Financial models (fynance.models)¶
Several financial models — deep learning, econometric and statistical architectures.
Attention
Scaled dot-product and multi-head attention modules for Transformer-based architectures.
Econometric models
Time-series models: MA, ARMA, ARMA-GARCH, ARMAX-GARCH.
Neural network models
Multi-layer perceptron and base class for PyTorch neural network models.
Recurrent neural networks
RNN, GRU and LSTM models with walk-forward training support.
Temporal Convolutional Network
Causal dilated convolutional network for sequences.
Transformer
Causal Transformer encoder with positional encoding.
Ensemble
Direction + magnitude stacking with an out-of-fold meta-model.
Loss functions
Differentiable Sharpe/Sortino/Calmar/Omega/directional losses.
Training utilities
Sample weighting and early stopping.
Rolling models
Walk-forward evaluation wrappers for time-series models.