PDA

View Full Version : Advanced Player Stats - The Season Edition



caduceus
03-29-2014, 02:17 AM
Here's the plus/minus and Roland ratings for the season. Roland rating per 40 minutes, measured over the course of a season is a very useful stat to evaluate a player's overall contribution to the team. It's calculated by taking the team's net points (team scoring minus opponent scoring) while the player is on the floor, minus the net points while the player is off the floor, then normalized to a rate of per 40 minutes to eliminate the variable of playing time. It measures both the offensive and defensive (and intangibles) correlation of a player's help to the team effort. The higher number, the better.

As you've probably seen, I've been posting the per game Roland ratings this year. Season ratings have much more statistical significance (i.e., less chance of random occurrence, one-off efforts, or hot/cold streaks). Players with more minutes played will have ratings with higher statistical significance (accuracy) than those with fewer minutes. Players with fewer minutes are more often playing in garbage time, so that should be taken into consideration as well. I can't vouch for total accuracy of the numbers, as some venues don't provide accurate player stint information (looking at you WVU and Pepperdine!), but over the course of a season, the number should be reasonably near true player quality. Please note that Roland ratings can only be used for comparison within a single team. You can't compare them to a player from another team (or season).

The sad part is that we get these numbers after the season is over, instead of while we are in the midst of the more exciting part of GU basketball. Nevertheless, it makes for interesting discussion. The following table shows for each player the season plus-minus, plus-minus per game and per 40-minutes, season Roland rating and per-40 minute Roland rating. The most useful stats (in bold) are plus-minus per 40, and Roland per 40.



Gonzaga Bulldogs - 2013-2014 Season Stats
Player GP Min Min/G +- +-/G +-/40 Roland Roland/40
K. Pangos 36 1240
34.4 406
11.3
13.1
404
13.0
D. Stockton 36 1013
28.1 308
8.6
12.2
208
8.2
P. Karnowski 36 919
25.5 275
7.6
12.0
142
6.1
S. Dower 34 913
26.9 264
7.8
11.6
162
7.1
G. Bell 30 854
28.5 241
8.0
11.3
164
7.7
K. Dranginis 36 850
23.6 230
6.4
10.8
52
2.4
D. Barham 36 639
17.8 222
6.2
13.9
51
3.2
G. Coleman 34 392
11.5 95
2.8
9.7
-192
-19.6
A. Nunez 24 178
7.4 53
2.2
11.9
-132
-29.7
R. Edwards 17 85
5.0 22
1.3
10.4
-136
-64.0
L. Meikle 15 57
3.8 -10
-0.7
-7.0
-231
-162.1


The notable findings I was able to glean from the table were:

1) Our starters were clearly the 5 guys that ought to be.

2) Mr. Pangos was easily the main piston driving the engine this season, followed by Stockton and then Bell.

3) Of the bigs, Dower was "da man". Karno was slightly behind (but I think he came on very strong toward the end of the season. Can't wait to see him next year).

4) Maybe a bit surprisingly, Barham was clearly the 6th man. He was efficient during his minutes, and not only for dropping threes. Perhaps due to either defense or rebounding, or something else. He really stepped up his game this year. Notably, his plus-minus in previous years was 39, 50 and -1. This year, 222?! WOW!

5) Dranginis played nearly as much as GBJ this year, and his plus-minus numbers are nearly that of the starters. He will be GREAT next year, I bet.

6) Coleman and Nunez are enigmas. They both had very good plus-minus numbers, but lousy Roland ratings all season. It suggests that while they played well on-court, the team overall played A LOT better when they were riding the pine. Maybe it's a defense issue. Maybe it's because these two just don't play enough with the starters. I'm not sure. The fact that they have very favorable numbers otherwise suggests that they are good individual players and will both be significant contributors to the team next year. Major upside on both of these guys. In particular, Nunez has really had very few minutes. He really could skyrocket next year.

7) Edwards has very favorable plus-minus numbers. Both he and Meikle have played well during non-garbage time, from what I've seen. Their lack of minutes (and presence of garbage time in their stats) likely don't reflect their quality of play.

I'm actually very pleasantly surprised at the numbers. If these numbers are any indication, I think next year is going to be a great season.


http://i.imgur.com/X3ZlMVn.gif

wnczagfan
03-29-2014, 04:48 AM
Thank, you, Caduceus!

gamagin
03-29-2014, 08:29 AM
Thanks, cad. Interesting. KP's numbers look head and shoulders above others. How much of that is due to mins played? If any ? The only mins/games played that didn't seem to affect his numbers was was GBj's. Or am I reading that correctly. In 1-2 KP played ave. 36 mins to DS's 28. Yet the Roland was nearl twice as high for KP. What specifically extended KP so far above the pack ? Thanks.

BTB
03-29-2014, 08:37 AM
I could be wrong, but doesn't this moreso just show who played the most? I understand how the math works, but on a winning team the players on the floor the most are going to have the largest +/- on court and the lowest +/- off court. Barham is the only one who's Roland is an outlier if you graphed Rolan vs. playing time.

Zagceo
03-29-2014, 10:26 AM
Thanks cad

Here's my 2 cents on coleman and nunez..........they both seem explosive on their first step's on offense but on defense they're the opposite. Slow to react and never really fought hard through screen action. Stockton and Bell should really be studied to show coleman and nunez how it's done. Maybe they don't like contact or they're to big to wedge themselves like DS and GB but they seemed to always get beat on screens.

caduceus
03-30-2014, 11:22 PM
Thanks, cad. Interesting. KP's numbers look head and shoulders above others. How much of that is due to mins played? If any ?

Minutes played is factored out of the calculation, and normalized to per-40-minutes, so none. If you compare the plus/minus (fifth) column to the Roland (eighth) column for each player, the difference between those two is the "off court plus/minus". That's the point differential for the team while the player was on the bench. For Pangos, it's +2. For Stockton, it's +100. Hence the difference.

The reason that the numbers seem to parallel playing time is related to effect, not cause. The players with high Roland Ratings (i.e. the "difference makers") will naturally get the most playing time (if they have a smart coach). Karno has a higher plus/minus (and playing time) than Dower, but the Roland Rating indicates that Sam was likely more helpful to the team than Karno over the course of the season.

One big point to remember, is that this is an indirect measure of player quality. It reflects how the team plays in relation to a player. Much like NumberCruncher's efficiency lineups, it doesn't tell you directly how good one individual player is. Over the course of a season, though, it's an excellent overall gauge of a player's value to the team (and takes into account those things that can't be measured easily, like defense and the value of rebounds, blocks and steals in relation to the score).


The only mins/games played that didn't seem to affect his numbers was was GBj's. Or am I reading that correctly. In 1-2 KP played ave. 36 mins to DS's 28. Yet the Roland was nearl twice as high for KP. What specifically extended KP so far above the pack?

Not sure what you mean re: GBJ's numbers. Minutes played are factored out.

Regarding your last question, Kevin's high Roland numbers are in part because the team just didn't score substantially over the opponent during the 200 minutes he was off-court (net +2 points vs. opponents for the whole season!). For Stocks off-court, it was +100 points; Karno, +133 points, and so on. Kevin was extremely valuable to the team, more than the usual statistics indicate.

ZagaZags
04-08-2014, 09:21 PM
Thanks Cad for all you do here. Awesome stuff.