AI Stock Advisor

apple stock prediction next 200 days
Apple prediction stock last years
tesla stock price

Inspiration

Becoming rich is what inspired us to make this app and what better than betting all your savings on one stock that will definitely gain in value according to our AI (Disclaimer: we strongly advise against using this website to make financial decisions, please use at your own risk).

What it does

It takes the history of a specific class and trains our model to predict the value of a stock based on the values of x number of days before that day, thanks to that it can graph a prediction of the next values of the stock in the next y number of days.

How we built it

For the dataset, we used Yahoo to get the history of the stocks. Then we used a sequential model where we had three layers of LSTM and one Dense layer. On the website, you can choose a stock that will be sent to the backend using where the model will be trained and from where the result will be sent to the website and graphed using JavaScript.

Challenges we ran into

So for the model, it is a great predictor of old stock prices because it predicts the value of the stock at a day based on the 30 days which functions well when you have those 30 days, but when you have to predict all those days before which is the case when you want to predict the price in the future it doesn't work well at all, so we had to add a random factor that somewhat solved the problem but a better solution would have been to change models but we didn't have time for that.

Accomplishments that we're proud of

We are proud that we have an app that is running and not some incomplete project.

What we learned

Jason learned how to use Flask for the backend and text file database, and Nizar learned how to use the ML framework Tensorflow to train the model and test it as well as how to make a simple API using Flask.

What's next for AI Stock Advisor

First, we need to find a better model for us to use, a model that can better predict the future, then we also have to work on our website UI which is for now very basic and standard