System and method for displaying and analyzing financial correlation data

摘要

A method for displaying a matrix of correlations or other statistical measures of co-movement associated with a plurality of financial instruments, portfolios, indices, or asset classes is disclosed. The method includes: converting the matrix of correlations or other co-movement measures into a probability transition matrix; defining a corresponding abstract distance measurement between any two of the plurality of financial instruments, portfolios, indices, or asset classes based on the probability transition matrix; assigning coordinates in a Euclidean space to each of the plurality of financial instruments, portfolios, indices, or asset classes, wherein a Euclidean distance between any two financial instruments, portfolios, indices, or asset classes in the Euclidean space corresponds to the corresponding abstract distance measurement; and displaying on a display device the plurality of financial instruments, portfolios, indices, or asset classes based on more significant dimensions of the Euclidean space.

主权项

1. A computer-implemented method for displaying on a display device an N×N matrix of correlations of N financial instruments, portfolios, indices, or asset classes using M dimensions, N being a large number, M being significantly smaller than N, while closely approximating distances between the N financial instruments, portfolios, indices, or asset classes as defined by the N×N matrix of correlations, the method comprising:
converting by a computer processor the N×N matrix of correlations into a probability transition matrix, each of the correlations representing a numerical correlation between two of the N financial instruments, portfolios, indices, or asset classes, the probability transition matrix comprising corresponding normalized versions of the correlations in the N×N matrix of correlations; defining by the computer processor a corresponding first distance measurement between any two of the N financial instruments, portfolios, indices, or asset classes based on the probability transition matrix, the first distance measurement measuring a closeness between corresponding said normalized correlations of said any two of the N financial instruments, portfolios, indices, or asset classes; using by the computer processor a diffusion map to assign coordinates in an M-dimensional Euclidean space to each of the N financial instruments, portfolios, indices, or asset classes corresponding to non-unit eigenvalues of the probability transition matrix such that a Euclidean distance between said any two of the N financial instruments, portfolios, indices, or asset classes in the Euclidean space closely approximates the corresponding first distance measurement; and displaying on the display device the N financial instruments, portfolios, indices, or asset classes based on the Euclidean space corresponding to a selection of M ones of the eigenvalues.