Monday, January 19, 2009

My Rating System, and why it is the way it is.

Josh from the Excellent Big Ten Geeks blog asked me about my rating system formula, and I thought I'd write up a full explanation.

The non-secret Formula:
Points + Rebounds + Offensive Rebounds + [2 x(Steals+Blocks)] + (2.5 x Assists) - [Missed FGs + (2 x Turnovers) + (Missed FTs/2)]

Divide by game for per-game rating (hpg), and divide by individual possession for per-possession rating (hpps).

Explanation/non-defense thereof:
I started working the formula up back when I was following the NBA in the early '90's, sort of as a way to figure out who might be best in on Fantasy NBA teams. A columnist at USA Today, maybe, proposed some model for comparing players, and I kind of went with it from there. I didn't think of applying it to the college game until a few years ago, when I came across the Big Ten Wonk. He had so many things covered in tempo-free ways, it was inspiring but also intimidating. Effective field goal % and rebounding rates and so much more, as you all well know. However, I like things in a tidier package for comparison's sakes, tho. Kyle Whelliston at Basketball State has a comparable model, but it's pretty much a straight addition of the box score.

Here's the basics: I wanted to reward not only the ability to score points efficiently, but also the other aspects of the game that are so important but not as hyped. And with the Big Ten Wonk's focus on tempo-free comparison, I realized that the possession was the important thing I wanted to factor in. I remember Bob Knight saying about freshmen, that they didn't understand that when they turned it over, that was two points the team lost. And in general, in per-possession stats, that's true: most teams average around 1 point per possession on offense and give up close to the same on defense. But guys like Scottie Pippen, Ekpe Udoh, and Jamarcus Ellis never seemed to get the credit they deserved because their per-game scoring wasn't near other players while their rebounding, passing (well, not Udoh), and ability to get defensive stops went unrecognized. I thought a formula like this might help even it out.

So, obviously points were points, and steals were plus two points, and turnovers were minus two. Offensive rebounds secure a possession where a team shouldn't have one, so that's plus another two points. Blocks are little more difficult, because they aren't always a sure stop, but I figured someone who block shots probably also intimidates or changes shots enough to balance that out: plus two for them as well.
I especially wanted to reward assists. I grew up in Southern Indiana, so passing was always a premium, but it's always irritated me that assists are valued so lightly in other computations (the NBA and Whelliston are examples, but just about every sport puts assists as one point). Especially in basketball, where assists can be on three-point plays or shots, or what should be an assist can so easily turn into a foul on the other team and two more points, to boot. But I figured with a rough count that 30+% of all points come from free-throws and the additional point from three-point makes, so I rounded up to 2.5 points per assist. Really, it should be 2.3 or less, but I was using missed free-throws as minus half-a-point, so for ease, I figured it balanced out. Also, since it's been revealed that the home games result in higher assist counts for the home team, my system could be criticized for putting a premium on the most subjective measure in the box score. But, over 16+ conference games evenly split between home & road, I hope it balances out, too.

I count missed shots as minus one point because offensive rebounds do happen, and some teams used the missed shot as a feature of their offense (Michigan State under Izzo, as well as the Fab Five). And I count rebounds as only one point, as I felt that defensive rebounds are something that basically should happen. Someone else from your team will likely grab that board if you don't. And that's perhaps unfair. But I saw a lot at both the college and professional level of big men grabbing rebounds when noone was near them and snarling or screaming. Besides the aesthetically unpleasantness, I just didn't think that say, DJ White should get as much credit for a rebound on the defensive board -which was often snagged over a teammate (namely Ellis), - as opposed to the rarer and I think harder task of getting a offensive rebound. But, at the end of the season, I may try a tweaked version where I count Def. rebounds as plus two points and missed shots as minus two points. I could also count missed free-throws as minus a full point, also.

The one thing I'd like to do, but sort of flies against the intent of my system, is to reward three-point makes with an extra point due to their value to any offense. It's hard to operate an offense if you don't have shooters to keep the defense from packing it in. But it would require valuing threes as more than just three points in the system, but really as four points each. But it might be necessary in recognizing the value that a guy like Jason Bohannon has to his offense, and especially if I start counting defensive rebounds as two points.

Any (collegial) thoughts or suggestions are, as always, completely welcome.

Thursday, January 1, 2009

PreConference Player ratings for 2009

Obviously, a rating system is not so valuable when it's out of conference, and the strength of schedule is a variable. Also, a lot of players get a few minutes here or there and then see basically no time in conference play. Their efficiency ratings tend to vary wildly (reserve centers especially tend to post per-possession ratings north of .8000, with four doing so this year), so I cut 20 players off the bat for, well, for being walk-ons and receiving walk-on minutes. I also cut anyone averaging under 10 possessions a game, which threw out most of the outliers, including all of those centers. So, by the time I started putting together averages, 36 of the original 151 players that've seen action for Big Ten didn't make the cut. As a result of this selective inflation, and due to the historical tendency for efficiency averages to drop during Big Ten play, don't take these ratings as definitive... anything. But I do think they'll still be useful in spotting some trends overall.

