 44 | 26.3% | +21.3pp | ⚠ Under audit |
| ⚾ MLB | WATCH | 920 | 463 | 50.3% | −4.6pp | Informational tier |
| 🏒 NHL | CORE | 39 | — | — | — | <5 per tier; settling |
| ⚽ Soccer | — | 0 | — | — | — | WC 2026 starts Jun 11 |
| 🏈 NFL | — | 0 | — | — | — | Preseason Aug 2026 |
Min 5 resolved signals to show win rate. CLV = Closing Line Value. ELITE tier 26.3% WR is actively under model audit — displayed honestly, not hidden.
Win rate tells you what happened. CLV tells you if your edge was real.
If you got +145 on a bet that closed at +120, you had positive CLV. You beat the market's final informed price. That's a real edge regardless of outcome.
A 45% win rate with +8pp avg CLV is a better long-term result than a 60% win rate at −2pp CLV. CLV separates luck from skill over large samples.
The model records the odds at signal generation and at market close. Every resolved signal shows both. Use /clv in the bot to see yours.
When a game ends, the resolver polls official box score data (BallDontLie API for player props, official stats APIs for match outcomes).
Signal IDs are UUID hashes generated at creation time. They cannot be edited retroactively. You can verify any signal ID directly in the bot with /wincard <id>.
The resolver doesn't know which outcome to "prefer." WIN/LOSS is written directly from official data. PUSH is tracked separately. Postponements are re-queued, not deleted.
Losses are never deleted. The ELITE tier 26.3% WR anomaly is shown on this page in full. We don't publish only the tiers that look good.
/modelwallFull model performance by sport/tier/wincard <id>Verify any individual signal/accuracyPlatform accuracy wall (7/30/90 day)/clvYour personal CLV breakdown/simulateBacktest on all resolved history/consensusCommunity count on active picksEvery signal on this platform is independently verifiable. No trust required — check the data yourself.
a3f7b91c). This ID is generated at creation time and cannot be changed retroactively./wincard <id> in the bot/accuracy 30d for platform-wide audit/simulate to backtest the entire archiveRun /modelwall to see the current model wall with all tiers, or /simulate to backtest on 2,392 resolved signals.