← ClaudeAtlas

pynecore-strategylisted

Write, backtest, optimize and deploy trading strategies and indicators as PyneCore (Pyne) Python scripts, and run them through the `pyne` CLI or the `ScriptRunner` API. Use this skill whenever a task involves PyneCore, Pyne code, a `@pyne` docstring, `pynecore.lib`, `ta.*` / `strategy.*` calls in Python, `pyne run`, `.ohlcv` data files, caching exchange candles via CCXT (Binance, Bybit), paper trading or live broker runs of a Pyne strategy, a PyneCore workdir (`scripts/`, `data/`, `output/`), backtests or parameter sweeps, `request.security` multi-timeframe logic, or FreqTrade integration, even when the user just says "write me a strategy", "backtest this idea", "why does my strategy not trade" or "add a stop loss" in a project with PyneCore files. Also covers drawing objects (`line`, `label`, `box`, `table`, `polyline`, `linefill`) with the `--viz` NDJSON format and the `strategy.risk.*` rules. Includes Monte Carlo robustness tests (trade resampling and price perturbation) not built into PyneCore.
theonetrade/pynecore-skills · ★ 2 · Data & Documents · score 78
Install: claude install-skill theonetrade/pynecore-skills
# PyneCore strategy development PyneCore runs **Pyne code**: Python files whose docstring starts with `@pyne`, executed bar by bar with series history, persistent state and a broker emulator. This skill is self-contained and installed globally: it needs no per-project configuration files. Everything project-specific (interpreter, CLI, workdir, data) is discovered at the start of a task. ## 1. Discover the environment first Run the bundled probe (standard library only) from the project root and read its output: ```bash python3 <skill>/scripts/pyne_env.py # add --json for machine-readable output ``` It reports the workdir (the folder with the `.pyne` marker or the `scripts/ data/ config/` layout, or a `workdir/` child) and which of its folders exist, the scripts with their role and `.toml`, the data files with symbol and timeframe, the configured plugins and symbol map, and the interpreter that can import `pynecore` (the project's `.venv`/`venv`/`env` first, then `$VIRTUAL_ENV`, then PATH) with its `pyne` command. Only the project folder is inspected. Rules that follow from it: - Call the CLI by its resolved path (`<venv>/bin/pyne`, or `uv run pyne` in a uv project) and run Python tooling with the same interpreter. The agent shell does not have the venv active. - Always set `PYNE_NO_LOGO=1` (and `NO_COLOR=1` when parsing output) on CLI calls. - No project yet, or no PyneCore installed: `python3 <skill>/scripts/init_project.py <dir> --smoke` builds the whol