Inspiration
We were inspired by National Bank's challenge to develop an algorithm that beats the market, given some advantages and constraints. All of us were new to finance and quant trading, so we also wanted to use this hackathon challenge as a learning opportunity.
What it does
QuackFlow is an adaptive, regime-aware trading algorithm that dynamically changes its behaviour based on real-time market conditions. It manages inventory, risk, and tries to buy/sell at the right time based on different strategies.
How we built it
We used the given Python template from National Bank that connects to a real-time market data simulation. We also had some help from online resources like ChatGPT and Gemini for guidance and financial knowledge.
Challenges we ran into
The most challenging part was to figure out how to implement the best money-earning algorithm that respected the constraints and didn't go bankrupt. Trying to infer the scenario of the market at any given time to choose which algorithm to trade with was also a challenge.
Accomplishments that we're proud of
According to our testing, our final algorithm never went bankrupt or violated the constraints and made a profit in all scenarios. Our earlier algorithms often lost a ton of money, so we're glad to be able to overcome it and make profitable algorithms.
What we learned
We learned a lot about the market and stocks. We also learned a lot about high-frequency trading and the many different approaches there are to algorithmic trading.
What's next for QuackFlow
Tweak how we detect different market types like flash_crash, mini_flash_crash, and stressed_market based on the values for drawdown, drop speed, and rebound (trial and error, or having a larger dataset for trend making) Make the HFT logic smarter, working on more advanced behaviour during crashes; the goal is not just to survive, but actually make more money from market crashes.
Log in or sign up for Devpost to join the conversation.