Crypto Trading Assistant

Crypto has attracted younger retail investors and newer trading institutions that are more likely to embrace systematic trading strategies that leverage pattern analysis. Trading is historically a risky endeavour with many investors relying on technical analysis and pattern recognition to predict price the movement of the volatile $2 Trillion market. Pattern detection is currently done manually and is a labor intensive task for trading houses as well as retail investors.

Our goal was to implement a model pipeline that uses deep learning model to help investors to incorporate alternative datasets into the pattern recognition process in order to more accurately detect the positive patterns that yields better future returns.

Our model consisted of four main elements:

1. Historical coin price for a time-series analysis
2. Candlestick analysis via computer vision to identify popular patters that indicate price movements
3. News publication sentiment analysis
4. Social media post sentiment analysis