srfm::stream::CoordinateNormalizer Class Reference

Rolling z-score normaliser using Welford's online algorithm. More...

#include <coordinate_normalizer.hpp>

Public Member Functions
	CoordinateNormalizer (std::size_t window)
	Construct with a given rolling window size.
void	update (double close) noexcept
	Ingest one new close price into the rolling window.
double	normalise (double close) const noexcept
	Return the z-score of value relative to the current window.
double	mean () const noexcept
	Rolling mean of the window. 0.0 if window not yet full.
double	sigma () const noexcept
	Rolling sample standard deviation. 0.0 if window not yet full.
std::size_t	count () const noexcept
	Number of ticks seen so far (saturates at window size).
bool	warmed_up () const noexcept
	Whether the window has been filled at least once.
std::size_t	window () const noexcept
	Configured window size.
void	reset () noexcept
	Reset all state as if no ticks have been seen.

Detailed Description

Rolling z-score normaliser using Welford's online algorithm.

Window size is fixed at construction time. Call update() with each new close price, then normalise() to obtain the current z-score.

CoordinateNormalizer norm{20};
for (const auto& tick : ticks) {
    norm.update(tick.close);
    double z = norm.normalise(tick.close);
    // z is 0.0 until the window is full
}

Definition at line 61 of file coordinate_normalizer.hpp.

Constructor & Destructor Documentation

CoordinateNormalizer()

srfm::stream::CoordinateNormalizer::CoordinateNormalizer ( std::size_t  window )

inlineexplicit

Construct with a given rolling window size.

Parameters

window

Number of ticks in the rolling window. Must be ≥ 2.

Definition at line 68 of file coordinate_normalizer.hpp.

Member Function Documentation

count()

std::size_t srfm::stream::CoordinateNormalizer::count ( ) const

inlinenoexcept

Number of ticks seen so far (saturates at window size).

Definition at line 141 of file coordinate_normalizer.hpp.

mean()

double srfm::stream::CoordinateNormalizer::mean ( ) const

inlinenoexcept

Rolling mean of the window. 0.0 if window not yet full.

Definition at line 131 of file coordinate_normalizer.hpp.

normalise()

double srfm::stream::CoordinateNormalizer::normalise ( double  close ) const

inlinenoexcept

Return the z-score of value relative to the current window.

Returns 0.0 if the window is not yet full or if the standard deviation is effectively zero (constant price series).

Parameters

value

Value to normalise (typically the current close price).

Note

noexcept — arithmetic only.

Definition at line 120 of file coordinate_normalizer.hpp.

reset()

void srfm::stream::CoordinateNormalizer::reset ( )

inlinenoexcept

Reset all state as if no ticks have been seen.

The window size is preserved.

Definition at line 156 of file coordinate_normalizer.hpp.

sigma()

double srfm::stream::CoordinateNormalizer::sigma ( ) const

inlinenoexcept

Rolling sample standard deviation. 0.0 if window not yet full.

Definition at line 134 of file coordinate_normalizer.hpp.

update()

void srfm::stream::CoordinateNormalizer::update ( double  close )

inlinenoexcept

Ingest one new close price into the rolling window.

Updates the internal circular buffer and the running mean/M2 accumulators using Welford's online algorithm. O(1) per call.

Parameters

close

Raw close price. Must be finite and positive.

Note

noexcept — arithmetic only, no allocation.

Definition at line 89 of file coordinate_normalizer.hpp.

warmed_up()

bool srfm::stream::CoordinateNormalizer::warmed_up ( ) const

inlinenoexcept

Whether the window has been filled at least once.

Definition at line 144 of file coordinate_normalizer.hpp.

window()

std::size_t srfm::stream::CoordinateNormalizer::window ( ) const

inlinenoexcept

Configured window size.

Definition at line 147 of file coordinate_normalizer.hpp.

---

The documentation for this class was generated from the following file:

• include/srfm/stream/coordinate_normalizer.hpp