-
Notifications
You must be signed in to change notification settings - Fork 1
Home
Welcome to the Baseball-Simulator wiki!
How does it currently handle extra innings?
I think this is OK, but not sure
Fixed. Now handles extra innings as it should. Interesting to note the possibility of extra innings adds .7 runs/game on average
Regress to the mean
If not enough data for a player, we should “regress” to the mean, using the league average of similar players (eg. a Pitcher should not bat .250)
Are combinations normalized
When we combine a pitcher and a hitter, make sure the results are normalized.
Platooning
Incorporate L/R splits
Assumptions to be modified
- A double moves all runners up 2 bases
- A single moves all runners up 1 base
Park effects
Research how we can add park effects to the model!
Notes on streakiness
Batting streaks hold little predictive power, and are not worth modeling.
Pitching “hot” streaks hold some predictive power, and should be modeled at some point
Notes on Batter vs. Pitcher stats
Don’t rely on a small sample size of batter vs pitcher stats, instead it is much more reliable to focus on the larger sample of his overall at-bats
Calculate League Averages
We need to calculate league averages using last years numbers, both for regressing to the mean, and for the batter/pitcher combination
Additional Things to be modeled
- Stealing
- Sac Fly’s
- Sac Bunt [ <1 out + batter < .230 ]
-Pinch Hitters
-Pitchers/Bullpen [ >3 runs let up & >= 2 men on base ] … etc
Final Notes
The current setup provides a good baseline on a couple different levels. It allows us to keep track of the time it takes to run the simulation as we add new pieces, and it also allows us to see how the runs/score/wins are affected.