Machine Learning Methods for Financial Forecasting: Application to the S&P 500

摘要: The world of quantitative finance is extremely intricate and the level complexity has increased over years. Therefore, Decision Support Systems (DSS) are more appreciated in finance. In recent years, machine learning statistical techniques have been applied to financial problems, people industry realized that these powerful and, usually, competitive with current systems. They therefore ready invest on research pertaining or related Artificial Intelligence new mathematical models options pricing, derivatives, currency trading, hedging, etc.The main purpose this project was study historical values S&P500, so as forecast future values. First, dissertation presents relative depth what may influence movement S&P500. Then, it takes a univariate approach looks at some potential forecasting benchmark well get feel both could be leading indicators S&P500 methods most appropriate for multivariate approach. next step will implement best model set indicators.A different were implemented obtained results very encouraging. final design performs better than outperforms Gately design, reference industry.It concluded implementing perhaps successful investment least short period time, until arbitrage opportunity absorbed, but useful longer time.