I am a mathematician who wants to learn about quantitative finance, in particular how machine learning can be applied to it.

I assume some machine learning techniques are more applicable than others in this field, so which machine techniques should I look into?

Are there any important academic papers or books it would be worth reading that focus on the latest developments in machine learning applied to finance?

  • 1
    $\begingroup$ I'm voting to close this question because the role of this community is not to teach people how to use Google. Although if you need help with that try " financial time series machine learning survey " and be amazed! $\endgroup$ – Quantuple May 20 '16 at 14:36
  • 3
    $\begingroup$ @Quantuple: Disagree. It's difficult to find something valuable from such a google search, even if you have experience in finance and math. For example, I've picked 2 files out of your google search - the first one called "survey" just lists the references without surveying anything, the second explains general machine learning, spending ONE page out of 60 on the financial model. To sonicboom: The following is a well written paper by guys who worked in the industry, but I would like to see more examples myself: cis.upenn.edu/~mkearns/papers/KearnsNevmyvakaHFTRiskBooks.pdf $\endgroup$ – LazyCat May 20 '16 at 15:02
  • $\begingroup$ @LazyCat With all due respect, I beg to differ (1) I think the survey in question gives a good overview of what methods people have investigated (at least up to 2010), with what inputs and in what proportion. For implementation details you have plenty of references. (2) Seriously? Who picks 2 links at random from a Google search and trust them as relevant? Plus second link points to a similar SE question... (3) Your paper deals with ML predictions over very short time scales, maybe it is not at all what the OP is interested in? The "survey" gives references to studies at different time scales. $\endgroup$ – Quantuple May 20 '16 at 15:38
  • $\begingroup$ Also, if a revolutionary and valuable method had been made public, then presumably, it would not be very profitable any more, would it? IMHO, surveys and the likes should only be used to get to know what was already done and help you do better think out of the box. $\endgroup$ – Quantuple May 20 '16 at 15:41
  • 3
    $\begingroup$ @Quantuple: it's easy to find a lot of references on ML in finance. The problem is that most of it is pretty useless. E.g. some of it written by people in academics with the sole purpose of advancing their career. So it would be really helpful, if someone with experience in applying these methods in practice would point out good stuff. Although with mixed success, I did try to use ML for intraday trading, I found the paper I refered to useful and well-written. If you have experience in the subject, I would be interested to know what papers/research on the topic you've read and found helpful. $\endgroup$ – LazyCat May 20 '16 at 16:00

I think the bible of machine learning in finance has become: Advances in Financial Machine Learning by Marcos Lopez de Prado 2018.


If you are interested in checking performance of a trading strategy using machine learning techniques I recommend using Quantopian for back-testing a Random Forest Classifier against SPY benchmark: https://www.quantopian.com/posts/simple-machine-learning-example-mk-ii

Machine learning can be useful in selecting alpha factors predictive of return as described in the following Quantopian notebook: https://www.quantopian.com/posts/machine-learning-on-quantopian

Machine learning can also be used to find clusters of stocks (K-means, GMM) for pair trading strategies, discover principal components (PCA) for mean reversion strategies, and predict latent market states (HMM, Kalman Filter).

I recommend Kevin Murphy's "Machine Learning: A probabilistic perspective" as an excellent resource for studying machine learning.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.