Stock market volatility prediction using possibilistic fuzzy modelling

This paper suggests a recursive possibilistic modelling approach (rPFM) for assets return volatility forecasting with jumps. The model employs memberships and typicalities to cluster data, and affine functions in the fuzzy rule consequents. The possibilistic idea provides model robustness to noisy and outlier data, essential for financial markets volatility modelling, which is affected by news, expectations and investors psychology. Computational experiments include actual intraday data from the main equity market indexes in global markets, namely, S&P 500 and Nasdaq (USA), FTSE (UK), DAX (Germany), IBEX (Spain) and Ibovespa (Brazil). Performance of rPFM is compared with well established recursive fuzzy and neural fuzzy modelling. The results show that rPFM produces parsimonious models with better accuracy than the alternative approaches7418119