PDA

View Full Version : Advanced Player Stats vs. San Diego (new and improved)



caduceus
12-30-2014, 01:50 AM
Not sure if I'm going to continue these yet, since after my last submission of post-game advanced stats there seemed to be some confusion (and negativity) about the usefulness of these numbers. As several NBA teams (and I'm sure college teams too) use these numbers in evaluating team play, I think there's some value to them. Roland and plus-minus do not tell you necessarily how well a player performed. They give indication about how the team performed vs. the opponent in relation to the players involved. If you reply that because you watched the game and thought these stats are bogus because so-and-so played well despite a negative RR, then you are not understanding the numbers. All players have good and bad moments. Some result in score changes, others don't. These stats supplement the box score stats by incorporating offense, defense, and all the other intangibles that result in the final score. They are not perfect measures (no stat in basketball is), and are only a very small snapshot of the season.

Since minutes make a big difference in how you interpret the Roland (and seem to be confusing some people), I've added a column to the stats -- RR/40. This is the Roland score standardized to 40 minutes of game play. The RR/40 suggests (but does not definitively indicate) the net point difference of the team vs. the opponent in relation to each player if they played all 40 minutes. In other words, an RR/40 of +5 would suggest that a player's involvement resulted in either the team netting 5 more points than the opponent per game, had they played the entire 40 minutes. Again, it's not a perfect benchmark (and neither are PPG, RPI or KenPom ratings), but it eliminates playing time from the calculation, giving a standardized rate of TEAM production per player. Take it for what you will, I'm just the messenger.


Comments and feedback about Zags' game play, stats, and methodology are always welcome. If you're simply going to rant about the statistical usefulness of these numbers, I've heard it before, and I understand the misgivings. There are plenty of other threads for you to enjoy.




On Court
Off Court
Player MIN RR RR/40 +/- + - +/- + -
K. Pangos 35 18 21 15 56 -41 -3 4 -7 G. Bell 29 20 28 16 44 -28 -4 16 -20 B. Wesley 30 12 16 12 49 -37 0 11 -11 P. Karnowski 27 -2 -3 5 34 -29 7 26 -19 K. Wiltjer 29 12 17 12 45 -33 0 15 -15 K. Dranginis 21 -10 -19 1 29 -28 11 31 -20 D. Sabonis 21 6 11 9 38 -29 3 22 -19

+ = 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 +/-
RR/40 = Roland Rating standardized to per 40 minutes

P.S. Few went 9 deep on the bench, but Nuņez and Melson had single digit minutes -- didn't include them.

ZagaZags
12-30-2014, 02:03 AM
Not sure if I'm going to continue these yet, since after my last submission of post-game advanced stats there seemed to be some confusion (and negativity) about the usefulness of these numbers. As several NBA teams (and I'm sure college teams too) use these numbers in evaluating team play, I think there's some value to them. Roland and plus-minus do not tell you necessarily how well a player performed. They give indication about how the team performed vs. the opponent in relation to the players involved. If you reply that because you watched the game and thought these stats are bogus because so-and-so played well despite a negative RR, then you are not understanding the numbers. All players have good and bad moments. Some result in score changes, others don't. These stats supplement the box score stats by incorporating offense, defense, and all the other intangibles that result in the final score. They are not perfect measures (no stat in basketball is), and are only a very small snapshot of the season.

Since minutes make a big difference in how you interpret the Roland (and seem to be confusing some people), I've added a column to the stats -- RR/40. This is the Roland score standardized to 40 minutes of game play. The RR/40 suggests (but does not definitively indicate) the net point difference of the team vs. the opponent in relation to each player if they played all 40 minutes. In other words, an RR/40 of +5 would suggest that a player's involvement resulted in either the team netting 5 more points than the opponent per game, had they played the entire 40 minutes. Again, it's not a perfect benchmark (and neither are PPG, RPI or KenPom ratings), but it eliminates playing time from the calculation, giving a standardized rate of TEAM production per player. Take it for what you will, I'm just the messenger.


Comments and feedback about Zags' game play, stats, and methodology are always welcome. If you're simply going to rant about the statistical usefulness of these numbers, I've heard it before, and I understand the misgivings. There are plenty of other threads for you to enjoy.




On Court
Off Court
Player MIN RR RR/40 +/- + - +/- + -
K. Pangos 35 18 21 15 56 -41 -3 4 -7 G. Bell 29 20 16 28 44 -28 -4 16 -20 B. Wesley 30 12 16 12 49 -37 0 11 -11 P. Karnowski 27 -2 -3 5 34 -29 7 26 -19 K. Wiltjer 29 12 17 12 45 -33 0 15 -15 K. Dranginis 21 -10 -19 1 29 -28 11 31 -20 D. Sabonis 21 6 11 9 38 -29 3 22 -19

+ = 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 +/-
RR/40 = Roland Rating standardized to per 40 minutes


95% of the GU board members love the advanced player stats. Thank you for posting them. :cheers:

http://25.media.tumblr.com/71785d53a76f7bd37913dc002eda0c60/tumblr_mn2x890NjM1qk6zdio1_400.gif

NumberCruncher
12-30-2014, 03:37 AM
Thanks, Cad, for all the good work. I like hard information. And opinions, too.

