ml4t-tearsheet
Installation
SKILL.md
Strategy Tearsheet
A single metric hides more than it reveals. A tearsheet shows cumulative returns, drawdowns, rolling Sharpe, monthly heatmap, and a metrics table - exposing regime dependence, tail risk, and decay that one number cannot.
The Problem
Reporting only Sharpe ratio misses critical failure modes. A Sharpe of 1.5 could come from steady 10 bps/day or from one massive gain followed by slow bleed. Max drawdown reveals survival risk. Rolling Sharpe reveals whether alpha is decaying. Monthly returns reveal seasonality. You need all of them.
The Pattern
WRONG
# Single number - hides everything important
sharpe = returns.mean() / returns.std() * np.sqrt(252)
print(f"Sharpe: {sharpe:.2f}") # "Looks great!" - but is the strategy dying?