Event Crumb Strategy Eval
TL;DR
event_ai_readthrough_power_overlay_v2 is the strongest iteration. It improves net Sharpe and drawdown versus the first-pass AI read-through strategy while still passing the correlation gate. I would not launch it live yet because net Sharpe is still below 1.5 and paper evidence remains missing, but it is now the best paper/live-shadow candidate from this family.
Equity Curves
Strategy-Eval Ranking
| Strategy | Bias | Net Sharpe | Net CAGR | Net MaxDD | OOS Sharpe | Max Corr | Corr Gate | Paper |
|---|---|---|---|---|---|---|---|---|
| Event AI Read-Through Power Overlay v2 | PASS | 1.37 | +32.66% | -23.92% | 3.21 | 0.68 | PASS | needs_input |
| Event AI Read-Through | PASS | 1.22 | +32.70% | -30.90% | 3.19 | 0.69 | PASS | needs_input |
| Event AI Read-Through Drawdown Guard v2 | PASS | 1.21 | +31.07% | -25.78% | 3.04 | 0.67 | PASS | needs_input |
| Event Crumb AI/Power/Nuclear | PASS | 1.20 | +31.26% | -40.25% | 2.77 | 0.94 | FAIL | needs_input |
| Event AI Read-Through Strict Leader v2 | PASS | 1.11 | +21.75% | -14.43% | 1.98 | 0.39 | PASS | needs_input |
| QSD AI Power Barbell Ref | PASS | 1.06 | +26.95% | -49.57% | 3.38 | 1.00 | FAIL | needs_input |
| Event Nuclear Deadline | PASS | 0.83 | +17.69% | -28.43% | 1.21 | 0.67 | PASS | needs_input |
| Event AI Read-Through Low Vol v2 | PASS | 0.79 | +16.22% | -34.79% | 2.38 | 0.67 | PASS | needs_input |
Backtest Snapshot
| Strategy | 3M Ret | 3M Sharpe | 3M MaxDD | 1Y Sharpe | 3Y Sharpe | Corr QSD | Latest Positions |
|---|---|---|---|---|---|---|---|
| QSD AI Power Barbell Ref | +39.95% | 3.89 | -11.17% | 3.13 | 2.03 | 1.00 | GRID:22.00%,UUP:20.00%,DELL:14.50%,AMD:14.50%,HPE:14.50%,MU:14.50% |
| Event Crumb AI/Power/Nuclear | +25.85% | 2.75 | -13.90% | 2.41 | 1.98 | 0.94 | GRID:26.83%,UUP:18.29%,AMD:11.25%,DELL:11.25%,HPE:11.25%,MU:11.25%,SOXX:9.88% |
| Event AI Read-Through Power Overlay v2 | +25.90% | 2.92 | -9.74% | 3.06 | 1.85 | 0.68 | UUP:20.00%,GRID:18.00%,DELL:15.50%,AMD:15.50%,HPE:15.50%,MU:15.50% |
| Event AI Read-Through | +40.67% | 3.62 | -10.19% | 3.06 | 1.72 | 0.69 | AMD:17.43%,DELL:17.43%,HPE:17.43%,MU:17.43%,SOXX:15.30%,UUP:15.00% |
| Event AI Read-Through Drawdown Guard v2 | +40.67% | 3.62 | -10.19% | 3.06 | 1.69 | 0.67 | AMD:17.43%,DELL:17.43%,HPE:17.43%,MU:17.43%,SOXX:15.30%,UUP:15.00% |
| Event AI Read-Through Strict Leader v2 | +14.82% | 1.96 | -10.67% | 1.84 | 1.51 | 0.39 | UUP:100.00% |
| Event Nuclear Deadline | -8.39% | -1.04 | -17.49% | 0.70 | 1.46 | 0.68 | GRID:80.00%,UUP:20.00% |
| Event AI Read-Through Low Vol v2 | +24.53% | 3.04 | -9.88% | 2.26 | 1.33 | 0.67 | UUP:28.00%,AMD:15.12%,SMH:15.12%,HPE:15.12%,SOXX:15.12%,DELL:11.52% |
Current Information Catalysts
The current event set uses NVIDIA's May 20, 2026 AI data-center confirmation, Broadcom's June 3, 2026 earnings date, Micron's June 24, 2026 earnings date, and the DOE Reactor Pilot Program's July 4, 2026 target as live catalysts.
Main caveat: this first pass backtests price-action proxies for event crumbs, not a full historical news/NLP database. That is why the conclusion is research/paper, not live promotion.