Explanation of the system is here and this PreConference spreadsheet full report is here.

Average HPPG: 11.4
Mode HPPG: 9.0

Average HPPS: .3239
Mode HPPS: .2918

Top Ten Per-Game performers
Manny Harris, Michigan (33.1)
Talor Battle, PSU (31.7)
Damian Johnson, Minnesota (26.9)
Evan Turner, Ohio State (26.4)
DeShawn Sims, Michigan (26)
Al Nolen, Minnesota (25.9)
Rob Hummel, Purdue (25.8)
Chester Frazier, Illinois (24.5)
Kevin Coble, Northwestern (23.2)
Kalin Lucas, Michigan State (23.1)

Top Ten Per-Possession performers
Manny Harris, Michigan (.6187)
Damian Johnson, Minnesota (.6079)
Talor Battle, PSU (.5363)
Rob Hummel, Purdue (.5193)
Al Nolen, Minnesota (.5167)
DeShawn Sims, Michigan (.5159)
Goran Suton, Michigan State (.5066)
Dominique Keller, Illinois (.5051)
Draymond Green, Michigan State (.4979)
JaJuan Johnson, Purdue (.4924)

Top 15 Freshmen
Colton Iverson (15.9/.4901)
JOhn Shurna (15.2/.4838)
Tom Pritchard (20.8/.4181)
Matt Gatens (17.0/.3898)
Lewis Jackson (11.6/.3134)
Draymond Green (6.6/.4979)
Ralph Sampson (14.4/.4741)
Delvon Roe (12.5/.3988)
Laval Lucas-Perry (12.3/.3772)
Korie Lucious (4.8/.3171)
BJ Mullens (11.4/.3723)
Anthony Tucker (13.2//.3622)
William Buford (9.7/.2832)
Verdell Jones (9.8/.2757)
Stu Douglass (9.0/.2734)

School reviews

Illinois
Definitely some surprises here. Dominique Keller (.5051) is tops on the team in per-possession efficiency, but with Mike Davis producing so well (20.9/.4498), I'm not shocked that Keller's PT is still low. But that'll increase in conference play, I bet. On the other end of the spectrum, Alex Legion only has a few games under his belt, but after a year of practicing with the team, I really expected a better start. Instead, Legion is second only to OSU's Walter Offutt as the worst per-possession performer (at .0423). Chester Frazier has been producing very
well (24.5/.4811), but once again, he's carrying a heavy load in minutes, although this year it seems to be justified.

Indiana
Uh, not much good news here, so let's start with the bad. Only one player is above average in per-possession efficiency, and the only player who falls in the average range is Kyle Taber (at .3194). There's not much to glean here in terms of who should be playing more, Crean's doing the best he can with a bad hand. But I would suggest play Tom Pritchard, Taber, and Verdell Jones until their legs fall off.
In the good news column, Pritchard's one the best freshmen in the conference (20.8/.4181), which should bring good returns in future seasons. Jones is only slightly-below average (.2757), and has shown the ability to make plays. Aesthetically, he bothers me with his poor decisions, but just keep Moore at point most of the time. Moore's okay for a freshman point guard, but that still compares badly outside of that paradigm. Devan Dumes had a recent resurgence, but unless he starts hitting threes consistently, Crean might as well give those minutes to anyone else, even Matt Roth, who's incredibly poor .1819 average is still above Dumes' .1732 rating.

Iowa
The Hawkeyes in general have been much better than I expected, and Jake Kelly & Jeff Peterson who were really poor in last season's efficiency standings, have been very good (.4533 & ..3556, respectively). Anthony Tucker has been a bit of pleasant surprise, until he got himself suspended (13.2/.3622). Fellow freshman Matt Gatens has been quite good as well (17.0/.3898), and Cyrus Tate has been as good as expected (15.9/.4781). However, interior play is a big concern, as Jarryd Cole (.1926) and Aaron Fuller(.2538) are pretty disappointing so far.

Michigan
I didn't expect Coach Beilien to revive this program so fast, especially after last season's efficiency stud, Ekpe Udoh, transferred in the offseason. But Michigan's small-ball offense and 1-3-1 defense have been effective this year. Obviously the key to all of this is Manny Harris' player-of-the-year level performance (33.1/.6187). He's been very good at scoring in every way, as well as cutting down on turnovers while increasing assists and not to mention rebounding like a power forward. In looking for more PT, CJ Lee has a case with his low-usage, above-average play (.3669), but it's unlikely to happen if Laval Lucas-Perry (12.3/.3772) keeps shooting lights-out from three.

