APBRmetrics

SGreenwell · Joined: 12 Feb 2005 Posts: 76 Location: Rhode Island

Hey, this is a bit off-topic, but there's not many places that seem to be interested in this sort of stuff. Slowly, over the past few months, I've compiled PERs for players in the Atlantic 10 conference. I've used only A-10 games, which limits the sample size (16 games) but also eliminates a lot of noise in terms of different opponents. The results seem to pass the initial smell test, as players like Delonte West and Jameer Nelson are ranked very highly for the 2004 season.

If you want to take a look at it, surf on over to my incredibly crappy Geocities site. I'm still working on adding a lot of the data (namely, things for all the non-URI teams) but I did want some input from people with backgrounds in this sort of field. I'm already trying to incorporate feedback from a non-stats orientated board I frequent for college basketball.

Ben · Joined: 13 Jan 2005 Posts: 266 Location: Iowa City

I like it. I wish there was more of this out there for college teams.

My 2 cents on Gross PER - I don't think any number of minutes with a PER of (say) 5 is really worth anything (i.e. it's at or below replacement level). I would set some replacement level (RL) and then set Gross PER = (PER - RL) * Minutes.

Kevin Pelton · Posted: Thu Jul 14, 2005 10:25 am Post subject:

Dwayne Jones didn't rate in the A-10 top 20 (in PER; he does in Gross PER)? That's surprising.

SGreenwell · Joined: 12 Feb 2005 Posts: 76 Location: Rhode Island

"Dwayne Jones didn't rate in the A-10 top 20 (in PER; he does in Gross PER)? That's surprising."

It wasn't so surprising to me, because I had seen him play for a few years. He's not a great offensive player, and even most of his limited PER value is tied up in blocked shots and rebounding. I think he led the A-10 in rebound rate; I don't have my spreadsheet handy right now to look it up.

Whoever told him to declare must have been solely looking at the fact that he's 7 feet tall. He also seemed like a good defender, but the A-10 isn't stacked with talented big men, and even the few they have are limited by their coaches not playing them more than 20 to 25 minutes a game.

"My 2 cents on Gross PER - I don't think any number of minutes with a PER of (say) 5 is really worth anything (i.e. it's at or below replacement level). I would set some replacement level (RL) and then set Gross PER = (PER - RL) * Minutes."

I just whipped up gross PER as something quick to use for actual value. Ideally, I would want to do something similar to Baseball Prospectus' VORP and incorporate replacement levels, but I'm not sure what that would be in college basketball.

The gap between the very best and the average is much deeper, as the median PER score is around 12. There's a lot of guys scoring in the 7 to 10 range who play major minutes and rack up a decent amount of gross PER. If anyone has suggestions for a replacement level threshold, I'd love to hear it.

One thing I've noticed is that college coaches do a poor job of optimizing their best players' minutes. Off the top of my head, I think La Salle is the only team that player three or more of their starters 30+ MPG, and it's clear why from looking at his bench's PER.

Maybe the stamina is different, but having played against some URI players in pickup at the URI gym, I've seen them go 30 to 40 minutes without a break. The pace is obviously different, but they also didn't have TV timeouts or anything. My lurking suspicion is that college coaches could play their better players a few more minutes each game.

Kevin Pelton · Posted: Thu Jul 14, 2005 10:07 pm Post subject:

Sure, Jones' offense is limited at best, but he still shot 50% and led the league in both rebounds and blocks. At the NCAA level, at least, he seems to be a fair comparison for Ben Wallace. So I think his rating that low doesn't pass my smell test.

SGreenwell · Joined: 12 Feb 2005 Posts: 76 Location: Rhode Island

Actually, because of this post, I found a hitch in my 2005 spreadsheet that was downgrading rebounding, which hurt some big men. Jones was the biggest benefactor from the fix, jumping up from a PER around 15.5 to 17.45. Once I revise the top 20 standings, he'll be in them.

He lead the A-10 in blocks, but blocks accounted for only 5.55 percent of possessions. He blocked a shot once every ten minutes, but I'm not sure if that's impressive in a league with only 3 or 4 legitimate seven footers.

I still hesitate to call him a good NBA prospect. West and Nelson put up PERs in the 25s when they went to the NBA, and both seem to be boderline starters at this point. I haven't gone back as far as David West and Cutino Mobley yet because of lack of data; I need to find some media guides.

Even with the new adjustments, a C/PF like Kevin Steenberge rates higher with a 22.51 PER. Steenberge is 6'11" with better stats across the board, except in blocks and rebounds. Jones is 6'11" with better "athletic" stats (blocks, rebounds) which might transfer better to the NBA; I'm not sure.

badgerbucco · Joined: 18 Jul 2005 Posts: 3

I have done the same for the Big Ten for the past two seasons. I used all numbers including non-conference and tried to compensate by adjusting for strength of schedule. Interestingly, the SOS adjustment doesn't seem to affect the final PER that much. If anyone is interested in seeing the sheets, let me know at jblakely1@wi.rr.com.