Men's Basketball: Coaching records in games decided by 5 points or less

Submitted by Bambi on December 10th, 2018 at 7:26 PM

Jordan Sperber (@hoopvision68) is a really good follow on twitter who puts out a lot of really god college basketball content. Earlier today he tweeted out a graphic of the W-L record of games withing 5 points of all coaches who have coached at least 50 such games in the KenPom era.

Link to tweet

It's a fun graphic to look at and try and find where different coaches are. Beilein is all the way to the right, coaching the second most such games (~137) with a winning percentage slightly above 55%.Tom Izzo is slightly to the left of Beilein in games played (~113) with a slightly lower winning percentage (right under 55%). Chris Holtmann is down at the 60ish game mark with only a 47%ish winning percentage.

Above a 60% winning mark we have some interesting names like Dana Altman, Brad Stevens, Archie Miller, Mark Few, Steve Fisher, Bill Self, Ed Cooley and Jim Boeheim. The two most interesting to me are Andrew Toole (head coach at Robert Morris since 2010), who's leading the pack as the only guy above 65%, and MGoBlog favorite Dan Dakich right at 60% at 50 games coached.

On the other extreme the names are less notable but we do have UConn coach Dan Hurley along with Fran McCaffery and Steve Pikiell to bring up the rear.



December 10th, 2018 at 7:51 PM ^

Nice plot. So the horizontal position should be relatively unimportant. It's a measure of how long you've been coaching (modulo the relative competitiveness of that coach's games in general). 50 games is enough beat down the noise. Coaches above 50% are just better coaches than coaches below 50% on this plot. 


December 11th, 2018 at 1:06 AM ^

50 games is enough to beat down the noise?  No.

Flip a coin 50 times. You're about as likely to get 56% or more heads or tails than you are to get between 44% and 56% of one or the other.

That said, games decided by 5 or 4 points or maybe even 3 points weren't coin flip games so of course the better coaches would tend to sit above 50% here.  They will win more games and some of those games will be decided by 5 or fewer points.

I imagine if this was done on 1 and 2 point games you'd see an almost perfect binomial distribution.


December 10th, 2018 at 8:17 PM ^

There is however another level to this.  If you are an underdog but helping your team play above their expectations more often than you are the favorite in a close game that you were expected to win it could mean that a lower percentage is still indicative of the better coach.  


December 10th, 2018 at 8:24 PM ^

This is interesting. One flaw in it though is how many 5 point games you've been involved in vs how many games youve coached. Having a high overall win % but a low number of 5 point games involved in would show more dominance, especially if that 5 point game % is high.


Only reason I say that is because naturally scatter charts like this make your eyes go to the top right plot point and assume that's the best, which isn't necessarily true; the top left point could be just as impressive


December 10th, 2018 at 8:41 PM ^

Kind of just looks like random distribution to me.  I wonder how much you can really read into which coaches come out looking better or worse from this.