Implementing an evolving ML/Deep Learning trading algorithm that learns on-the-go with plans for future optimization using FPGAs.
Our goal is to develop a trading model that doesn't just learn from static data but adapts in real-time as new data comes in. This will make our model super responsive to the ever-changing market conditions.
Traditional ML models usually train on a fixed dataset. But markets are anything but fixed—they’re dynamic, unpredictable, and always evolving. Online learning allows our model to continuously update and improve with new data, making it more relevant and accurate in real-time trading scenarios.
We'll start by developing and testing the online learning algorithm using dummy trading data. As we progress, we’ll explore more sophisticated aspects, like maybe compiling our algorithm down to FPGA-friendly languages to optimize our algorithm and also look into retrieving old/delayed market data.