{"id":1369,"date":"2017-01-06T16:22:09","date_gmt":"2017-01-06T22:22:09","guid":{"rendered":"http:\/\/kenpom.com\/blog\/?p=1369"},"modified":"2017-12-21T15:17:46","modified_gmt":"2017-12-21T21:17:46","slug":"basketball-hangovers-are-difficult-to-diagnose","status":"publish","type":"post","link":"https:\/\/kenpom.com\/blog\/basketball-hangovers-are-difficult-to-diagnose\/","title":{"rendered":"Basketball hangovers are difficult to diagnose"},"content":{"rendered":"

Last Saturday, the two presumed favorites in the ACC, Duke and North Carolina, lost their respective conference openers. The teams that beat them, Virginia Tech and Georgia Tech, proceeded to perform incredibly poorly in their next game. That might make one wonder if the basketball hangover exists. In other words, does a team typically\u00a0have a letdown after it plays Duke (or some other high-profile team)?<\/p>\n

To figure that out with some accuracy, one will need to control for many variables.\u00a0But that can be time consuming and not something your humble correspondent can accomplish on a Friday morning. For now, let’s see how far we can get.<\/p>\n

In\u00a0all of my experiments, I have limited the possible scope of the hangover to regular-season conference games. For my first try, I will\u00a0look at a team’s record after they played the potential hangover team. When an ACC team plays Duke, what it the outcome of their very next game?1<\/sup><\/a><\/span><\/span> I am using all games from\u00a0the 2001-2002 season through last season. Let’s see the ten-best and ten-worst hangover teams by this method.<\/p>\n

