Crypto Price Prediction

I got into crypto during college like a lot of people, but I was more interested in the technical challenge of prediction than actually trading. Crypto markets are notoriously volatile and unpredictable, which makes them a perfect testing ground for machine learning models. For this project, I built prediction models using both LSTMs and Transformers, feeding them technical indicators like momentum and volatility metrics. The Transformer architecture was particularly interesting because of how it handles sequential data. We published a paper on our findings. While I'm not claiming we solved market prediction (nobody has), we did find some patterns that traditional analysis misses. It taught me a lot about time series analysis and the limits of prediction.