Continuation of 2021’s WallStreetBots with a twist -- crypto price prediction.

Introduction

The volatile nature of cryptocurrency prices and its being driven by supply and demand makes machine learning an appropriate tool to predict their prices.

Crypto price prediction depends on knowing people’s beliefs about them and how these sentiments will affect trading behaviors based on the current market conditions. This project touches on both aspects of the task. First, we will scrape data from different sources (including Twitter, Reddit, and The New York Times) to create NLP sentiment indicators describing how society feels about crypto. We will then combine these indicators with intuitive trading predictor features based on market data to use as inputs to state-of-the-art machine learning techniques to predict the future price of cryptocurrencies. Our predictions will form the basis for an automatic trading strategy to be tested on an online platform already developed by WallStreetBots last year.

We are looking for developers to work in the three subteams of WallStreetBots: (1) market predictor feature engineering team (finding intuitive predictors based on market data to use as inputs to the final prediction model), (2) prediction model team (designing and fine-tuning specific deep learning models to predict the crypto price based on input features), (3) NLP team (designing clean NLP sentiment indicators to use as inputs to the final prediction model). We look forward to your participation in the project.

Developers should be proficient at Python and have experience with basic machine learning frameworks/libraries. Knowledge and interest in trading is a plus. Time commitment is 1-3 hours per week.