PDA

View Full Version : Advanced Player Stats vs. BYU



caduceus
12-27-2014, 06:50 PM
Looks like StatSheet has gotten their act together and are finally tracking player-by-minute performance this year. Interesting numbers:



On Court
Off Court
Player MIN RR +/- + - +/- + -
K. Pangos 38 3 5 82 -77 2 5 -3 G. Bell 23 25 16 54 -38 -9 33 -42 B. Wesley 29 -23 -8 57 -65 15 30 -15 P. Karnowski 33 11 9 75 -66 -2 12 -14 K. Wiltjer 31 -9 -1 65 -66 8 22 -14 S. Melson 8 -21 -7 13 -20 14 74 -60 K. Dranginis 24 23 15 62 -47 -8 25 -33 D. Sabonis 14 5 6 30 -24 -1 57 -56


+ = Team points scored while on court
- = Opponent points allowed while on court
+/- = Team points scored minus Opp points allowed
Off Court = Team points scored/allowed while off court
RR (Roland Rating) = On Court +/- minus Off Court +/-



Bell and Drangs appeared to have helped the team the most. Seems that defense was a big key to this win.

Zag's RPI holding at #6 after a top-50 win (BYU still hanging in the 40's).
KenPom rating holding at #7 (#3 offense / #33 defense efficiency).


For BYU (Roland Ratings):
Kafusi: -21
Halford: -15
Haws: -13
Collinsworth: +21
Winder: +21

Malastein
12-27-2014, 08:37 PM
I don't take too much stock in these numbers on a game to game basis anymore. The problem is that they are indicative when runs occurred, which can entirely be based on who was hot at what point. Over the long haul, these numbers mean much more and would be troubling if a player always stayed in the negative but still played major minutes.

caduceus
12-27-2014, 10:04 PM
The problem is that they are indicative when runs occurred, which can entirely be based on who was hot at what point.

Not sure how that's a problem. That's how basketball works. Runs = good offense and/or bad opponent defense. Not runs = bad offense and/or good opponent defense. Runs + Not Runs = final score.

I agree that single game stats don't tell you who's a good player or not in a season -- that's pretty obvious. It does however tell you who was involved (or not) when the team played well (or not), and to what degree (points). It can tell you a lot about defense, teamwork and intangibles compared to the standard player stats (kinda like NumberCruncher's cool "which lineup is best" analysis). Statistically, not as strong for only a 40 minute game, but nonetheless interesting (for some of us).

thegloriousgoateeofKP
12-27-2014, 10:08 PM
Cad, do you have a running total of each player's RR for the season? That'd be a much more indicative sample size

caduceus
12-27-2014, 10:34 PM
2014-15 YTDPlayer MPG +/- Roland K Pangos 32 222 175 B Wesley 28 192 115 G Bell 26 193 117 K Wiltjer 26 167 65 P Karnowski 24 160 51 D Sabonis 20 141 13 K Dranginis 18 80 -109

Includes garbage time, so Drangs' (and maybe Sabonis') play is probably better than the numbers indicated. Didn't include the rest of the bench (too few minutes).

*Disclaimer: Numbers are only as good as StatSheet's record keeping.


Edit to add: Numbers look about what you would expect, I think.

Edit 2: T̶h̶e̶s̶e̶ ̶a̶r̶e̶ ̶s̶e̶a̶s̶o̶n̶ ̶R̶o̶l̶a̶n̶d̶ ̶t̶o̶t̶a̶l̶s̶.̶ ̶ ̶F̶o̶r̶ ̶a̶ ̶(̶s̶t̶a̶n̶d̶a̶r̶d̶i̶z̶e̶d̶)̶ ̶"̶R̶o̶l̶a̶n̶d̶ ̶R̶a̶t̶i̶n̶g̶"̶,̶ ̶I̶ ̶t̶h̶i̶n̶k̶ ̶y̶o̶u̶ ̶d̶i̶v̶i̶d̶e̶ ̶t̶h̶e̶ ̶R̶o̶l̶a̶n̶d̶ ̶b̶y̶ ̶t̶o̶t̶a̶l̶ ̶m̶i̶n̶u̶t̶e̶s̶ ̶p̶l̶a̶y̶e̶d̶,̶ ̶t̶h̶e̶n̶ ̶m̶u̶l̶t̶i̶p̶l̶y̶ ̶b̶y̶ ̶n̶u̶m̶b̶e̶r̶ ̶o̶f̶ ̶g̶a̶m̶e̶s̶ ̶p̶l̶a̶y̶e̶d̶.̶ ̶ ̶I̶'̶m̶ ̶t̶o̶o̶ ̶l̶a̶z̶y̶ ̶t̶o̶n̶i̶g̶h̶t̶.̶

Edit 3: I was wrong. Roland Rating is per 40 minutes for college basketball. So Roland Rating = Roland divided by total minutes played times 40.

webspinnre
12-28-2014, 12:59 AM
My biggest problem with +/- numbers and calling them "advanced statistics" is that they're so dependent on who all is on the floor with you (or currently on the bench), and who is on the court for the opponent, and who is on the bench. There are so many confounding factors present, that while they're indicative, I'd be cautious about saying how indicative. Are there any serious analytic/statistical studies done on the value of +/- numbers, even when done in a somewhat more useful manner as presented by RR?

thegloriousgoateeofKP
12-28-2014, 07:34 AM
Yes, good observation, webspinnre. I'm getting pretty deep into hockey analytics, where we do something similar: any shot attempt a player is on the ice for, regardless of whether he took the shot, is counted as a +1 if his team took it or a -1 if the opponent took it.

Recently, though, we've begun performing "score adjustments," accounts for the score in the game when an event takes place. Because a team that's trailing pushes the play and takes a lot more shots. I wonder if we could do something similar with these numbers to eliminate garbage time, and situations when the Zags are up by 15+, even if it's the first half

willandi
12-28-2014, 08:48 AM
Looks like StatSheet has gotten their act together and are finally tracking player-by-minute performance this year. Interesting numbers:



On Court
Off Court
Player MIN RR +/- + - +/- + -
K. Pangos 38 3 5 82 -77 2 5 -3 G. Bell 23 25 16 54 -38 -9 33 -42 B. Wesley 29 -23 -8 57 -65 15 30 -15 P. Karnowski 33 11 9 75 -66 -2 12 -14 K. Wiltjer 31 -9 -1 65 -66 8 22 -14 S. Melson 8 -21 -7 13 -20 14 74 -60 K. Dranginis 24 23 15 62 -47 -8 25 -33 D. Sabonis 14 5 6 30 -24 -1 57 -56


+ = Team points scored while on court
- = Opponent points allowed while on court
+/- = Team points scored minus Opp points allowed
Off Court = Team points scored/allowed while off court
RR (Roland Rating) = On Court +/- minus Off Court +/-



Bell and Drangs appeared to have helped the team the most. Seems that defense was a big key to this win.

Zag's RPI holding at #6 after a top-50 win (BYU still hanging in the 40's).
KenPom rating holding at #7 (#3 offense / #33 defense efficiency).


For BYU (Roland Ratings):
Kafusi: -21
Halford: -15
Haws: -13
Collinsworth: +21
Winder: +21

Have to agree with this! Wesley, Karno and Melson are THE reason BYU almost won!










NOT!!!!!!