@Wellenreiter said in Machine Learning and A.I. predictions for 2018:
t care for real trad
yeah i think you are spot on. the real problem with any analysis of the stock market is there are always things going on that the raw data doesn’t tell you. A rock solid buy can be completely undermined overnight by some unknown coming to light. or maybe just someone badmouthing it on a popular media outlet. (the opposite is true as well - bad buy being made “a winner” for a time anyway).
As much as anything, this gives me an outlet for putting my latest data mining work to good use. I mentioned t-sne, if i can solve the last piece of the puzzle with that (the first half seems to work well) i’ll really have something. outliers (as they always are) have been a problem. neural networks by there nature will, if possible, build a scenario for those that decision trees generally dont. if i can find the right way to handle that i might have something even more worth while. :)