Trident Pulse Machine Learning Algorithm
Model Overview
Trident Pulse is an XGBoost model trained on 2 months' worth of trade-level data from Tesla (July–August 2024). This model uses a pool of over 700 indictors, covering a wide range of statistics including volume, momentum, percentiles, market structure, time between events, and more. These features are then aggregated over multiple periods to give the model the best chance of forecasting future market moves.
Performance And Use Notes
This model is designed for and performs best in trending market conditions with high volume. In such scenarios, it has on multiple occasions produced results with a Sharpe ratio greater than 1.5 after heavy trade costs have been factored in. In sideways market conditions, however, the model tends to perform poorly and often breaks even or incurs losses.
