Backtest
Simulate signal logic without confusing it for proof.
Real model calibration from live resolved markets, then a deterministic simulated strategy cohort to demonstrate how the signal gates behave. Simulation is machinery evidence, not a performance promise.
Backtest lab
Simulation with live calibration beside it
Live Brier
0.003
10 resolved
Sim trades
42
EV gated
Sim P/L
+$14,392
not proof
ELECTIONS
5 trades60% favorable
COMPANY
4 trades75% favorable
CRYPTO REGULATION
6 trades50% favorable
Backtest
Real model calibration from live resolved markets, then a simulated strategy P&L showing how the signal logic would perform on a synthetic cohort.
Backtest context
This page intentionally separates live calibration from simulated strategy machinery. It is useful for testing gates and consistency, but the public proof still lives in the timestamped track record.
Engine calibration - live markets
Building: 10 of 30 required resolved predictions. Reliability diagram appears at 30. Track record
Signal calibration - simulationSIMULATED DATA
Deterministic synthetic cohort (seed 42, n=160) demonstrating the calibration machinery. Not a performance forecast.
0.182Brier score on 160 simulated resolutions
Strategy simulation - trade the engine's signalsSIMULATED DATA
Equal $1,000 stake on every simulated market where EV after fees and slippage cleared the 5% hurdle - the same gate the live signal classifier uses.
42
trades
of 160 markets
47.6%
favorable rate
simulated outcomes
+34.3%
avg return
net of modeled costs
+$14,392
paper P&L
simulation only
$12,434
max drawdown
peak to trough
0.96
consistency
not annualized
Performance by categorySIMULATED DATA
| Category | Trades | Favorable rate | P&L |
|---|---|---|---|
| ELECTIONS | 5 | 60% | +$6,475 |
| COMPANY | 4 | 75% | +$4,578 |
| CRYPTO REGULATION | 6 | 50% | +$3,310 |
| GEOPOLITICS | 9 | 22% | +$2,504 |
| FED | 4 | 75% | +$1,873 |
| MACRO | 3 | 67% | +$1,588 |
| CRYPTO | 3 | 67% | -$133 |
| CPI INFLATION | 4 | 25% | -$2,865 |
| AI REGULATION | 4 | 25% | -$2,939 |