Module cointegration 

Engle-Granger cointegration test and pair-trading statistics.

Pure math — no I/O, no async, no external dependencies beyond std + rust_decimal.

Algorithm

1. OLS regression: y = alpha + beta * x + residuals
2. ADF test on residuals: tests if the spread is mean-reverting
3. Half-life via Ornstein-Uhlenbeck: hl = -ln(2) / ln(1 + theta)
4. Pearson correlation between the two price series

Structs

EngleGrangerResult

Result of an Engle-Granger cointegration test.

Enums

CointegrationMathError

Errors from cointegration computation.

Constants

MIN_OBSERVATIONS

Minimum number of observations for a meaningful cointegration test.

Functions

engle_granger

Run an Engle-Granger cointegration test on two price series.

pearson_correlation

Pearson correlation coefficient between two series.

spread_stats_to_decimal

Convert spread mean and std from f64 to Decimal for the port layer.