Machine learning based forecasting of significant daily returns in foreign exchange markets

Asset value forecasting has always attracted an enormous amount of interest among researchers in quantitative analysis. The advent modern machine learning models introduced new tools to tackle this classical problem. In paper, we apply algorithms hitherto unexplored question instances significant fluctuations currency exchange rates. We perform analysis nine using data on four major pairs over a 10 year period. A key contribution is the novel use outlier detection methods for purpose. Numerical experiments show that substantially outperform traditional and finance techniques. addition, recently proposed method PKDE produces best overall results. Our findings hold across different pairs, significance levels, time horizons indicating robustness method.