LiveEV
MLB Live trader
Real results, in units — see homepage for why we don't show raw dollar amounts.
Since inception
Units over time
Win rate by purchase price
| Price | Win rate | Trades |
|---|---|---|
| 40-49¢ | 83.3% | 6 |
| 50-59¢ | 56.8% | 88 |
| 60-69¢ | 67.3% | 171 |
| 70-79¢ | 78.4% | 153 |
| 80-89¢ | 84.0% | 144 |
| 90-99¢ | 94.1% | 289 |
| 100-109¢ | 100.0% | 7 |
| 130-139¢ | 100.0% | 1 |
Model calibration
Three different questions, side by side: is the model's confidence accurate (Delta), does the strategy make money at each confidence level (Margin), and what the average trade in each bucket actually returned, after Kalshi fees (Return). These aren't the same thing on purpose — the bot only ever buys when the price paid sits meaningfully below the predicted probability, so Margin is the number that determines profit, and Return puts every bucket on the same footing (Margin alone makes higher-price buckets look thinner than they really are, since it's not normalized by the money staked). A small red Delta with solidly positive Margin/Return is normal and expected, not a warning sign.
Trade counts per bucket are still small enough that a Delta under ~10pp is within normal statistical noise either way.
| Predicted | Actual win rate | Trades | Avg price paid | Margin | Return after fees | Delta |
|---|---|---|---|---|---|---|
| 60-64% | 61.9% | 21 | 53.3¢ | +8.6pp | +17.7% | -0.6pp |
| 65-69% | 56.0% | 100 | 57.9¢ | -1.9pp | -5.8% | -11.5pp |
| 70-74% | 74.8% | 119 | 65.2¢ | +9.6pp | +12.5% | +2.3pp |
| 75-79% | 72.2% | 90 | 70.5¢ | +1.7pp | +1.0% | -5.3pp |
| 80-84% | 72.7% | 77 | 75.7¢ | -3.0pp | -5.1% | -9.8pp |
| 85-89% | 82.3% | 62 | 81.7¢ | +0.6pp | +0.1% | -5.2pp |
| 90-94% | 89.1% | 193 | 89.5¢ | -0.4pp | -0.8% | -3.4pp |
| 95-99% | 95.9% | 197 | 95.9¢ | +0.0pp | -0.2% | -1.6pp |