by Ken Pomeroy on Monday, April 21, 2014
Whose listed weight is the most misleading? You didn’t know you wanted to know that and I didn’t either, until I started trying to predict a player’s weight based on other information. And the only reason I did that was because sometimes a player’s weight is not revealed. Mostly in cases of players on Utah Valley’s roster as the Wolverines are the lone team in college hoops that treats its players’ weights like Bill Belichick treats his players’ injury information.
You might think it is really dumb to be bothered by such minutiae, and that’s fine, but you are not the intended audience of this site. Go to some other place and read “normal” things about “people” and stuff. In the real world, when one entity out of 351 refuses to do something the other 350 are doing, it can create a headache. I mean, that “effective weight” stat people have been demanding is not possible because of the Wolverines.
But there is no need to let Utah Valley hold us back from having a complete data set of players’ weights. One can look at available players’ weights and all of their accumulated statistics and figure out what matters when predicting what a player’s listed weight will be. And this produces a pretty decent estimate. By pretty decent, I mean a standard error of 13.5 pounds for the 2014 season by using data from the 2007 through 2013 seasons to build the model.
It turns out some non-statistical data is useful in making predictions. Naturally, height is the most important thing and fortunately, all teams are willing to publish this information. But other stuff helps, too. A player’s age would figure to be predictive, but I don’t have that. However, eligibility class is a decent proxy and useful. Given players of the same height, the older player will tend to weigh more. Each year is worth about two pounds on average.
But there’s more! Take two players of the same height, but one is listed by his team as a center and the other is listed as a forward/center. The latter will weigh about five pounds less than the former. You can roughly apply that five pound difference as you ascend to pure forwards, guard/forwards, and pure guards. Also, players on better teams tend to weigh more, as do players in better conferences. These factors are included in the estimates.
One last non-statistical piece of information that’s important is uniform number. It’s understood that taller players tend to choose larger numbers, but all things being equal, heavier players wear bigger numbers as well. The player wearing 55 tends to be five pounds heavier than his statistical twin wearing zero.
I suspect one could find value in a player’s hometown as well. That would take quite a bit of effort, and I think I have a decent enough model already. I mean, I’m just trying to figure out whose weight is the most misleading. The fact that I’m writing 1,000+ words on this is silly enough.
There are true statistical indicators that help, too and there aren’t many surprises in this area. Heavier players of the same height get more rebounds, use more possessions, commit more fouls, and take fewer threes. Lighter players tend to be better shooters from the free-throw and three-point lines, get more steals and blocks, and dish out more assists.
Enough with the blah, blah, blah. Let’s look at some output. Take a look at who the model expects were the ten heaviest players from the past season.
Act Pred Player Ht Wt Wt Mamadou N’Diaye 7-6 290 307 Sim Bhullar 7-5 360 295 Omar Oraby 7-2 270 269 Boris Bojanovsky 7-3 235 267 Chad Posthumus 6-11 265 263 David Wishon 7-2 265 262 Michael Ojo 7-1 292 262 Jordan Bachynski 7-2 248 261 Przemek Karnowski 7-1 305 261 Ian Markolf 7-1 260 261
One thing that jumps out is that only two players exceed a predicted weight of 269. Maybe everyone should stop listing weights, because the guessing game is always going to be friendly to the larger gentlemen in the game. Actually, this isn’t a horrible aspect of the model. Only about one percent of the D-I population has a listed weight greater than 269 (38 of 3,497 players that participated in at least 10% of their team’s minutes). So it’s going to be impossible to guess at which players under seven feet would be in that exclusive group solely using statistics.
The fact is, heavy people rarely to play basketball at a high level. Those that succeed are either very tall, or just incredibly special. We’ll get to those cases in a second. It also doesn’t help this effort that programs may be less-than-truthful about a player’s weight when it exceeds 300.
Here are the ten statistically-lightest players in the country.
Act Pred Player Ht Wt Wt Jordan Wilson 5-7 155 155 Marcel Smith 5-6 150 156 Keon Johnson 5-7 160 157 Kevin Alter 5-6 147 158 Justin Robinson 5-8 165 159 Christavious Gill 5-8 165 160 Blake Provost 5-10 170 160 Raymond Taylor 5-6 145 160 Parker Wentz 5-9 160 161 Zavier Turner 5-9 170 161
There are a lot of names that play for low to mid-major programs on this list simply because you don’t see too many 5-6 to 5-9 guys playing at high-majors. The model does a better job guessing the weights of the lighter people because the weight of college basketball players is not normally distributed. The difference between the heaviest and seventh-heaviest player is 65 pounds, while there are 27 guys that are within 15 pounds of the lightest player (Richmond’s Kendall Anthony, at 140 pounds).
The more fun lists are the ones where the model missed. Here are the ten guys the model most overestimated.
Act Pred Player Ht Wt Wt Ben Moore 6-8 185 227 Raphiael Putney 6-9 185 225 Josh Mendenhall 6-8 180 218 Tate Stensgaard 6-8 190 227 Josh Sharp 6-7 185 222 Ty’Quan Biting 6-8 185 222 Marcus Holt 6-10 200 236 John Bohannon 6-11 210 246 Onochie Ochie 6-6 180 216 Corey Allen 6-1 147 183
This group is a mix of guys who were just incredibly light for their listed height - Allen, Mendenhall, and Putney were the lightest players in the country listed at 6-1, 6-8, and 6-9, respectively - and guys that “played heavy”. That description applies to guys that rebound well, shoot mostly twos, commit a bunch of fouls, and don’t get as many blocks or steals as you’d think.
Also, I’m guessing some of these guys truly do weigh more. John Bohannon, for one, was listed at 210 pounds for each of his four seasons at UTEP, despite the fact he was given an extra inch this season.
What about the guys “playing light”?
Act Pred Player Ht Wt Wt BeeJay Anya 6-9 325 215 Joshua Smith 6-10 350 246 J.J. Davenport 6-6 325 245 Sim Bhullar 7-5 360 295 Justin McBride 6-10 310 249 Kennedy Meeks 6-9 290 233 Davante Gardner 6-8 290 234 Ako Kaluna 6-7 276 223 Antonio Campbell 6-8 279 227 Steve Forbes 6-9 295 243
BeeJay Anya never shoots (he took 8.5% of N.C. State’s shots) and is a woeful rebounder for his size (7.1 OR%/13.5 DR%), but the man can reject some shots. He swatted roughly one-in-eight opponents’ two-point attempts. It’s a profile that is typical of the skinny shot-blocking specialist. The kind of guy that a coach wishes would add 25 pounds in the offseason.
If you’re listed at 300+ pounds, you’ll be a candidate for this list simply because the model thinks someone of that weight would be pretty pathetic playing basketball. Five of the six 300-pounders in the country represent the top five on the list. (Prezemek Karnowski ranks 18th.) But Anya’s the unique case in this group because he projects as lighter than the average 6-9 player while being listed as the heaviest 6-9 player in the land. Come on, Mark Gottfried, let’s give him the “290” designation in next year’s program. (And maybe list him as F-C instead of a pure forward?) He’s earned it.
Finally, here are some bonus scatterplots. It’s not too difficult to spot Anya in the second one.