Strategy Backtest

Dev Tools

Quantitative strategy backtesting—implement, run, and tune trading rules on historical data; performance metrics (CAGR, max drawdown, Sharpe, win rate) and simple parameter sweeps. Keywords: backtest, algorithmic trading, Backtrader, moving average, MACD, RSI, walk-forward, risk.

Install

openclaw skills install strategy-backtest

Strategy Backtest — Historical Performance & Optimization

Overview

Supports systematic trading strategy workflows: backtest rules on history, optimize parameters (e.g. grid search), and report results. Typical building blocks include moving-average crosses, MACD, RSI, and custom signals—implemented with libraries such as Backtrader or similar.

Trigger keywords: backtest, trading strategy, quant, algorithmic trading, Sharpe, drawdown, optimize parameters, walk-forward

Prerequisites

pip install pandas numpy backtrader matplotlib

Capabilities

Backtest engine — run strategies on historical OHLCV (or vendor-specific) data.
Performance analytics — annualized return, max drawdown, Sharpe-like ratios, win rate (definitions must match your implementation).
Parameter search — grid or bounded search over strategy parameters with out-of-sample caution (see references/strategy_backtest_guide.md).

Commands

Command	Description	Example
`backtest`	Run a backtest	`python3 scripts/skills/strategy-backtest/scripts/strategy_backtest_tool.py backtest [args]`
`optimize`	Parameter optimization	`python3 scripts/skills/strategy-backtest/scripts/strategy_backtest_tool.py optimize [args]`
`report`	Emit a backtest report	`python3 scripts/skills/strategy-backtest/scripts/strategy_backtest_tool.py report [args]`

Usage (from repository root)

python3 scripts/skills/strategy-backtest/scripts/strategy_backtest_tool.py backtest --strategy ma_cross --symbol SPY --period 3y
python3 scripts/skills/strategy-backtest/scripts/strategy_backtest_tool.py optimize --strategy ma_cross --fast 5-20 --slow 20-60
python3 scripts/skills/strategy-backtest/scripts/strategy_backtest_tool.py report --format markdown

Use symbols and venues appropriate to your data feed (e.g. SPY, QQQ, or local indices)—the examples above are illustrative.

Output format (for the agent’s report)

# Strategy backtest report

**Generated**: YYYY-MM-DD HH:MM

## Key findings
1. [Finding 1]
2. [Finding 2]
3. [Finding 3]

## Metrics snapshot
| Metric | Value | Notes |
|--------|-------|-------|
| CAGR | X% | … |
| Max drawdown | Y% | … |
| Sharpe (if defined) | Z | window & rf assumption |

## Analysis
[Grounded in actual run outputs—no fabricated fills or equity curves.]

## Risks & limitations
- Past performance ≠ future results; costs, slippage, and survivorship bias matter.

References

Core

Community & critique

Notes

Base all numbers on script or notebook output—never invent trades or metrics.
Mark missing fields as unavailable instead of guessing.
Not investment advice; comply with local regulations for research vs advisory.
Prefer stating assumptions (fees, spread, leverage, timezone, corporate actions).