HistoricalVolatility

Summary

HistoricalVolatility is RTTA's streaming implementation of: Annualized rolling standard deviation of log returns.

Update API

result = rtta.HistoricalVolatility().update(close)

The update(...) call consumes one observation using close. advance(...) uses the same inputs when the caller wants to update state without materializing a Python return value.

Theory Of Operation

HistoricalVolatility estimates realized volatility from close-to-close log returns over a rolling window and annualizes with \(\sqrt{\texttt{periods\_per\_year}}\) (default 252).

Recurrence

Let \(c_t = close_t\), \(n\) the window, and \(P\) periods per year.

\[r_t = \ln\frac{c_t}{c_{t-1}}\]

\[\sigma_t = \operatorname{stddev}_n(r_t), \qquad HV_t = \sigma_t \sqrt{P}\]

stddev uses the same population form as StdDev (\(1/n\) second moment). The first sample has no return and yields 0 when fillna=True, else NaN.

The return value is the current scalar indicator value.

Composed Primitives

StdDev

Implementation Notes

The recurrence is implemented in src/rtta/indicator.cpp in class HistoricalVolatility.

Reference

Background reference