c6c550a2be
Wire Databento Schema::Trades into OFI feature extraction so VPIN and Kyle's Lambda use real buy/sell classification instead of tick-rule proxy. Trade data pipeline: - trades_loader.rs: DbnTrade struct, load_trades_sync(), binary-search get_trades_for_bar() for O(log n) time-aligned trade windowing - ofi_calculator.rs: feed_trade() accumulates real buy/sell pressure into VPIN, Kyle's Lambda, and trade imbalance calculators - data_loading.rs: loads trades from --trades-data-dir, feeds per-bar trades to OFI calculator before calculate() - download-trades-job.yaml: K8s job for ES.FUT trades from Databento - job-template.yaml: sync trades data from MinIO + --trades-data-dir arg Offline RL (CQL/IQL): - experience_dataset.rs: bincode save/load for pre-collected datasets - iql.rs: Implicit Q-Learning (Kostrikov 2021) — expectile value network, advantage-weighted action extraction - CLI: --offline, --dataset-path, --collect-dataset flags Cleanup: - Remove FeatureVector51/MarketFeatureVector type aliases → FeatureVector - Fix stale dimension comments across 18 files (54→43/51) - Fix feature_dim default (54→43) 2758 tests pass, 0 compile errors, 0 clippy warnings. 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;