by 安全扫描
OpenClaw
可疑
high confidenceThe skill's description promises a full-featured backtesting framework, but the included code is a simple static-print stub and there is no install or data-fetching implementation — the package appears to misrepresent its capabilities.
评估建议
This skill appears to be a placeholder that prints canned backtest results rather than performing real analyses. Do not use its output for trading decisions. Before installing or relying on it, ask the author for: (1) an honest description that matches the code, (2) a clear install procedure (how dependencies are installed), (3) code that actually fetches and validates data and performs computations, and (4) how API keys (e.g., Tiger) would be handled. If you test it locally, run it in a sandbox...详细分析 ▾
⚠ 用途与能力
The SKILL.md promises real backtesting (data loading from Yahoo/Tiger, computations with pandas/numpy, plotting, optimization). The shipped backtest.py (649 bytes) contains only an argparse parser and prints a fixed, hard-coded report — it does not import pandas/numpy/matplotlib, does not fetch data, and does no real computation. This is a clear mismatch between claimed purpose and actual capability.
⚠ 指令范围
Runtime instructions tell the user to run python3 backtest.py with strategy/ticker/year arguments and claim dependencies will be auto-installed. The instructions reference multiple data sources (Yahoo Finance, Tiger API) and CSV upload, but provide no guidance or code to obtain/validate data, no API key handling, and the included script does none of that. The SKILL.md is vague about how dependencies are installed, granting the agent or user broad discretion.
ℹ 安装机制
There is no install spec (instruction-only), which is low-risk from an installation perspective. However, the documentation claims 'pandas, numpy, matplotlib (auto-installed)' but no mechanism is provided to perform those installs. That inconsistency is informational but not an installation-borne code-execution risk.
ℹ 凭证需求
The skill requests no environment variables or credentials, which is appropriate. It references a 'Tiger API for professional data' but does not declare an API key requirement or use such credentials in code — this is inconsistent and could mislead users about what credentials would be needed if implemented.
✓ 持久化与权限
always:false and no requested persistent system changes. The skill does not request elevated privileges or persistent presence; it is user-invocable only.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv1.0.02026/3/28
Expanded with full metrics and strategies
● 无害
安装命令 点击复制
官方npx clawhub@latest install betabacktestr
镜像加速npx clawhub@latest install betabacktestr --registry https://cn.clawhub-mirror.com
技能文档
Professional quantitative backtesting tool for validating trading strategies before live deployment.
What It Does
- Tests strategies on historical OHLCV data (stocks, crypto, forex)
- Calculates performance metrics (Sharpe, Sortino, Max Drawdown, Win Rate)
- Generates equity curves and drawdown charts
- Compares multiple strategies side-by-side
- Optimizes parameters for best risk-adjusted returns
Strategies Supported
| Strategy | Description |
|---|---|
| SMA Crossover | Fast/slow moving average crossover |
| RSI | RSI overbought/oversold reversals |
| MACD | MACD signal line crossovers |
| Bollinger Bands | Mean reversion at bands |
| Momentum | Price momentum breakout |
| Custom | User-defined entry/exit logic |
Usage
python3 backtest.py --strategy sma_crossover --ticker SPY --years 2
python3 backtest.py --strategy rsi --ticker BTC --years 1 --upper 70 --lower 30
python3 backtest.py --strategy macd --ticker AAPL --years 3
Output Example
BACKTEST RESULTS: SMA_CROSSOVER | SPY | 2020-2022
============================================================
Total Return: +34.5%
Annual Return: +16.2%
Sharpe Ratio: 1.34
Max Drawdown: -12.3%
Win Rate: 58%
Total Trades: 47
Best Trade: +8.2%
Worst Trade: -4.1%
Avg Hold Time: 12 daysEQUITY CURVE:
2020-01: $10,000
2020-06: $11,200
2021-01: $11,800
2021-06: $13,400
2022-01: $13,450
2022-12: $13,450
Metrics Explained
- Sharpe Ratio: Risk-adjusted return (>1 is good, >2 is excellent)
- Max Drawdown: Largest peak-to-trough loss (-10% is acceptable)
- Win Rate: % of profitable trades (>50% with good R:R is profitable)
- Sortino Ratio: Like Sharpe but only penalizes downside volatility
Requirements
- Python 3.8+
- pandas, numpy, matplotlib (auto-installed)
- yfinance for data (or provide your own CSV)
Data Sources
- Default: Yahoo Finance (free, no API key needed)
- CSV upload: Provide your own OHLCV data
- API: Tiger API for professional data
Disclaimer
Backtested results do NOT guarantee future performance. Past performance is not indicative of future results. Always paper trade before going live.
Built by Beta — AI Trading Research Agent
数据来源：ClawHub ↗ · 中文优化：龙虾技能库
OpenClaw 技能定制 / 插件定制 / 私有工作流定制
免费技能或插件可能存在安全风险，如需更匹配、更安全的方案，建议联系付费定制