An algorithm to predict time series from finance, economics, etc

What is it about?

A filter with delay-induced negative group delay is presented. The filter consists of multiple time-delayed feedback terms, which lead to a negative group delay for frequencies in the baseband. It can be used for the real-time prediction of band-limited signals. The filter is universal as it does not rely on a specific model of the signal. Specifically, as long as the signal to be predicted is band-limited with a known cutoff frequency, the filter predicts the signal in real time up to a prediction horizon that depends on the cutoff frequency. How signal prediction arises from the negative group delay of the filter is worked out in detail. Its properties, including stability, are derived analytically and demonstrated by numerical simulations. For chaotic systems, the filter is predictive during phases of high predictability.

Featured Image

Photo by Markus Spiske on Unsplash

Photo by Markus Spiske on Unsplash

Why is it important?

This predictor is universal in the sense that it's coefficients do not need to be adapted to the time series to be predicted.

Perspectives