250% more alpha
41x Sortino | 39% lower max drawdown
vs. traditionally selected models
Stop deploying models based on RMSE. Start optimizing for alpha with FIS (Forecast Intelligence Score).
| Model | FIS | MAPE | Sharpe | Max DD |
|---|---|---|---|---|
| ARIMA FIS Winner | 0.65 | 15.3% | 1.12 | -18.2% |
| Prophet Lowest Error | 0.51 | 0.09% | 0.87 | -22.4% |
| LSTM | 0.48 | 12.8% | 0.94 | -19.7% |
| XGBoost | 0.43 | 8.2% | 0.76 | -25.1% |
Traditional metrics recommend Prophet (0.09% MAPE — lowest error).
FIS recommends ARIMA (0.65 FIS - superior tail-event capture).
Prophet optimizes for baseline accuracy but misses volatility spikes. ARIMA has worse average error but captures tail events 27% earlier.
Rank trading models by decision utility (FIS) + risk-adjusted returns (Sharpe). Know which model to deploy before going live.
Pre-modeling dataset analysis. Get health scores, model recommendations, and preprocessing guidance before training.
End-to-end automated training, evaluation, and deployment for time series models.
CSV with your model predictions + actual prices. Works with any models you've trained (LSTM, Prophet, ARIMA, XGBoost, etc.).
FIS/CER for decision utility + Sharpe/Sortino/Drawdown for risk-adjusted returns. See which model wins and why.
Deploy the highest-ranked model. Our recommendations combine forecast quality with realistic trading performance.
Predictions + actual prices
FIS, Sharpe, Drawdown, model ranking
Highest utility, lowest risk