Introducing “Jesse”, an open-source AI built-in crypto trading framework
Have you ever had a trading strategy in mind but didn’t know how to actually test it without losing money? Does the Monte Carlo simulation sound cool but hard to implement? How about the Genetic algorithm? Do you wish you could use machine-learning’s help in your trading?
You may think all of this sounds too good to be free and open-source, or that only giant hedge funds deserve to have such tools for trading. Behold to meet Jesse, an open-source AI built-in crypto trading framework.
I have named the project “Jesse” in the honor of the legendary trader Jesse Livermore. Only if Livermore had access to Jesse!
Enough chit-chat; let’s talk about a few cool features of Jesse:
- A simple yet advanced interface to implement your custom strategies.
- Use of WebSocket connection to interact with the exchange.
- An advanced backTest mode that takes care of back-testing your strategy with historical data.
- Genetic algorithm integration used for tuning profits (or any other metrics).
- Deep learning integration using Google’s TensorflowJS.
- Written in TypeScript, running on nodeJS, which is one of the most adoptable/common languages on the Internet.
- A LOT more!
Jesse is not designed to necessarily give you fishes, but to teach you how to fish. It is designed to find objective trading strategies that are testable and actually work.
My goal is to turn Jesse into an advanced trading framework from both programming and trading aspects. That being said, whether you’re a trader or a developer, your contribution to the project would be welcomed.
Jesse is way bigger than what I’ve explained so far. Ask me questions, write comments, etc; so I get some ideas.
I’m working hard on the initial release. The repository will be available at https://github.com/jesse-ai/jesse. Don’t forget to spend 3 seconds to star it as a way of support.