eec5494344
- v_range search bounds [5,30] → [2,10] to match EMA-normalized reward range [-0.45,+0.32] (max discounted return ≈1.2) - Replace val_loss proxy (max_Q - reward)^2 with Sharpe-based validation metric: returns -Sharpe so lower = better, economically meaningful - Add compute_per_action_q_values(): samples 200 states from replay buffer, logs per-action Q averages (S100/S50/Flat/L50/L100) at each epoch — exposes whether agent differentiates exposure actions - Remove dead max_q_values computation (leftover from old proxy) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
ml
10-model ML ensemble for the Foxhunt HFT system, built on Candle v0.9.1.
Models
- DQN (Rainbow) — deep Q-network with prioritized replay, dueling heads, noisy nets
- PPO — proximal policy optimization with GAE, LSTM policies, clip-higher
- TFT — temporal fusion transformer for multi-horizon forecasting
- Mamba2 — state space model for sequence prediction
- Liquid Networks — biologically inspired networks for non-stationary data
- TLOB — transformer-based limit order book analysis
- KAN — Kolmogorov-Arnold networks
- xLSTM — extended LSTM architecture
- TGGN — temporal graph neural network
- Diffusion — diffusion-based generative model
Key Modules
ensemble— model ensemble coordination and confidence aggregationhyperopt— PSO-based hyperparameter optimization with per-model adapterstrainers— unified training loops (DQN, PPO, supervised)inference—InferenceAdaptertrait for predictioncheckpoint— model checkpointing and restorationevaluation— walk-forward evaluation pipeline
Usage
use ml::dqn::DQN;
use ml::ppo::PpoTrainer;