Hacker News Reader

>- Stock Market Prediction

I can assure you this is being done. Anything that works is kept secret.

Ditto, understandably, it's kept secret but would be nice to at least see some of the models, presented outside the trading context. That being said, hft world still primarily uses linear/logistic regression.

We see a lot of finance work(trend prediction in markets being 1 of these things. For weather forceasting, moji (HUGE weather in app in china) does a lot of deep learning based on data collected from their mobile app.

Yandex.Weather relies heavily on deep convnets for short term forecasting.

http://healthcare.ai/ is an early-alpha tool for healthcare data using Python and R.

In chaotic systems (like weather or stock market) there is a fair share of randomness, which cannot be predicted no matter what approach we use.

Not to mention huge externalities. In the case of stock markets, things like financial crises, central bank decisions, political events (brexit etc).

Some of those externalities could be predicted by deep learning (the crash of the housing market for example)

I think these could have easily be predicted using common sense ;)