I trained my model on the data frame that consisted of all trade data rows which means that the model calculated features using the characteristics from past trades and to clarify it more, lets suppose there is a trade X in the data frame, the code will not use the previous candles to calculate the features, instead it will get data from the rows of previous trades to calculate the features + to train the model The model performed very well and gave me 70% accuracy using this approach ( I was getting 30% before this) I just want to know that is this a good approach like am I missing some important considerations? Continue reading...