Michigan State
Well, these ratings might give Spartan Weblog's kj some relief. Kalin Lucas, while not the best point guard, is the best per-game performer (23.1) on the MSU team due to his incredible 7:1 assist/to ratio. Lucas' poor-shooting keeps him out of the top 5, tho, but just one more made three per game would put him with the conference's elite. A quick glance down the Sparties' assist-turnover stats shows Lucas' value to the team, as only Travis Walton has a positive ratio among the major minute earners. But Walton still turns it over too much for a senior combo guard. Goran Suton's
return is helping the offense tremendously(20.2), and he's already MSU's best per-possession performer(.5066), and with Lucas and #11 Raymar Morgan (21.0/.4579), gives the Sparties three top-15 per-game producers. The per-possession news isn't as great, but it's still pretty good, and in particular reserve Draymond Green (.4979) is looking to be featured a little more, as his per-possession rating is second only to Suton's on the team. MSU looks good for a conference title, I think.

Minnesota
The center position is paying dividends for Tubby Smith, as his three of his top five per-possession performers are freshmen Colton Iverson(.4901) and Ralph Sampson (.4741) followed by senior reserve Jonathan Williams(.4589). Lawrence Westbrook (.3405) & Blake Hoffarber (.3246) are slightly above average, but outside of the outstanding Damian Johnson (.6079) & Al Nolen(.5167), there actually isn't much efficient depth here on a per-possession basis. Former Juco POY Devron Bostick has been somewhat disappointing (.2853), serving more as a role player than the expected impact wing. Honestly, former walk-on Travis Busch has been as good as a non-center bench option as coach Smith has (at .3259).

Northwestern
The big turnaround for NU is no mystery. Freshmen John Shurna, Kyle Rowley, Luka Mirkovic, and David Curletti have provided quality inside play to complement to last year's starters Craig Moore, Kevin Coble, Michael Thompson, and Ivan Peljusic. Jeff Ryan has been effective enough off the bench (.3306), far better than as a part-time starter last season, but the presence of Shurna upfront has allowed Coble to return to an effective combo-forward role (23.2/.4546). One note of concern, outside of Shurna's team-leading per-possession efficiency (at .4838), the other freshmen bigs are all well below average. This suggests that we shouldn't quite believe in NU making the NCAA tourney quite yet. Still, the future is looking much brighter for coach Carmody.

Ohio State
One of the major stories of the early season was OSU's resurgence and Evan Turner's incredible play. However, you'll note that Turner's production is quite good on a per-game basis, he's not in the top ten per-possession. His rating falls down to 17th there, which is still very good, but I think he's not going to be conference POY. Frosh center BJ Mullens hasn't been quite as good as expected, but he's been solid per-possession (.3723), and Dallas Lauderdale's incredible improvement (18.1/.4637) has kept Mullens on the bench a little more than I expected. However, per-possession, noone else is really even at an average efficiency level. Power Forward Nikola Kecman should be available for conference play and might add another shooter and rebounder for OSU. Matta's club in good shape for a NCAA bid, but the flat loss to WVU does call the resurgence into some question.

Penn State
Talor Battle has been incredible so far this year, 2nd in per-game production (31.7) and 3rd in per-possession efficiency (.5363). Stanley Pringle has also been quite good (20.8/.4465), but outside of Jamelle Cornley's slightly above-average efficiency (.3402), there isn't a lot of guys who should be playing more. Jeff Brooks is above-average (3294), but he can only take minutes from Cornley and Andrew Jones (.3125), who is the only other average player. And now Andrew Ott is looking to get some PT as well, but the early returns haven't bad, but not really great either. As always, the defense and strength of schedule are questions for the Nittany Lions going into conference play. But PSU has a shot at the post-season due to Battle's fantastic play.

Purdue
The Boilermakers are certainly one of the best teams in the conference, but outside of JaJuan Johnson's improvement (19.4/.4924) and freshman point Lewis Jackson (11.6/.3134), there isn't a lot of great news here. Sure, the guards are fine, Marcus Green is still great off the bench (possessing a very good .4161) and Robbie Hummel is still great(25.8/.5193). But Keaton Grant has regressed back a couple years, taking a significant step backward (.2196) from last season's performance, and E'Twuann Moore isn't looking like the all-conference guard he was last season (dropping to .3475). But Purdue usually seems to be better in conference play than
outside it, so look for most of these averages to improve soon.

Wisconsin
Jon Leuer is the big winner of the Badgers in per-possesion terms (.4792), but the core of Marcus Landry, Jon Krabbenhoft, and Trevon Hughes has been quite good as well. Keaton Nankivil has been a pleasant surprise as a starter at the center spot,
but his average performance (9.0/.3109) still has left the Badgers significantly worse off on interior than last season, when Brian Butch & Greg Stiemsma were the best 1-2 punch in the Big Ten. The freshmen class has been extremely disappointing so far, with only Jordan Taylor and Rob Wilson getting any significant time, and both of the them posting the fifth and fourth lowest per-possession ratings (.0593 & .0594 respectively). But coach Ryan always develops his players slowly, and is one of the best in the business, so Badger fans shouldn't worry about the future.