US 11,836,794 B2
Crisis prediction based on persistence homology of data
Yuri Katz, Island Park, NY (US); and Alain Biem, New York, NY (US)
Assigned to S&P Global Inc., New York, NY (US)
Filed by S&P Global Inc., New York, NY (US)
Filed on Jul. 24, 2020, as Appl. No. 16/937,943.
Prior Publication US 2022/0027989 A1, Jan. 27, 2022
Int. Cl. G06Q 40/04 (2012.01); G06F 17/15 (2006.01)
CPC G06Q 40/04 (2013.01) [G06F 17/153 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
using a number of processors to perform the steps of:
collecting a number of financial data points;
normalizing the number of financial data points into a time series;
merging the time series into a number of aggregates according to a number of sliding windows, wherein the number of sliding windows comprise a number of different time periods and a sliding step increment;
computing a periodic change in an increasing and convex transformation for each aggregate;
applying multi-dimensional time-delayed coordinate embedding to each aggregate;
applying the number of sliding windows to the time-delayed coordinate embedded aggregates;
deriving a number of time series of variances and a number of point clouds within each sliding window;
computing a number of persistence homologies and time series norms for the number of point clouds;
correlating the time series norms with the time series of variances; and
outputting a warning of an impending financial crisis if the correlation of the time series norms with the time series of variances exceeds a predefined threshold.