Universal randomised signatures for generative time series modelling
By Francesca Biagini, Lukas Gonon, Niklas Walter
Rating
1661
Battle Count: 124
Relevance
8/10
Highly relevant for generating realistic financial time series data, which can be used for model testing, risk assessment, and strategy development in quantitative trading
Implementation Complexity
7/10
Requires understanding of advanced concepts like randomised signatures and neural SDEs, but code is provided to aid implementation
Reproducibility
4/5
Code is available on GitHub, detailed methodology and experiments are provided
About this paper
Methodology: Randomised Signature-based Generative Model. Problem types: Time Series Forecasting, Generative Modeling.
The interactive Everscope explorer (charts, battles, favorites) loads below.