runPercentRank: Percent Rank over a Moving Window

View source: R/percentRank.R

runPercentRankR Documentation

Percent Rank over a Moving Window

Description

This function computes a running/rolling percentage rank.

Usage

runPercentRank(x, n = 260, cumulative = FALSE, exact.multiplier = 0.5)

Arguments

x

Object coercible to xts or matrix.

n

Number of periods to use in the window or, if cumulative=TRUE, the number of observations to use before the first result is returned. Must be between 1 and nrow(x), inclusive.

cumulative

Logical, use from-inception calculation?

exact.multiplier

The weight applied to identical values in the window. Must be between 0 and 1, inclusive. See details.

Details

The computation for a percentage rank can vary depending on the weight given to values in the window that are equal to the value being ranked. This weight can be set using the exact.multiplier argument which defaults to 0.5.

exact.multiplier = 0 scores equal values in the lookback window as always being greater than the value being ranked. exact.multiplier = 1 scores equal values as being below the value being ranked. Any multiplier between 0 and 1 counts that proportion of the equal values as being below the value being ranked.

The value of exact.multiplier has the most impact when the window is relatively small or when the number of discrete values in the window is small. For non-repeating values, changing exact.multiplier = 0 to exact.multiplier = 1 for a window of size N will shift the resulting percentile rankings by 1/N. It is equivalent to changing the question from, "how many values are < the value" to "how many values are <= the value".

Value

A object of percent ranks over a n-period moving window of the same class as x and y or a vector (if try.xts fails).

Note

This computation is different from the one used in Microsoft Excel's PERCENTRANK formula. Excel's computation is rather strange and gives inconsistent results as it uses interpolation to rank values that are not found within the lookback window.

Author(s)

Charlie Friedemann

References

The following site(s) were used to code/document this indicator:
https://en.wikipedia.org/wiki/Percentile_rank


TTR documentation built on May 29, 2024, 3:54 a.m.