No-CDN static report. Rebuilt with next-tradable-day signal execution after the bias review.
The backtest now applies signals on the next tradable day rather than the same close. This reduced the best passing candidate's net Sharpe from 2.06 to 1.70, but it still passes the bias gate. Other candidates remain blocked by the shift test and should not be promoted.
| Strategy | 3M Return | 3M Sharpe | 3M MaxDD | 3Y Sharpe | 3Y MaxDD | Net Sharpe | Bias | Max Corr | Latest Positions |
|---|---|---|---|---|---|---|---|---|---|
| Low-Corr Story Overlay | +36.09% | 2.65 | -13.31% | 2.26 | -34.42% | 1.89 | FAIL | 0.65 | UUP:20.00%,LITE:13.60%,CIEN:13.60%,MRVL:13.60%,STX:13.60%,MU:12.80%,WDC:12.80% |
| Infra Bottleneck Barbell | +39.49% | 3.05 | -12.08% | 1.97 | -37.16% | 1.70 | PASS | 0.70 | AMD:11.67%,UUP:10.00%,GRID:10.00%,VRT:10.00%,PWR:10.00%,MRVL:8.33%,AVGO:8.33%,CIEN:8.33%,WDC:6.67%,STX:6.67%,DELL:5.00%,HPE:5.00% |
| Static Story Equal | +19.54% | 2.63 | -11.01% | 1.94 | -30.72% | 1.89 | FAIL | 0.82 | ANET:10.00%,AVGO:10.00%,BOTZ:10.00%,EWT:10.00%,GRID:10.00%,MU:10.00%,NVDA:10.00%,PLTR:10.00%,UUP:10.00%,VRT:10.00% |
| Theme Rotation | +17.70% | 1.61 | -20.75% | 1.71 | -33.72% | 1.50 | FAIL | 0.67 | AMD:25.00%,DELL:25.00%,HPE:12.50%,SOXX:12.50%,STX:12.50%,WDC:12.50% |
| Physical AI Rotation | +13.00% | 1.42 | -17.03% | 1.36 | -25.60% | 1.24 | FAIL | 0.68 | IRBO:22.00%,CAT:22.00%,ROBO:22.00%,ROK:19.21%,TER:14.79% |
| Strategy | Bias | Net Sharpe | Net CAGR | Net MaxDD | OOS Sharpe | Max Corr | Corr Gate | Paper |
|---|---|---|---|---|---|---|---|---|
| Low-Corr Story Overlay | FAIL | 1.89 | +68.96% | -34.57% | 3.87 | 0.65 | PASS | needs_input |
| Static Story Equal | FAIL | 1.89 | +61.10% | -30.72% | 3.05 | 0.82 | FAIL | needs_input |
| Infra Bottleneck Barbell | PASS | 1.70 | +58.28% | -37.32% | 3.38 | 0.70 | PASS | needs_input |
| Theme Rotation | FAIL | 1.50 | +54.41% | -33.90% | 2.74 | 0.67 | PASS | needs_input |
| Physical AI Rotation | FAIL | 1.24 | +30.34% | -25.73% | 2.45 | 0.68 | PASS | needs_input |
| Theme | Story | Example Basket |
|---|---|---|
| AI factories | AI as infrastructure; energy/chips/infrastructure/models/apps | NVDA, AVGO, AMD, ANET, VRT, DELL |
| Optical networking | CPO/silicon photonics for AI factory scale-out | ANET, AVGO, MRVL, CIEN, COHR, LITE |
| HBM / storage | Memory bandwidth and AI data platforms | MU, WDC, STX, HPE, DELL |
| Power / cooling / grid | Power, cooling, grid optimization for AI factories | VRT, ETN, PWR, CEG, VST, GRID |
| Physical AI | Robots, autonomy, digital twins, industrial automation | BOTZ, ROBO, IRBO, TER, ROK, ISRG |
| Agentic software | Agents as enterprise software interface | MSFT, META, GOOGL, PLTR, NOW, CRM |
{
"GTC_2026_stack": "NVIDIA Newsroom 2026-03-03: five-layer stack: energy, chips, infrastructure, models, applications.",
"GTC_2026_physical_ai": "NVIDIA Newsroom 2026-03-16: physical AI and robotics ecosystem.",
"Computex_2025_ai_factories": "NVIDIA Blog 2025-05-18: AI factories, agents, robotics, Taiwan supply chain.",
"CES_2026_physical_ai_agents": "NVIDIA Blog 2026-01-05: Rubin, agentic AI, physical AI, autonomy.",
"GTC_DC_2025_DSX": "NVIDIA Blog 2025-10-28: DSX blueprint for gigawatt-scale AI factories, power/cooling/grid.",
"GTC_2025_CPO": "optics.org 2025-03-19: co-packaged optics and silicon photonics for AI factories."
}