Performance Notes

Benchmarks are intentionally lightweight and reproducible:

uv run python examples/benchmark_efficiency.py

Current local baseline after the executable-weight expansion optimization:

elapsed_seconds=1.186
ic_rows=299
lag_analysis_rows=20
lag_returns_rows=5710

The previous local baseline was about 2.38s for the same synthetic factor evaluation. The improvement comes from reusing prepared market data and using a vectorized as-of expansion for portfolio weights instead of nested Python loops over execution dates and assets.