Spider Stocks

AAPL Stock Chart Page (Ex: Estimate Up)
AMZN Stock Chart Page
META Stock Chart Page (Ex: Estimate Down)
GOOG Stock Chart Page

Inspiration

We wanted to combine a neural network to a real word application. Predicting stocks seemed like a logical choice with it being hard to analyze all the different trends a stock might have.

What it does

Gathers stock data from previous days and months to create predictions for future closing prices of stocks.

How we built it

It uses a multiple inputs containing different stock trends that feed into a layer based neural network. The network then produces a sigmoid function that creates a "confidence" percentage that the stock's closing price will be higher than the day before.

Challenges we ran into

Finding free API's for the stock market that produce valuable data for the neural network. Creating a nice frontend that is easy to read for users. Frontend is way more difficult than it seems.

Accomplishments that we're proud of

Creating a trained network that estimates day to day changes of stock price at a 60% success rate. Test data will be determined Monday afternoon.

What we learned

Putting together an entire project from back to front end and how both sides work together simultaneously.

What's next for Spider Stocks

Track the next few weeks of stock data to see if the model is able to maintain a 60% correct estimation rate, and if not, how to better select parameters that help fit that data.