^N100 Index model

• 

  Percent error of predictions (7/21-7/28)

Inspiration

I used to wonder how stocks work and if there's really a good way to predict whether or not a stock will fall or not. With machine learning, I quickly found out there was.

What it does

Predicts stock prices, specifically from ^N100

How we built it

Used PyTorch in a jupyter notebook with an LSTM cell RNN system

Challenges we ran into

Very new to PyTorch, so most of the time was spent trying to find out how things work

Accomplishments that we're proud of

The model's results are actually not bad, financial predictions are within a 10% percent error

What we learned

RNNs are definitely difficult to preprocess and get them working

What's next for ^N100 Index model

Probably work on trying to get more stocks into the model so that it is a multilabel model

Built With

• python
• pytorch