-
This is my financial trading system using ML.
-
See Notebook
-
The other example using daily stocks: Project
Momentum strategy with machine learning
-
- Form time/dollar bars with tick data
-
Getting Trading Signals
- Momentum strategy (RSI..)
- Additional ML regime detector
-
Trading Rules
- Enter rules with trading signals
- Exit rules (triple-barrier method)
- Binary Labeling (Profit or Loss)
-
Strategy Enhancing ML Model
-
Get Features (X)
- Market data & Technical analysis
- Microstructure features
- Macroeconomic variables
- Fundamentals
- public sentiments with NLP
-
Feature Engineering
- Feature scaling
- Dimension reduction
- Feature Analysis with feature importance
- Feature selection
-
Machine Learning Model
- Cross-validation (time-series cv / Purged k-fold)
- Hyperparameter tuning
- AutoML with autogluon and select the best model
- Results (accuracy, f1 score, roc-auc)
- Trading
- Bet Sizing
- Trading Simulation
- Results
- Cumulative returns, Sharpe Ratio, max drawdown
- ETH/USD 5 min data (2019.1.1 ~ now)
- The trading rule is based on Triple-Barrier Method introduced in Lopez De Prado (2018).
- Advances in Financial Machine Learning, Lopez de Prado (2018)
- ta, https://github.com/bukosabino/ta
- autogluon, https://github.com/awslabs/autogluon