by Ken Pomeroy on Tuesday, March 26, 2013
Explaining the game-to-game variation in a team’s performance is difficult. Sometimes, a team’s star player just plays well, or the opposing team plays more poorly than usual, or a team avoids foul trouble, or most of the dozen or so 50/50 calls go a particular team’s way. These explanations are uninteresting and not satisfactory, though, and so observers must come up with more meaningful explanations whether they are real or not. Be it a famous alum giving a pep talk before the game, a team focusing on starting the game better, or an injured player returning to the lineup.
Evaluating the impact of of a player’s injury is perhaps something that highly-trained humans can do better than computers. Not that computers can’t help in this regard, but it’s a complex problem, which is why I’ve been disappointed in the approach to this sort of analysis in some cases this season. Maybe it just seems like it started this season, but instead of critically evaluating how a player’s absence affects a team, the experts now just look at a team’s record with and without that player on the floor.
There’s some value to this, of course, but given the varying quality of opposition a team faces during the season, looking simply at W’s and L’s is not the most robust approach. Besides, this is something a computer can do as well as a human. Humans are selling themselves short if this is their method of choice.
This approach first came to my attention when Miami beat Duke by 27 on January 23. They had lost to Florida Gulf Coast in November when Durand Scott was serving a suspension. They lost to Arizona and Indiana State without Reggie Johnson. But after the Duke win, it was frequently noted that the Hurricanes hadn’t loss with both players available. Keep in mind that at that point, Miami had played just six games with a full contingent which is a frighteningly small amount of data from which to draw conclusions. But that didn’t stop people from declaring that Miami could be one of the best teams in the nation and that any games played without Johnson or Scott should be ignored.
(It’s an interesting twist that Miami’s breakthrough win over Duke occurred without Ryan Kelly in Duke’s lineup. Upon Kelly’s return in February, he was given similar status to Reggie Johnson. All of Duke’s games without Kelly, including that lopsided Miami loss, were suddenly stricken from the record. The conclusion was that the Blue Devils were unbeatable with Kelly in the lineup.)
It figures that Reggie Johnson’s availability helps Miami, but given that he’s played no more than 24 minutes in any game since his return, one should be skeptical whether his presence alone lifts Miami from a middle-of-the-pack ACC team to a national title contender, especially considering Johnson’s 88 offensive rating this season. Perhaps the losses without him in the lineup are data points that should be discounted, but they shouldn’t be completely ignored. As the ACC season continued, we saw Miami play very well at times, and somewhat less inspiring at other times.
This concept continued with similarly fuzzy logic when evaluating Oregon. The Ducks struggled big-time in February, going from 7-0 in conference play to 12-6. The low-point came in lopsided losses to Colorado and Utah in the final two games of the regular season, when the Ducks controlled their own destiny for a solo regular-season Pac-12 title.
At that point, there was the natural urge to explain the difference in play between late-season Oregon and early-season Oregon. That was largely done using the Ducks’ freshman point guard Dominic Artis, who played a somewhat prominent role in the Oregon offense as the Ducks jumped out to a 17-2 record. Artis injured his foot and missed the Ducks next nine games, where Oregon compiled a 5-4 record as Johnathan Loyd took on most of Artis’s vacated minutes. Artis has played the Ducks’ last eight games and Oregon has gone 6-2. And they’ve been dominant in their two NCAA tournament games, fueling the argument that Oregon is somehow historically-underseeded based on their performance at full strength.
Nevermind that Artis has averaged just 17 minutes a game since his return, and hasn’t really been the same player he was in November and December. He’s made 24 percent of his twos, 32 percent of his 3’s and has been to the free throw line a total of seven times since coming back. In Oregon’s overtime win over Washington in the Pac-12 quarterfinals, with the Ducks tournament hopes possibly hanging in the balance, he didn’t even play in overtime. He was also in the lineup for the disappointing losses to Colorado and Utah. Attributing all or even most of Oregon’s stunning improvement in the tournament to his presence is misguided.
The response to such logic is that Artis brings intangibles to the table. No doubt intangibles exist, but it’s also the knee-jerk rebuttal to any statistical argument. In this case, the lack of tangibles weakens the position that the Ducks are different with Artis. You could more easily point to the improved play of Loyd himself, or Carlos Emory, and Oregon’s incredible run defensively since the Pac-12 semifinals to explain Oregon’s improvement in the tournament. For whatever reason, Oregon’s last 80 minutes have been their best of the season, but Artis’s presence has little to do with it.
Finally, let’s take a blind look at three guys that made a late return to teams that are still alive.
with w/o G* MPG* 2PM-A 3PM-A FTM-A A 23-4 5-4 8 17 5-21 6-19 5-7 B 13-2 15-6 5 24 6-8 9-24 12-14 C 20-1 9-4 6 30 15-28 9-26 22-26
*Since returning to the lineup
You might guess that A is Artis since I’ve revealed his statistics previously. C is Kelly, whose impact has also been overstated. Kelly does mean a lot to Duke, but he doesn’t make them unbeatable as was suggested after his 39-point game against Miami. His offensive numbers have been decidedly less spectacular since then. He hasn’t hit a three-pointer in the ACC or NCAA tournaments and scored a single point against Creighton. (To his credit, he was mostly responsible for Doug McDermott’s somewhat-inefficient 21-point game).
Player B is Wichita State’s Ron Baker, who most recently put up a 16-point game against a Gonzaga team oddly willing to let him put his marksmanship on display. Baker’s presence is obviously not as important as Malcolm Armstead or Carl Hall or Cleanthony Early, and I’m not saying Baker turns the Shockers into a Final Four team. But if Wichita State gets there, Baker’s return will have played a role. It’s not hard to make the case it’s a larger role than Artis’s, yet Baker’s return has received approximately zero fanfare.
I expect the discrepancy has more to do with groupthink than anything. Oregon was in the national spotlight, stumbled in the middle of Pac-12 play, and Artis was a good enough reason to attribute to those losses. Oregon was surely going to lose some Pac-12 games anyway, though, just as Duke would have lost some of their ACC road games if Ryan Kelly had played in them. But people need some explanation as to why Oregon has looked so good in the tournament when they struggled so much at the end of the regular season.
Despite the 9-0 start to its season, Wichita State barely cracked the AP top 25, and was promptly demoted after a road loss to Tennessee. Many fewer people have been interested in their situation than Oregon’s. However, the Shockers have also dramatically improved their play in the tournament and in their case, it’s a bit more factual to attribute part of this improvement to a late change in personnel.