My project is about to predict the price movement (up or down) of the DJIA and NDX100 with SVM and Random Forest. At present I used the Brent oil futures, Gold futures, DAX, Nikkei 225, S&P 500, Shangai Composite, USD/EUR, USD/GBP, USD/JPY as my features. However, not all features are related with the NDX and DJIA. What I ask for is to use only the related features from my features set. Feature importance scaling or recursive feature elimination may be used to eliminate the problem. Only the related features will be used and results will be provided with accuracy scores, ROC curve and F1 scores.