What problem does it solve? Evaluating a trading strategy requires more than a single Sharpe ratio: you need risk metrics, parameter optimization, out-of-sample validation, and forward-looking risk projections. This Skill implements the full 5-stage backtesting methodology (Data → Research → Metrics → Parameterisation → Validation) in flat numpy/pandas Python, so quantitative researchers can rigorously test strategies without heavy dependencies. ## Core Features & Use Cases - 30+ Risk and Performance Ratios: Sharpe, Sortino, max drawdown, VaR/cVaR (empirical, normal, Johnson SU), Kelly fraction, Rachev A/B/C, profit factor, risk of ruin, all as flat vectorized numpy functions. - Event-Driven Engine with 8 Strategies: SMA crossover, RSI mean-reversion, Bollinger contrarian, MACD, ADX trend, momentum, and more, with commission and slippage modeling and proper signal shifting to avoid look-ahead bias. - Validation and Simulation: Walk-forward cross-validation with IS/OOS splits, parameter sweeps, Monte Carlo search, Markowitz portfolio optimization, and forward simulation using Johnson SU marginals with t/Gaussian/Clayton/Gumbel/Frank copulas. - Use Case: Load a CSV of SPY prices, run a walk-forward validation of an SMA(50/200) crossover strategy, compare in-sample vs out-of-sample Sharpe ratios, then project 10,000 forward wealth paths to estimate the probability of a 30% drawdown. ## Quick Start Run the backtesting CLI on the bundled S&P 500 returns CSV to compute all 30+ risk and performance ratios, then execute the 4-level validation suite to verify mathematical correctness.
