- Posts: 1313
- Joined: Sat Mar 03, 2018 12:07 am
- Location: Usually Somewhere Else
I think Charlie hit it squarely for me.
There are definitely some strange black box things that no one can explain fully because there is lacking transparency.
But, I think there is some possible confirmation bias with the 100 win team results.
100 win teams are generally powerhouses in their division. They are generally well matched to it too.
It is very plausible they wouldn't be well matched against another division opponent. My last four 100 win teams split--two lost in semis, two won finals.
I think it is also not ironclad that run diff is predictive in all situations. Real baseball is a little different due to the general amount of parity. Most of our leagues lack parity. And it is very easy to create statistical situations that are not normally distributed, such that the assumptions inherent in a W-L prediction from a run diff calc are not high confidence.
Whether it is due to that lack of parity creating outlier scores in some matchups, or just team composition (such as a smallball team that wins small and occasionally loses big), there are valid reasons that are possible.
Then again, I just had a playoff result that seemed to contain a lot of low probability results that were one sided...so...
But the momentum thing--I agree this falls somewhere in the black box of a Max Rules setting, and if SOM wants to disprove it, open the black box and let us see when "rolled" results have been altered to "correct the statistics".
There are definitely some strange black box things that no one can explain fully because there is lacking transparency.
But, I think there is some possible confirmation bias with the 100 win team results.
100 win teams are generally powerhouses in their division. They are generally well matched to it too.
It is very plausible they wouldn't be well matched against another division opponent. My last four 100 win teams split--two lost in semis, two won finals.
I think it is also not ironclad that run diff is predictive in all situations. Real baseball is a little different due to the general amount of parity. Most of our leagues lack parity. And it is very easy to create statistical situations that are not normally distributed, such that the assumptions inherent in a W-L prediction from a run diff calc are not high confidence.
Whether it is due to that lack of parity creating outlier scores in some matchups, or just team composition (such as a smallball team that wins small and occasionally loses big), there are valid reasons that are possible.
Then again, I just had a playoff result that seemed to contain a lot of low probability results that were one sided...so...
But the momentum thing--I agree this falls somewhere in the black box of a Max Rules setting, and if SOM wants to disprove it, open the black box and let us see when "rolled" results have been altered to "correct the statistics".