Sorry, you need to enable JavaScript to visit this website.
Skip directly to content

Join Daniel Bloch to explore price jumps with mahalanobis distance and signatures.

13:00 EDT / 18:00 BST / 19:00 CEST

Event Abstract:

In order to estimate the hidden states of dynamical systems, Jump Models cluster financial observations along time series and impose a cost on jumping from one cluster to another. We extend these models by letting the conditioned observations be sampled from a distribution. Further, we consider two seemingly unrelated problems: 1) Detecting and predicting price jumps by identifying whether a new observation results in a price jump relative to previous observations 2) Anomaly detection of segments of a time series, that is, we treat fixed size segments of a time series as a normal corpus, and search for outliers.

We observe that while these problems are clearly different, the former can be reformulated in terms of the latter and we can therefore associate jump indicators to data driven metrics for anomaly detection. We propose a three-step process: 1) jump detection: we combine the Mahalanobis distance with path signatures to come up with a data-driven jump indicator, 2) training: we use this M-distance jump indicator as an input feature for Jump Models, 3) prediction: we use new incoming data to predict its hidden state.

This approach aims at enhancing the model's accuracy and predictive capabilities by leveraging the strengths of Mahalanobis distance on detecting data points after the jump as outliers from previous distribution. We conducted an extensive analysis on simulated data, examining the structure, benefits and limitations of the approach, and found that we could retrieve the true hidden states with high accuracy without using future information.

About the Speaker:

Daniel Bloch has managed teams of quant researchers in top tier banks, developing and implementing option pricing and risk models. He was also a portfolio manager on multi-strategies systematic trading across continents, using multifractal analysis and machine learning. Currently, Daniel conducts research on mathematical finance and AI, focusing on dynamical models applied to forecasting the stock and option market in order to maximise return and minimise risk.