All dashboards, latest picks, TP/SL, and historical performance in one place
This hub aggregates 12+ trading systems. Here's what the ML terms mean:
| Term | What It Means (Simple) |
|---|---|
| XGBoost | Hundreds of tiny decision-makers vote together. Like asking 200 people and going with majority. Used by Mercury 2 (3 models) and ML Battleground. |
| LightGBM | Similar to XGBoost but faster. Builds decision trees one leaf at a time. Used by Crypto ML Edge (16 per-coin models) and Mercury 2 (gainer predictor). |
| Random Forest | 100 separate decision trees, each trained on random slices of data, then vote. Like a jury with 100 members. Used by KIMI, Claude Gainer, Alpha Engine. |
| GRU Neural Net | Artificial brain that reads price history in sequence (like reading a story). Remembers important moments, forgets noise. Used by ML Battleground System C. |
| DSR Gate | Deflated Sharpe Ratio — a statistical test asking "is this model's performance real or just luck?" Models must pass to generate picks. |
| Heuristic mode | Rule-based fallback when ML hasn't trained yet. Uses simple formulas (win rate, Sharpe) instead of a trained model. Upgrades to ML automatically. |
| Walk-forward | Training on past data, testing on future data the model never saw. Prevents "memorizing" history. Like studying chapters 1-8 then taking a test on chapter 9. |
Each system card below shows its ML methodology and validation status. Green badges = model passed validation. Red = needs retraining.
| Rank | System | Category | Win Rate | Sharpe | Active Picks | Quality |
|---|---|---|---|---|---|---|
| Loading leaderboard... | ||||||