We thought of this idea while trying to invest in the stock market and realized that with our backgrounds in Machine Learning, python, and html, we could create something to help us make some money on the side.
What it does
Our website and machine learning model takes any stock ticker as input and analyzes the patterns and price history of the stock to output a 30 day prediction of the stock price.
How we built it
We used python machine learning methods from scikit and used flask and html5 to create the website. We also used a yahoo finance API to get data for our machine learning model.
Challenges we ran into
We initially wanted to do this project with deep learning which produced much more accurate predictions of the stocks, however, this required too much time and power from our computers which is why we chose to switch to basic machine learning. The style of the website was also foreign to us which caused us to have trouble with implementing basic items such as buttons while making them look good. We also started off with using a TD Ameritrade API instead of yahoo finance, however, this API returned a json file instead of a csv file which is what our machine learning model needs. Eventually, we found a way to convert the json into csv however we decided to use yahoo finance for ease of use and efficiency.
Accomplishments that we're proud of
Throughout our testing of our machine learning model, we have achieved upwards of 97% accuracy for our machine learning models and have reached up to 99.7% accuracy. We also successfully incorporated the stock input from the yahoo finance API to our machine learning model. We also successfully built an html website which looks appealing at first glance and is also functional.
What we learned
We learned a lot about the matplotlib package in python as well as a lot about html and css for web development. We also learned a little about stock trading and certain patterns that would lead for profit and patterns that would lead to loss.
What's next for Finding Fortune
We plan on expanding our website to include top movers on a given day as well as recommendations for users. We could also try to use deep learning again but on a limited scale. We also plan on incorporating news to our analysis of stocks and creating a way to show all of the info in a visual way to our users.