Inspiration
As financial markets have many layers and variables to account for, most predictions seem random and unable to filter the signal from the noise. The goal of this project was to understand how these variables influenced the value of a stock and how I could find tune my model to predict such value.
What it does
It accounts for many components of available stock information such as volume, bid amount, sell amount , volatility, etc. It uses these values and applies a scoring algorithm that scales on each value based on the model that was created, and generates a final overall score for each stock. After all the stocks are ranked, each second, the program chooses to sell the lowest score and buy the highest score stock from the possible options.
How I built it
Using bloombergs api to connect to their simulated server and python as a made backend framework.
Challenges I ran into
Due to the limiting api request amount of 3 times per round (20 min per round), I had issues with linking this correctly and realized later I was calling the request too many times instead of having it being called a single time.
Accomplishments that I'm proud of
Using my physics background in understanding complex multi variable systems and modeling, I was able to do finical research through the internet and friends to develop a strong approximation of each variables contribution.
What I learned
A lot about the financial market
What's next for Bloomberg Code B - Algorithmic Trading
Likely I will make a new algorithm, implementing machine learning to understand the data on a deeper level to make better cognitive decisions, applied to real data sets such as the NASDAQ.
Log in or sign up for Devpost to join the conversation.