Algorithmic Trading AI with Python Day 3 Data Finalization

All right, we are wrapping up our data prep for our algorithmic trading AI using Python. Last night I thought about some small tweaks to how we store our data. When we pull data from YFinance, it gives us both the market open and close, but I realized we do not need to overcomplicate it.

We can just step forward by the time period we care about and use those prices directly. That way, our “close” becomes our “future price” and we are set. We are cleaning up our folder a bit, getting rid of some old CSVs, and keeping our main price data in a simple file called “download” now.

We are using BTC as our main ticker because it is super active, has a ton of news, and is really volatile. This is perfect because we want lots of up and down moves and lots of news data for sentiment. Down the road, we could add things like gold, Tesla, and Apple since they also have a lot of volume and movement.

One thing I wanted to change was the name of our main download file, just to keep it short and easy. As for our data, we are matching news events to price changes and calculating the percent difference so we can feed that straight into the model. We hit a bug where current price or future price came up “None”, turns out they were out of sync or missing values because of our indexing, but we fixed it by checking for missing data and skipping those rows.

The script is working now and giving us the differences we want, like $62 or $141 between current and future price. This way, we can clearly see moves that could mean profit. Next, we just need to run this through the AI, which we are building with PyTorch transformers and embedding models.

Thanks to everyone who’s hanging out and helping! If you want to grab the latest code or play around with it, it is all up on Discord.