EarningsOracle

Inspiration

• We were inspired to make this project because of our lack of experience in the AI field, as well as the stock market. Covering completely uncharted territory was going to be a good way for our team to get our of our comfort zone.

What it does

EarningsOracle attempts to predict the earnings of a stock based on 3 types of data: public sentiment, technical data, as well as news during the time of earnings. We used OpenAI's API to then take this data and find patterns in price movement and all of these analysis.

How we built it

Made fully with React

Challenges we ran into

Learning to work in a React environment as opposed to a browser environment brought a lot of unexpected challenges to our team. We also had to test run many different APIs and AI models, eventually settling on FinnHub, StockNews, and OpenAI.

Accomplishments that we're proud of

We were able to retrieve data from online and get an AI to predict the price action. Though we set high goals for our first Hackathon, we feel like we've created something we are proud of.

What we learned

We learned that before we completely settle on one environment, we should test to see if APIs and AIs that we want to use are compatibile with one another.

What's next for EarningsOracle

We hope to move away from OpenAI, as right now we need to basically jailbreak it to bypass restrictions. In the future we hope to use a different model, and eventually create our own model that we can fine tune and backtest more accurately.

Built With

• react