1
$\begingroup$

I am returning to studying the markets after ten years spent in banking. I would like to ask for directions, what approaches to price prediction are currently used - I want to catch up quickly. Papers are often behind a paywall, so I would like to invest time/money in areas that are 'hot'.

One of the last papers I recall studying was on reproducing kernel hilbert spaces (Krejnik, Tyutin, 2010) - its this approach still used? If yes, has there been some notable developments?

Improve this question
$\endgroup$
7
  • 2
    $\begingroup$ If anything is hot right now and works - it wouldn't be available freely on the internet. Also it is return prediction that most people do, not price (technically they are the same but one is stationary and the other is not). All I mean is that you should be more specific than price prediction - everyone is trying (usually unsucessfully) to predict whether stocks will go up or down, and there are many, many, many approaches that have been tried. $\endgroup$
    – rubikscube09
    Jun 18, 2021 at 14:07
  • $\begingroup$ Thanks for pointing out my error. To rephrase - what are the currently accepted gold standards for returns modeling? Back in my days, arima-garch was taught everywhere and the fit was poor. Then rkhs came and it seemed better. Ten years have passed - has there been some development in returns modeling using these methods? Or is there some other commonly accepted modeling approach? $\endgroup$
    – Daniel Bencik
    Jun 18, 2021 at 15:17
  • 1
    $\begingroup$ There is a one-to-one mapping between return prediction and price prediction; if you have one, you have the other. Whether you call it return prediction or price prediction is a matter of taste. $\endgroup$
    – Richard Hardy
    Jun 18, 2021 at 16:18
  • 1
    $\begingroup$ @RichardHardy yes - I just wanted to make sure that no one is training models using prices. $\endgroup$
    – rubikscube09
    Jun 18, 2021 at 16:52
  • 1
    $\begingroup$ Daniel: If you want a relatively short, "get back to speed" approach, it's probably best to read Ernie Chan's books. Marcos Lopez De Prado also has a book on machine learning that seems popular but I haven't read it yet. I think getting up to speed would be faster if you read practitoner type books rather than papers but that's just my take. And, as rubikscube09 mentioned, If anything does really work, it's not going to be in the literature, books or papers, anyway.. But those books could provide what's generally popular. $\endgroup$
    – mark leeds
    Jun 18, 2021 at 21:57

0

Reset to default

Your Answer

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