Skip to content

Modify scripts readme to add reclamm training section#82

Open
jubeira wants to merge 4 commits into
QuantAMMProtocol:training-pipelinefrom
balancer:training-pipeline
Open

Modify scripts readme to add reclamm training section#82
jubeira wants to merge 4 commits into
QuantAMMProtocol:training-pipelinefrom
balancer:training-pipeline

Conversation

@jubeira
Copy link
Copy Markdown

@jubeira jubeira commented Apr 29, 2026

Quickstart guide to train reclamm pools (docs only, no code).

Juan Ignacio Ubeira and others added 4 commits April 29, 2026 16:15
…s timestamps

- scripts/download_data.py: force mpire WorkerPool to start_method='spawn'
  so binance_historical_data workers don't fork after JAX threads (macOS).
- get_binance_vision_data: vectorize timestamp normalization with numpy;
  handles Binance Vision's 2025+ microsecond klines and runs ~20× faster.
- forward_fill_ohlcv_data: cast date_range via datetime64[ms] before int64
  — pandas 3.x switched pd.date_range default unit from [ns] to [ms].
- resample frequency "1H" → "1h" (pandas 3.x dropped uppercase alias).
- NOTION_GUIDE.md: setup section now reflects venv path and training-pipeline
  branch checkout.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Agent-facing guide covering env/venv usage, data download caveats,
do_run_on_historic_data + HODL baselines, Optuna tuning, and the
continuous-vs-fresh-deposit pitfall. Drops NOTION_GUIDE.md (content
now lives in the team Notion page).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Ignore .venv/, .claude/, results/, and optuna_studies/ created by the
simulation/tuning workflow.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants