predictor_cor: Analyzes correlations between an asset and a predictor...

In jeanmarcgp/xtsanalytics: A library of functions to do advanced analytics of time series data (xts matrices).

Description

Computes the class correlations between an asset (or any equity curve) and a predictor. The predictor can be a binary class or a real value centered about 0.

Usage

predictor_cor(data, price_col = "prices", timer_col = "GC",
  plot = TRUE)

Arguments

data	The xts matrix data containing the equity curve of the asset or portfolio, and the classifier's prediction column. e.g. market timer. The predictor column may be binary (-1, +1) or real-valued centered about zero.
price_col	The column name or column number of the equity curve under test. Default = "prices".
timer_col	The column name or column number of the prediction (or timer). Default is "GC".
plot	Logical value indicating whether or not to output a scatter plot annotated with the class correlations and a legend.

Details

In the real valued case, it is assumed that the sign represents the direction of the market, while the real value is a conviction level. For example, a predictor value of 0.0001 could be considered a weak conviction of an up market, whereas a predictor value of -0.1 could be construed as a strong conviction of a down market. Person linear correlations (function cor) is used to compute the correlation between the actual future market direction (internally represented as +1, -1), and the real value of the indicator

Value

Returns the class correlations and other information as a list.

$pred_N_up:

Number of up predictions by the predictor.

$actual_N_up:

Number of actual up observations of the asset.

$up_cor:

The predictor correlation for the up class (predictor value is positive).

$pred_N_down:

Number of down predictions by the predictor.

$actual_N_down:

Number of actual down observations of the asset.

$down_cor:

The predictor correlation for the up class (predictor value is negative).

jeanmarcgp/xtsanalytics documentation built on May 19, 2019, 12:38 a.m.