Ever since breakthroughs in self-driving cars, super-human chess bots and robotic manipulation thanks to advanced Reinforcement learning, this latter has been one of the most trending research concepts in artificial intelligence and machine learning.
Cryptocurrencies are electronic and decentralized alternatives to governmentissued money, and cryptocurrency trading has been one of the hottest topics in social media in the last year with the huge rise in most cryptocurrency prices and the hype from tech moguls like Elon Musk.
This project aims to join these two concepts and create a reinforcement learning system capable of favorable crypto trading, in this thesis we first of all present the host organization for which I interned and did my end-of-studies project,their values,their goals and their main clients. After that we present the context of the project, the problematic and its objectives.Furthermore, we dive deep into what cryptocurrencies are on a technical level.Moreover, we explain reinforcement learning, how it works and the reinforcement learning technique we used to solve our problematic called (Deep-Q-learning). Finally, we present our algorithm solution, how it works, how it was trained and the results obtained.