Strongest Hangover                      Weakest Hangover<\/strong>\r\nArizona St.            98-150 .395      NJIT                   22- 14 .611\r\nUT Rio Grande Valley   27- 41 .397      Pepperdine            125- 83 .601 \r\nLipscomb               84-127 .398      Oregon                150-102 .595      \r\nMiami OH               91-135 .403      Hawaii                128- 90 .587\r\nNorth Carolina A&T     89-128 .410      Creighton             142-100 .587\r\nPrinceton              72-103 .411      Harvard               105- 75 .583\r\nRice                   93-133 .412      Bethune Cookman       125- 90 .581\r\nSan Diego              85-120 .415      Davidson              131- 97 .575\r\nPortland               83-115 .419      Jackson St.           145-108 .573\r\nAlabama A&M           106-146 .421      Gonzaga               118- 89 .570\r\n<\/pre>\n
Arizona State has a great reputation as a party school.\u00a0Are opposing teams getting caught up in the culture enough to affect them in their next game? Initially, I thought I might have\u00a0broken a scandal that would\u00a0rock the college basketball world. But\u00a0after further thinking, I don’t think so. These\u00a0lists are largely an exercise in random-number generation modulated by two things.<\/p>\n
First, note that historically poor teams tend to be in the strong hangover group and stronger teams tend be in the weak hangover group. The reason for this is that if you’ve just played Gonzaga, you are very likely not playing a team as good as Gonzaga in your next game. It’s a small thing, but over 200 games one can start to see the effect.<\/p>\n
For some of these teams, the larger issue is one\u00a0of travel partners. The Pac-12 and Ivy League have maintained consistent travel partners over the span of the study period. If a team has just played Arizona State, then a little under half the time its next opponent will be Arizona. Likewise Princeton’s opponents are often\u00a0playing Penn next, Oregon is\u00a0paired with a typically-weak Oregon State and Harvard is similarly paired with Dartmouth.<\/p>\n
But one drink\u00a0doesn’t usually cause a hangover. It’s that excessive celebration that becomes the issue. Likewise, it may not be that simply\u00a0playing<\/em> a specific team causes a letdown. It’s the wins over those teams that does it. So for my next trick, let’s take a look at how a team’s opponents perform after those opponents experience a victory over the potential hangover team.<\/p>\n
Strongest Hangover                       Weakest Hangover<\/strong>\r\nArmy                   57-82   .410      Gonzaga                17- 5   .773\r\nUT Rio Grande Valley   19-27   .413      NJIT                   12- 4   .750\r\nMiami OH               43-59   .422      Louisville             54-19   .740\r\nNorth Carolina A&T     54-72   .429      UMass Lowell           19- 7   .731\r\nRice                   63-80   .441      Oral Roberts           41-16   .719\r\nSacramento St.         62-78   .443      Monmouth               79-34   .699\r\nMiami FL               54-64   .458      Southern Utah          99-46   .683  \r\nLIU Brooklyn           56-65   .463      Yale                   50-24   .676\r\nAuburn                 72-83   .465      Pepperdine             79-38   .675\r\nOhio                   48-55   .466      IUPUI                  64-31   .674\r\n<\/pre>\n
This approach presents a new set of problems. If you were good enough to beat Gonzaga, then you were probably good enough to beat the next team on your conference schedule. A conference\u00a0outlier in either direction has\u00a0a better chance of showing up on these lists, but only looking at wins has reduced the sample dramatically, which adds more noise.<\/p>\n
Ok, let’s cut the crap and add some sophistication. For my final experiment, let’s look at how teams did in their next game compared to expectation. I’ll take my final ratings to produce an expected point differential in every game and then compare the margin of the next game to what was expected. Here are the teams that produced the strongest and weakest hangovers using this method.<\/p>\n
Strongest Hangover                     Weakest Hangover<\/strong>\r\nAbilene Christian     -3.49  16-17     Northern Kentucky     +3.47  26-14\r\nUT Rio Grande Valley  -2.64  19-27     Nebraska Omaha        +3.25  18-11\r\nGeorgia               -2.13  66-60     Grand Canyon          +3.03   9- 5\r\nNorth Dakota St.      -2.11  29-24     Wichita St.           +2.36  38-35\r\nMiami OH              -2.06  43-59     New Mexico            +2.28  59-31\r\nUTEP                  -2.02  55-43     IUPUI                 +2.20  64-31\r\nHouston               -1.98  58-61     Oral Roberts          +2.11  41-16\r\nHampton               -1.92  41-34     Lafayette             +2.04  68-43\r\nLehigh                -1.88  45-43     Cal St. Bakersfield   +1.96  10- 7\r\nLipscomb              -1.83  46-52     Albany                +1.93  64-40\r\n<\/pre>\n
This method should control for most of the issues with scheduling difference and other structural quirks in each conference. There are still some issues you could point to with respect to inaccuracies in my ratings method. But even so, I think we’d expect to see something more than continued random-number generation.<\/p>\n
Historically, teams beating\u00a0Abilene Christian struggle the most in their next game, falling short of the expected margin by an average of 3.49 points. Among teams that existed throughout the span of the study, a win over Georgia has apparently provided the biggest hangover. On the flip side, Wichita State and New Mexico, two teams with good track records over the past decade-and-a-half, provide the biggest benefit for teams that beat them. We can say with some confidence that a\u00a0big win over those teams doesn’t have a negative impact\u00a0for the next game.<\/p>\n
Furthermore, teams beating Duke have under-performed by an average of just .19 points in their next game. Teams beating Kansas have under-performed by .71 points, and teams coming off a win over UNC have over-performed by .95 points. There doesn’t seem to be a pattern that’s related to the prestige or quality of the team being beaten.<\/p>\n
These lists largely illustrate how difficult it is to identify emotional factors. Humans generally identify these traits after the fact because they don’t have much skill doing so before the game. There are a bunch of things\u00a0that affect the outcome of a game, and while having just beaten the best team in your conference may be one of them, that knowledge appears to only be useful in combination with many other things.<\/p>\n
 
 <\/span>[+<\/a>]<\/span><\/p><\/div> 
 \r\n\r\n
References<\/caption> 
 ^<\/span>1<\/a><\/th> If the next game is against a non-conference opponent or a tournament game, it is ignored.<\/td><\/tr>\r\n\r\n <\/tbody> <\/table> <\/div><\/div>