## NCAA Basketball Win Probability

Live win probabilities for NCAA basketball games are now available at wp.advancednflstats.com/bball. The model's approach is very similar to my football model. Basketball is a much simpler sport, though. There's no field position, down, or distance in basketball. Score and time remaining are really the only significant factors.

When the NFL season ended, I still had a framework for real-time win probability graphs. All I needed to change was the data and model behind it. In fact, I could pretty easily do it for other sports such as the NBA, NHL, or even MLS. I'm not planning on launching AdvancedBballStats.com or anything. This is all just for grins.

I've written up some details on the model at a guest post over at Dave Berri's Wages of Wins site. Comments and suggestions are always appreciated.

### 12 Responses to “NCAA Basketball Win Probability”

1. Jon says:

Although basketball doesn’t have the same strategy elements as football, there are some interesting potential applications of WP in Dr. Naismith’s creation–when to start fouling, when to slow down the game, the value of simply possessing the ball, or how much the ref’s bogus call really swung the game.

I'd be interested in a discussion of stuff like this and when to rotate players in or when to bench a guy with foul trouble. If someone is out there doing this, I'd like to know about it.

PS - the link that worked for me is
http://wp.advancednflstats.com/bball

2. Brian Burke says:

Link fixed. Thanks Jon.

3. Anonymous says:

Have you seen Bill James' Lead Calculation formula?
http://www.slate.com/id/2185975
How closely does his formula match your win probability?

4. Mike says:

Looks like the current games page has stopped updating.

5. Mike says:

Brian,

I noticed with your model that, for example, a 3-point lead at halftime produced a 76% likelihood of winning for the leading team. (Taking this from two tourney games last night.)

But I recently read a PennU (Wharton) study where the authors found that a 3-point lead at halftime produced only a 67% likelihood of winning (see the final graph at that link). Quite a discrepancy with your numbers. How do you explain the difference, and which number do you think is more accurate?

6. Brian Burke says:

Home court advantage I'd guess. At least until I have time to review the study you linked to more. I read about that study in the news recently. Do you have a link to the full paper?

That brings up a good point. For now, home court advantage is still in effect in my system. When I have time Saturday, I'm going to remove it to account for the fact that almost all remaining games are at neutral sites.

Which do I think is more accurate? From what I understand, the other model accounts for all kinds of things, such as team talent levels, timeouts, possession arrow, etc. I don't--which isn't always a bad thing. A team being up or down signals team strength, and if our variable of interest is who wins, then that's a good thing to leave in the model. If your variable of interest is something more academic or abstract, like the 'value of being down by 1,' then you would want to account for that and remove it from the analysis.

7. Anonymous says:

Hey Brian,

Love your CB Win Probability page.

I notice at the beginning of every tournament game you have the "visiting" team's Win Probability set at 47% or thereabouts. Two questions: Is that arbitrary? How did you come up with that? And, since most of these are neutral site games, is there really any home/away advantage to be had there?

Ty from Bucks Diary

8. Brian Burke says:

Ty-No, not arbitrary. That's the home court advantage. And, you're right about the neutral site games. When I get some time later today or tomorrow I'm going to remove the HCA. Just about all the games are now neutral site games.

9. Brian Burke says:

Home court advantage removed for the rest of the year as of now.

10. Mike says:

Brian,

You can get a .pdf of the full paper I mentioned above here.

11. Brian Burke says:

Thank you. I'm definitely going to read it.

12. Brian Burke says:

Texas Dawg-I just read the paper. I think the results are interesting, and as far as I can tell convincing. One reason I buy it is because I found a similar effect in the NFL. In the 4th quarter, teams down by 1 with the ball win more often than teams up by 1 with the ball.

My theory was that this is explained by risk/reward strategy choices rather than Prospect Theory motivation. I think that, generally, teams play on the conservative side of the optimum risk/reward balance. (They should typically be playing more aggressively by passing more or deeper, or by going for it on 4th down.) Teams down by 1 will play more aggressively by necessity, and teams up by 1 might tighten up, giving the trailing team the advantage.

The authors of the paper dismiss this theory pretty quickly, saying just 1 point wouldn't cause a strategy change in basketball. I think that's probably true, but it's just as true that such a small lead wouldn't cause much of a change in motivation.

See this article from last August.

### Leave a Reply

Note: Only a member of this blog may post a comment.