I found a number of Python models online that will predict the outcome of each day's MLB games. One of them in particular claimed to be .638 against the line. Not sure if that was for 2025 or this year, but I was sold and downloaded the model.
Using ChatGPT, I've made a system that checks every morning for that day's games and then downloads the lines on those games from eight sportsbooks. It then uses results of completed games to "train" the model... I admittedly don't really know what training means other than it supposedly gets better with each training session. Each morning the model is "trained" based on what happened the night before. I'm also using the model to make predictions on that day's games.
Today is the first day that I've run this. I'm downloading scores a few times an hour, and keeping track of the results as the night goes on. It isn't real money at this point, but this is a fun little project I've started. I'm 7-4 on the night with four of the 15 games not yet complete. Despite the house advantage, I'm still up on the night. I'd never bet on every game with real money. I'm looking for particular types of games that the model is good at predicting.
Some of tonight's results:
EDIT - The Astros just cost me $100 by losing to the pitiful Angels.
