One of our famous mathematicians, James Simons, used an extension of the Baum-Welch algorithm to 'crack' the wall street when he started trading on the stock market. Now, as Google, all informations his team used to trade are hidden to the public. I would like to do such thing, but it'll be very hard.

I think it should be great to start with using the Baum-Welch Algorithm. Some guys suggest me to look up A Spectral Algorithm for Learning Hidden Markov Models and Learning Parametric Output HMMs. Another recommend me just starting with deep neural network.

To implement a great model, what are you suggesting to use first? In fact, I just don't know if I need to learn deep learning first or starting implement my algorithm. What is it advised to learn first?

Thanks in advance!

  • 1
    $\begingroup$ I do not think this is going to be answerable on this site. Even if someone here knows something about data science used for marked investment and wants to share, then they do not know "where to start" would apply to you. You'll get the same kind of answers as the "some guys" . . . you should also bear in mind that data science is widely known and large financial companies have used it for a long time. The chance of a lone actor succeeding to "crack" Wall Street is very low if all you do is implement already existing models. $\endgroup$ Apr 8 '17 at 23:19
  • $\begingroup$ @NeilSlater You are absolutely right to say that. Know that James Simons started also to implement such model alone. Finally, he has succeeded to work with other guys on this project. As far as I am concerned, I will also start implementing this model alone, and sometimes I will work with other guy. It will take time, but it is a necessary evil to doing what you like. Sorry, my english is a bit poor. $\endgroup$
    – J.Doe
    Apr 8 '17 at 23:28
  • 1
    $\begingroup$ I don't think at this stage it matters what you start to learn. Start with basic machine learning - e.g. Andrew Ng's Coursera course: coursera.org/learn/machine-learning - it only takes a few weeks and then you will be ready to start exploring the harder questions. There is nothing in that course about markets, HMMs or deep learning. But without the basics, those technologies are very hard to understand properly. $\endgroup$ Apr 8 '17 at 23:33