1
$\begingroup$

I have searched many websites and forums describing stock price forecast using LSTM.

They shared two things in common: one is that all the sources make predictions with same set of data and none of these websites show a prediction for next 10 days.

So I would like to know whether it's possible to predict stock prices for the next few days (n days) using LSTM. And why do most sources use the same dataset?

$\endgroup$
1
$\begingroup$

The reason why you find the same dataset everywhere is that financial data is extremely secretive and expensive and nobody is willing to share their financial information because in doing so they will be effectively losing an edge in the market. For that reason public financial data is mostly gathered by academics and not too often.

Second, LSTM can make predictions for the next n-hours, n-days, n-months and even n-years like the article STOCK PRICE PREDICTION USING LSTM,RNN AND CNN-SLIDING WINDOW MODEL by Sreelekshmy Selvin et. al. and many others show. You just need to build a good training vector and the target can be anything you want. As it's natural in financial markets, accuracy decreases as time increases.

$\endgroup$
1
  • $\begingroup$ And more often that not, the correlation between the predicted delta and the actual delta is 0. $\endgroup$ – Matthieu Brucher Dec 20 '18 at 16:10

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.