This project aims to build an AI model because a user might spend a lot more time pouring over the stock market. This bot can do a bunch of cool things that help get information to make better trades.
It will anticipate the trading decision within a day. And the AI trading bot will update every day to make itself smarter.
In order to make the AI trading bot, we have collected the stock trading data (concretely, the forex trading data), then create an AI model that will learn how to trade by learning this data. The AI model is created with a Deep Reinforcement Learning model that will learn by trial and error, if a trade is bad, the AI will be penalized, otherwise, the AI will be incentivized. For visualization, we used Huggingface to make a trading board for a user to observe the trading process.
In terms of the AI model, we built it from scratch, The visualization is hard because there are too many candles to plot.
Our AI model is quite smart when it comes to daily trading. Our visualization is dynamic and astonishing.
We learned a lot from this hackathon, concretely, we learned how to build an AI model doing stock trading, then we learned how to use a new tool for deploying our model. And the greatest part is that our team worked together for 24 hours to make a great product, we learned how to collaborate, how to prioritize the task, and have fun together.
In the future, we want to make the bot have the capability to predict the trading action for various datasets such as Forex, Bitcoin, etc. We also want to upgrade our AI bot to make it smarter and faster.
gradio
pandas
plotly
python
pytorch