beckdawg
Well-known member
- Joined:
- Oct 31, 2012
- Posts:
- 11,750
- Liked Posts:
- 3,741
you're conflating two disparate things, games for a team and at-bats for an individual. considering that the length of a season is 162 games, i don't understand your comment about volatility.
162 at bats isn't enough data points to remove the statistical noise. 162 games isn't enough games to remove statistical noise. MLB is the best of the sports leagues in that regard but it doesn't mean it's a great sample size. Typically when you talk about statistical data sets you want 500 of whatever you're talking about as a bare minimum and you'd like to have 1000+. Point here being that run differential would be more accurate over a larger sample size because you are able to remove the statistical noise. In the case of baseball, 162 games isn't great in that regard and that's why run differential is spotty in terms of record prediction. The example CSF77 is a perfect example of this. While an 8-0 game does occasionally happen, it isn't very common. So, over larger sample sizes that evens out more. The larger the sample size the less impact individual games like that have.