Kenpom - Least Likely Comebacks

Submitted by Blazefire on March 30th, 2013 at 11:43 PM

So, we all know there was a very tiny chance of Michigan completing its comeback against Kansas on Friday, as evidenced by the win probability chart Brian frontpaged earlier.

Kenpom records that at the worst, with just over two mnutes to play, Michigan had just a 0.62% chance to win.If that game happend 200 times, the math says Michigan would win one of them. (And another one would either be Schroedinger's Basketball Game or that weird 3/10ths of a child that American families somehow manage to produce on the regular, but I digress).

Was that the most unlikely comeback this season? Not even close. Kenpom Blog recounds 13 others that were even LESS likely. You'll need a membership to read about each game, but the article mentioning each one is free.

The winner? Northern Kentucky over Jacksonville, trailing by 13 with just 2.27 to go, had a .22% chance. It just goes to show, never, never EVER say die.



March 31st, 2013 at 12:08 AM ^

0.62% works out to 1 in 161 or 162.

I don't know, but thinking about it, it doesn't seem like it was quite as bleak as that percentage suggests.  With the 3-point shot, you can potentially come back from a 10-point deficit in four possessions, and 2+ minutes offers you enough time.  We had enough time there to score 14 points and give up four as well.  I guess that combination of events doesn't come up that often, but I feel like the percentage should be a little higher.

At any rate, the percentage should definitely be higher when one team has Trey Burke at PG and the other has Elijah Johnson.


kevin holt

March 31st, 2013 at 12:17 AM ^

How the fuck do you come back with 2.27 seconds and down 13?

Make a three, down ten. Foul before the inbound, they miss the free throw, you shoot the entire court and it somehow goes in with less than a second left. Down 7. Same story, clock runs out. WTF? Did Jacksonville shoot toward their own hoop by accident, or foul the other team despite being up by 13? I'm seriously not believing something stupid didn't happen.

Also can you say die after the game? I think Kansas said a fuckload of die after the buzzer.


March 31st, 2013 at 2:25 AM ^

rally started with a 4 point play, Jacksonville turned it over 4 times in the last 2 minutes, at least twice quick enough between scores to suggest NKU pressed and got turnovers with easy layups off of it.  NKU put Jacksonville on the line 4 times while trailing, Jacksonville went 2-7 resulting in getting the ball back 3 times without Jacksonville scoring.




March 31st, 2013 at 12:55 AM ^

Some useless facts about NKU:


(1) they are in FGCU's conference (Atlantic Sun).


(2) George Clooney attended school there (didn't graduate).


(3) If one counts Alaska's hockey team, the NKU Norse are only the 3rd D-1 school to have a nickname that starts with the letter "N".


March 31st, 2013 at 2:54 AM ^

2:22 Glenn Robinson III Steal. 62-72  
2:18 Glenn Robinson III made Dunk. 64-72  
2:02   64-72 Elijah Johnson Turnover.
1:55 Mitch McGary made Layup. Assisted by Trey Burke. 66-72  
1:36 Kansas Full Timeout.
1:22 Foul on Mitch McGary 66-72  
1:22   66-73 Travis Releford made Free Throw.
1:22   66-74 Travis Releford made Free Throw.
1:16 Trey Burke made Three Point Jumper. 69-74  
1:12 Michigan Full Timeout.
0:41   69-74 Ben McLemore missed Jumper.
0:41 Jordan Morgan Defensive Rebound. 69-74  
0:32 Tim Hardaway Jr. missed Three Point Jumper. 69-74  
0:32 Glenn Robinson III Offensive Rebound. 69-74  
0:29 Glenn Robinson III made Layup. 71-74  
0:28 Michigan Full Timeout.
0:21 Foul on Caris LeVert 71-74  
0:21   71-75 Elijah Johnson made Free Throw.
0:21   71-76 Elijah Johnson made Free Throw.
0:14 Trey Burke made Layup. 73-76  
0:13 Foul on Trey Burke 73-76  
0:13   73-76 Elijah Johnson missed Free Throw.
0:13 Tim Hardaway Jr. Defensive Rebound. 73-76  
0:05 Trey Burke made Three Point Jumper. 76-76  
0:05 Kansas Full Timeout.
0:01   76-76 Naadir Tharpe missed Three Point Jumper.
0:00 End of the 2nd Half.



March 31st, 2013 at 8:28 AM ^

It's interesting to go back and sift through the box scores and play-by-play for some of these  games actually to see where teams actually found the "extra gear", if you will.

For example, in UCLA's loss to Cal Poly, with 12:03 to go, Cal Poly was 8-21 from the field and UCLA was 14-26, but for the final 12 minutes, Cal Poly went 16-33 and UCLA went 10-27. In the Brown-Providence game, looking at the play-by-play, Brown made three three-pointers in the final 55 seconds to the Friars' single free throw. The Texas-Oklahoma game mentioned by Kenpom saw the lead change twice even in OT, but a few Oklahoma personal fouls and Texas going 2-2 from three-point land and 5-7 from the free throw line seemed to seal that one.

As we discovered in person with the PSU game, but also with Kansas, being on the losing end of finishes like that seems doubly painful, but when you're on the winning end of fantastic finishes such as these, it results in no less than 3,500 appreciation / posbang threads, and that as it probably should be. 


March 31st, 2013 at 9:33 AM ^

There are 200ish teams playing 30ish games in college bb. That's 6000ish game, which means there should be 40 or so 1:150 odds comebacks. If there were only 14, it tells you a lot about the ability of anyone to estimate tail events. It also tells you about how bad humans are at intuiting liklihoods. Kenpom win probability metrics were clearly never meant to be used for this type of analysis. it was a great comeback, but we need to not read too much into it. It wasn't destiny, it was luck and a bit of perseverance.


March 31st, 2013 at 10:04 AM ^

I don't think that's how this works.  For a 1:150 comeback to occur the game needs to develop a certain way, with a team trailing 10-15 points in the last 3-4 minutes.  If the game is closer it won't qualify as a comeback.  If it's more of a rout there comes a point where you could replace the trailing team with the Miami Heat, it just won't happen.  That eliminates a lot of games.

The 1:150 odds are for a team in a certain kind of situation, and not every game leaves a team  in a position where they can, or have to, put together the kind of run Michigan did at the end in order to win.

(BTW: I think there's actually something like 300 teams playing D-1 basketball now, don't remember the exact figure.)

almost as old …

March 31st, 2013 at 10:13 AM ^

Insightful comment re: tail events, but might not your view be thrown off a bit by doubling the number of games?  One assumes that the teams play each other, so the 6000ish games become 3000ish games, so your statistifcally predicted 40 should be about 20, at which time 14 is not quite the unusual nuber it seems. 


March 31st, 2013 at 10:11 AM ^

If there were 13 other games with a lower probability by Kenpom's standards, then maybe there's something wrong with the calculations.  I don't believe the chances of comebacks like that are that low.


March 31st, 2013 at 10:44 AM ^

...the only thing objective statistics can't factor in is the relative emotional gravity of the game situation. The late-round tournament scenario makes the comeback even more dynamic vis-a-vis a comeback during the regular season when wins and losses don't have quite the same dramatic impact.