Okay, so today I decided to tackle something I’ve been seeing all over the sports betting side of the internet: a “Daily Home Run Predictor.” Sounds fancy, right? I figured, why not try to build one myself? I’m no statistician, but I do love baseball and tinkering with data, So, I started.
Diving In
First thing I did was try and find a reliable source that could provide me with the data I needed.I figured I needed stats, lots of them.
Gathering the Raw Materials
So, with data sources in hand, I started pulling in information.I decided to focused on things.
- Batter Stats: Obvious, right? I grabbed things like recent home run totals, at-bats, maybe even their home run percentage over a certain period. The more the merrier, I figured.
- Pitcher Stats: Just as important. I looked for the pitcher’s ERA, how many home runs they’d allowed recently, and maybe even their home runs allowed per nine innings.
- Park Factors: I’d be a fool. Some parks are just easier to hit homers in.
This part took a * was a lot of copying, pasting, and trying to make sure I didn’t mess anything up.I have to do to much manual work for my * I will automate this in the future.
The “Algorithm” (and I use that term loosely)
Now for the “brains” of the operation. Look, I’m not building some super-complex AI here. I basically wanted to create a simple scoring system. I did this:
- Batter Score:I gave points for recent home runs. More home runs, more points. I also factored in their home run rate – a guy who hits a homer every 10 at-bats is more valuable than a guy who hits one every 30.
- Pitcher Score: This was the opposite. Pitchers who gave up lots of homers got points subtracted. A low ERA helped, but I really focused on those home run numbers.
- Park Factor Score:If a game was in Coors Field(A notorious hitter’s park), I added some bonus points.
I added all those scores up for each batter-pitcher matchup. The higher the total score, the more likely I figured that batter was to hit a dinger that * is basic but effective.
Testing and Tweaking (and a lot of guessing)
I picked a few games and ran my little “predictor.” Did it work perfectly? Of course not! I made a lot more work for myself.
- I realized I needed to weigh some stats more than others. Recent performance probably matters more than what a guy did two months ago.
- I started looking at even more stats, like how a batter performs against left-handed vs. right-handed pitchers.
The tweaking is never-ending, It has made things more difficult and I need a better way to keep things orgainized.
The Verdict (So Far)
My “Daily Home Run Predictor” is very much a work in * provides me with a very simple score and I have to interpet the data to come up with a prediction, but I can’t see myself giving up anytime soon.
It’s fun to play with, and hey, even if I’m not hitting every prediction out of the park (pun intended), I’m learning a ton about baseball and data along the way. I’m already thinking about ways to make it better. Maybe I’ll add weather data? Or how about the size of the outfield? The possibilities are endless!
