Feb

16

 Specs,

Just wondering if any of you have been doing anything related to Machine Learning for modeling the markets?

Years ago, reading The Education Of A Speculator, I learned that Victor was not fond of Neural Networks and the like.

Anything changed?

Regards,

Newton

Mr. Isomorphisms writes: 

You're quite late to the party if just starting.

qplum.co was at a supposed quant conference claiming he had the greatest returns (at the youngest age) of any Tower partner. I was skeptical. For reference the same conference had a bloomberg guy claiming sharpe >5 to a twitter signal (whoops, trading costs not included. Wtf)

2-3 years ago finance bros were all over NLP looking to beat the market. Now the papers are out from that movement. arXiv has turned itself inside out with DL + markets nonsense. I feel like this (and even DL) jumped the shark several years ago. If you want to learn about NN, read Eugenio Culurciello's post (originally on his github) on neural net architectures. The big advances in image recognition were due to flickr users tagging massive amounts of data.

S Mallat and R Vershynin have done some good work in the last 5 years while everyone else is going crazy. Even N Matloff is apparently caught up in the eddy, making points about NN that L Wasserman published 20 years ago. Francois Chollet is moving on. You should too, unless you have a good reason not to. As someone else mentioned, tensorflow, caffe, model zoo, are good fun to play with, if that's your only goal. otoro.net made some fun video games with convnet.js. Think a little more creatively, like that, if you want to play this game.


Comments

WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '12342' AND comment_approved = '1' ORDER BY comment_date

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search