STUDY OF PREDICTION MODELS FOR TIME SERIES

We present the concept which approaches the string theory to the field of time series forecast and data analysis through a transformation of currency rate data to the topology of physical strings and branes. We introduce new type of prediction models for financial time series based on string invariants. The performance of the first versions of prediction models is compared to support vector machines and artificial neural networks on an artificial and financial time series. We propose a string angular momentum as an another tool to analyze the stability of currency rates except the historical volatility. Next we investigate the fundamental properties of the space of time series data. We provide the proof that the space of time series data is a Kolmogorov space with T-0-separation axiom using the loop space of time series data.