Thursday, December 1, 2011

Calvin, the MIAA, and Player Efficiency Ratings

I’m a sucker for one-number metrics. In basketball, I love to judge teams by RPI; in baseball I love to judge players by WAR. Fans often have reservations when it comes to describing the quality of play of an entire team or individual by a single number (surely you can’t measure and quantify every single contribution or intangible!), but, at the same time, we’re obsessed with (somewhat) ordered lists and rankings.

Top 25 polls, All-Americans, players of the week, regional rankings, MVP selections, all-tournament teams, and all-conference teams are part of our “normal” routines. Each of these require us to either measure and quantify (or simply ignore the “unquantifiable”). In selecting or debating these types of awards and rankings, we are forced to determine what, exactly, is important to us.

Single-number ranking systems don’t (or shouldn’t) claim to be perfect. Just like your favorite top-25 ranking will always have flaws, an RPI ranking system will have flaws. But unlike your favorite top-25 (or all-conference team or whatever), an RPI system (or player efficiency ratings, or whatever) will always apply the criteria evenly across the board. Again, not perfect, but consistent (that’s always the key!).

Anyhoo, going back to my love of WAR (wins above replacement) in baseball, I began to play around with similar numbers for basketball. I found John Hollinger’s NBA player efficiency rating (PER) a few years ago, and I’ve been toying with it to adapt it to the MIAA since. PER is a convoluted formula (so I won’t try to explain it here), but I think it does a pretty good job of organizing players in an expected way. For example, in each of the last two MIAA seasons, PER “correctly predicted” 10 of the 12 all-conference players (and it also did well in predicting first/second team).

I’ll show you what the numbers currently look like for the league, but first, here’s a rule-of-thumb guide to what the numbers mean.

PER Range
Approximate Role
25+
MVP Candidate
20-25
All-MIAA Candidate
15-20
Solid Starter
13-15
Fringe Starter
7-13
Off The Bench
0-7
Reserve

Aside: PER is set up so that a rating of 15.0 is equivalent to league average play. Since teams usually play a 10 man rotation (or there about), and since the ‘better’ players (generally) see more minutes than the ‘worse’ players do, league average comes out to mean something between a fringe starter and a solid starter. In other words, think of ‘average’ as ‘quite useful’. This also means that “below average” players also carry quite a bit of value. Many teams will have a “below average” player (or two) in their starting lineup, and pretty much all of the players coming off the bench should be considered “below average” as well. For example, last year’s rating for Tyler Kruis was 14.0. I know seven other MIAA teams that would have loved to have that type of below averageness in their rotation.

Here’s how Calvin’s team looks according to PER this season:

Player
PER
Tyler Kruis
28.3
Matt DeBoer
24.8
Mickey DeVries
17.9
Tom Snikkers
16.8
Mitch Vallie
14.8
David Rietema
14.6
Jordan Mast
11.9
Bryan Powell
11.7
Adam DeYoung
9.7
Brian Haverdink
6.3
Ryan Nadeau
5.2
Tyler Dykstra
-5.0
Nate Van Eck
-6.6
Brent Henry
-26.3

We need to be careful that we’re not taking these numbers to be ‘true talent’ indicators (especially at this point in the season. This is an indication of how each player HAS played, not how each player WILL play going forward.

Looks like Coach Vande Streek was very right in adding Matt DeBoer and Mickey DeVries to the starting lineup. So far they’ve been giving the team very solid production while they’re in the game.

On obvious limitation of this system is that it doesn’t judge quality of defense outside of steals and blocks (and defensive rebounding, if you want to count that). There are plenty of solid defenders that don’t get steals, and probably also plenty of poor defenders that gamble a lot and rack up a number of steals. So, if a player is an exceptional defender, he may (rightly) receive more playing time than his PER would otherwise suggest.

Tyler Kruis is playing (almost) off the charts. If he could sustain this level of play for the rest of the year, he’d be in the MIAA MVP conversation. Certainly he’d be an All-MIAA first team player. If you’re looking for some references from the recent past, John Mantel put up a 26.0 PER in his senior year (he was third in the league that year), Danny Rodts put up a 20.4 rating last year, and Will Bowser of Hope finished last season with a 25.4 rating (Michael McClary was the MVP with an outrageous 35.4 rating).

Here’s the MIAA’s current* top 15 players according to PER:

Player
Team
PER
Kolin Kazen
Albion
37.6
Nate Snuggerud
Hope
34.3
Peter Bunn
Hope
29.8
Joe Prepolec
Kalamazoo
29.4
Eric Lewis
Adrian
29.3
Tyler Kruis
Calvin
28.3
Sean Gallant
Adrian
27.6
Joe Wilson
Kalamazoo
26.9
Tim Pearcy
Trine
25.6
Ian Jackson
Trine
25.5
Matt DeBoer
Calvin
24.8
Grant Carey
Kalamazoo
24.7
Braden Knight
Trine
23.5
Neil Smith
Trine
23.1
David Krombeen
Hope
22.4

*Hope’s game against Western Michigan from Wednesday night hasn’t been added to their MIAA team page, and since that’s where I get my data from, it’s not included in this set yet.

Most of these ratings will come way down once we get a few more games under our belts. I found it quite interesting that there are more Kalamazoo and Trine players on the list than there are Albion, Calvin, and Hope players combined. Could be a function of the difficulty of each team’s schedule, could be an indication that the traditional powers might struggle a bit more than we anticipated, or it could just be randomness.

I’ll try to keep the list updated. Perhaps I’ll do it weekly as the season goes on.

 Want to stay up to date on the happenings at FFTMAG? Follow me on Twitter, “like” us on Facebook, and grab our RSS Feed. Need a ride to a game? Check out the Ride Board and post your needs.

No comments:

Post a Comment