Fanservice: Win Probabilities for Every Scenario/Format in a Seven Game Series

Posted on 09/22/2010 by


I know, I know, you’re probably saying to yourself: another piece on seven game series and win probabilities, what’s wrong with this guy.

I’m well aware that this is my fifth piece on the subject(see here, here, here and here). I know I should be working on the rookie model and the next version of The Build. But see, I made a promise to one of my readers and here at Arturo’s Silly Little Stats we’re all about the fanservice.

Official Assistant GM of this blog, Devin Dignam of NBeh? caught an error on my part and I gave him a no-prize:

No Prize

His prize was to name his post and he came up with the following suggestions:

  • What is the relationship between college PAWS and NBA WP48? David Berri did some work on it but it was more general. Can you find any hard numbers on the subject? If we look at a specific player’s college PAWS, what is the likely progression for their NBA WP48 in year 1, year 2, year 3, etc?
  • Are there any statistical differences between left-handed and right-handed players? I’m left-handed, and left-handed issues always interest me. Based on my playing experience and my non-statistical theory, left-handers may be better at blocking shots because their dominant hand is on the same side as most shooters’ shooting hand. They also may be more likely to commit turnovers because their dominant dribbling hand is on the same side as defenders’ dominant hand (although I never notice this when I play, it’s just a logical offshoot of the blocking theory). As a consequence, there may be a greater proportion of left-handed forwards and centres (blockers) and a smaller proportion of left-handed guards (ball handlers). Of course entering all the handedness data may be tedious, so I found this list that could help.
  • Does the change in series format from 2-2-1-1-1 to 2-3-2 in the NBA finals influence the outcome? Not sure how to measure this. Maybe compare the number of upsets between the two? Of course the sample size of 2-3-2 series isn’t very large. I dunno. Given that the better team usually wins and that the better team may not have homecourt advantage…maybe there is no difference between the two.

See, this is why I’d hire Devin. Similar thought patterns. Number 1 sounds suspiciously like the rookie model doesn’t it? Number 2 is interesting but it sounds hard enough to be a longer term project. Number 3 however is ready to go today .

Now don’t worry the rest of this post is math free. I did the ridiculously long probability equation required for this during my lunch break and I won’t post it here.

(Image courtesy of

Enough talk let’s start with the 2-2-1-1-1 format:

But it looks prettier as a graph:

What do we learn here? Home team winning in 5 (if the home team is better) or losing in six (if worse) seem to be the smart bets unless the teams are even in which case the bet is home team in 7.

And now the  2-3-2 format:

And of course the pretty chart:

What do we learn here? For the the 2-3-2 Home team winning in 6 and not 5 (if the home team is better) or losing in five not six (if worse) seem to be the smart bets unless the teams are even in which case the bet is home team in 7.

Now please keep in mind that this blog does not in any way shape or form promote the practice of gambling.

Posted in: Uncategorized