I am in no way willing to acknowledge that any team in particular has a streak they haven't beat since 2008. But what information does a winning streak tell us about what the expected win % is for a team?
Horribly bad assumptions upfront:
- Win expectancy is constant throughout the season, I know this is not true.
- Win variance is constant throughout the season.
Our Null Hypothesis, Ho, is that the teams win % is W%.
Our Alternative Hypothesis, Ha, is that the teams win % is > W%.
- Given that the season is 56 games long so far, n = 56.
- Given that the max streak for that team so far is k.
- We will reject Ho for P less than 5%.
P =1-((1-W%*X))/(k+1-(k*X))(1-W%) *1/X^(n+1) ;
X=1+((1-W%) (W%)^k)+(k+1) ((1-W%) (W%)^k )^2
So if a teams current streak is 7 games in a season only 56 games old we would expect them to have a win % of at least 40.14% or higher. Which means we would expect that team to lose no more than 97 games in the entire season.
Not that these results apply to any team in particular, or that I would update them if needed tomorrow.