prInvestor: Pattern Recognition based Financial Time Series Investment System

Predictability of financial time series (FTS) is a well-known dilemma. A typical approach to this problem is to apply a regression model, built on the historical data and then further extend it into the future. If however the goal is to support or even make investment decisions, regression-based FTS predictions are inappropriate as on top of being uncertain and unnecessarily complex, they require further analysis to make an investment decision. Rather than precise FTS prediction, a busy investor may prefer a simple decision on the current day transaction: buy, wait, sell, that would maximise his return on investment. Based on such assumptions a classification model is proposed that learns the transaction patterns from optimally labelled historical data and accordingly gives the profit-driven decision for the current day transaction. Exploiting a stochastic nature of an investment cycle the model is locally reduced to a 2-class classification problem and is built on many features extracted from the share price and transaction volume time series. Simulation of the model over 20 years of NYSE:CSC share price history showed substantial improvement of the profit compared to a passive long-term investment.