foxhunt

Files

jgrusewski 195f880fb4 feat(hyperopt): add batch_size to PPO and Liquid parameter spaces

PPO: add batch_size as 6th param (index 5)
- Static bounds: 512–8192 (PPO needs large batches for variance reduction)
- Default: 2048 (was hardcoded)
- Dynamic: scales with VRAM (200MB overhead, 0.015 MB/sample)
- Wire into PPOConfig in train_with_params()

Liquid CfC: add batch_size as 8th param (index 7)
- Static bounds: 8–512
- Default: 32 (was hardcoded)
- Dynamic: scales with VRAM (100MB overhead, 0.06 MB/sample)
- Wire into CfCTrainConfig in train_with_params()
- NUM_PARAMS: 7 → 8

All 176 hyperopt tests pass, full workspace compiles clean.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-02-28 19:20:08 +01:00

backtesting

perf: eliminate heap-alloc Decimal→float casts across 19 files (36 instances)

2026-02-28 02:29:04 +01:00

common

feat(common): re-export financial_types as FinancialPrice/Quantity/Money/Ratio

2026-02-28 09:49:00 +01:00

config

refactor: restructure repo — crates/, bin/, testing/ layout

2026-02-25 11:56:00 +01:00

ctrader-openapi

refactor: restructure repo — crates/, bin/, testing/ layout