Stock OverFlow

• 

  james

Inspiration

"Number one rule of Wall Street. Nobody... and I don't care if you're Warren Buffet or if you're Jimmy Buffet. Nobody knows if a stock is gonna go up, down, sideways or in fucking circles. Least of all, stockbrokers, right?" - Mark Hanna

What it does

Takes data mined from all NASDAQ stocks, cleans this data and puts it into a neural network that can be used to predict next day closing prices.

How I built it

Extensive data collection and cleaning, using Google Compute Engine. Then running these training points through Tensor flow in a 3 layer neural network.

Challenges I ran into

High amount of computation and incomplete data sets.

Accomplishments that I'm proud of

Learned to use Google Cloud Computing to take advantage of high core-count processors and the use of multi threading. Also learned Tensorflow in a night.

What I learned

Predicting stock prices is much more difficult than it seams. The data is based on countless peoples' irrational consumer choices, and finding correlations between inputs and results can be difficult.

What's next for Stock Overflow

With more relaxed time constraints, and potentially better data, we could increase the inputs of our neural network and potentially find higher correlation.

Built With

• data-mining
• google-cloud-computing
• machine-learning
• multi-threading
• os
• pandas
• python
• tensorflow