Martin Centre Mad Man
12-30-2014, 04:34 AM
http://m.quickmeme.com/img/9b/9b4c5146e6cd2feeaa2067cb08cd0987076663267b84f939f5 1e29e2b966bdd4.jpg

caduceus
12-30-2014, 05:01 AM
Sorry.... :(

This graph will explain everything:

http://3.bp.blogspot.com/-EEJ97PdVZHc/TV2TyOSzcRI/AAAAAAAB72Q/_h87oqrh_T8/s1600/funny_graphs_16.jpg

zagfan24
12-30-2014, 06:12 AM
I love the statistical breakdown. Definitely appreciate it and hope you keep it up.

zag67
12-30-2014, 07:39 AM
I agree and thanks for the work. I think that one game is hard to read but as you see multiples, then some trends may show up.

rijman
12-30-2014, 08:32 AM
These stats are interesting. Watching in person I thought Draino was having a very positive impact on the game for the Zags, these stats indicate otherwise. I may have had more of a positive impact on the Zags rooting from the stands, although I don't have any stats to support my conclusion.

zagitup
12-30-2014, 08:47 AM
Please keep it up Cad...many appreciate the work you do here. Don't let the snarky comments (e.g. "lying stats") discourage you from continuing on. Thx again!

JokerZag
12-30-2014, 09:02 AM
Love this view. Keep them coming. Any chance you can publish the BYU game stats ? It was such a dynamic game, curious what the +/- shows. thanks

webspinnre
12-30-2014, 09:15 AM
These stats are interesting. Watching in person I thought Draino was having a very positive impact on the game for the Zags, these stats indicate otherwise. I may have had more of a positive impact on the Zags rooting from the stands, although I don't have any stats to support my conclusion.

Remember that these numbers don't show causation, they show correlation, which may or may not point to causation.

In any case, while I may have been a bit critical of +/-, that doesn't mean it isn't interesting fodder for discussion and isn't worth posting on the boards.

amaronizag
12-30-2014, 09:31 AM
This is a totally legitimate thread, but as a stats guy, I can tell you that I rely on KenPom while the Roland Rating doesn't interest me in the least. To ANYONE that watched the game it becomes immediately apparent that the numbers are meaningless. If you just looked at the RR and drew conclusions about how the game came down without actually watching the game, your conclusions would be way off base because you would conclude that we were better off with Sabonis on the floor than Karno and Drano. It could also be concluded that it was poor defense in the paint by Karno that allowed all of the points to be scored when it was actually 3 point shots that kept SDU in the game. What does SDU's ability to hit threes have to do with whether Karno was in the game? Ask yourself how a shooting slump and 7 minute scoring draught are reflected in the RR and how that reflects on Karno, Draino, and Sabbonis. I contend that shooting drought is responsible for the RR numbers but in a totally invisible way that would lead one to erroneous conclusions about the game and how it was played IF you didn't watch the game. But I did watch the game so once again I am left wondering what the RR numbers are supposed to tell me because they don't reflect the reality of the game.

caduceus
12-30-2014, 09:43 AM
Love this view. Keep them coming. Any chance you can publish the BYU game stats ? It was such a dynamic game, curious what the +/- shows. thanks

http://guboards.spokesmanreview.com/showthread.php?51894-Advanced-Player-Stats-vs-BYU

rijman
12-30-2014, 10:18 AM
I do appreciate the stats and effort. I was going for funny in a previous post, may have come off snarky.

It would be interesting to see these stats from many games to get a feel for them. Using Dranginis as an example, I would be interested to see if his games with similar stats would show completely different RR and RR/40 results from game to game.

caduceus
12-30-2014, 10:47 AM
I thought Drangs played great. Karno too. Everybody had a positive +/-. Just because a player has a negative RR doesn't mean they played badly. It doesn't even mean that the team was outscored while they were on the court (that's plus-minus!). It often simply means that the team played well with the player on the court, and played (for whatever reason) to an even better degree when they subbed out.

What I find most interesting is GBJ's numbers in this game. He didn't score much, but I think his defense had a huge influence on the game, perhaps the biggest. That doesn't show up on the normal stat sheet.

RR/40 over multiple games or a season is a much more powerful stat, of course. I'll do that sometime, provided I have the numbers available. Still, I think it's fun to look at these after watching (or before re-watching) a game.

zagfan24
12-30-2014, 02:50 PM
What does SDU's ability to hit threes have to do with whether Karno was in the game? Ask yourself how a shooting slump and 7 minute scoring draught are reflected in the RR and how that reflects on Karno, Draino, and Sabbonis. I contend that shooting drought is responsible for the RR numbers but in a totally invisible way that would lead one to erroneous conclusions about the game and how it was played IF you didn't watch the game. But I did watch the game so once again I am left wondering what the RR numbers are supposed to tell me because they don't reflect the reality of the game.

Basketball is a complex sport. The eye test is meaningful, but IMO these statistics give us insight into what we miss while watching the game, whether due to bias, misconceptions, not seeing the whole floor, not understanding the complex offensive or defensive principles employed, etc.

To your example - a post player can have a huge impact on the opposing team's 3 pt shooting. How well he hedges on ball screens, how much help he requires from guards in the paint (thus leaving their man open), how well he gets down the floor in transition, how well he calls out screens even when he is away from the ball -- all of these and more would apply. IMO, this is the specific benefit of RR. Basketball is an incredibly interdependent game and this particular statistic helps "unpack" that a bit. It's not comprehensive but it helps shed light on the subtle ways any given player impacts the team when he is on the floor (or not).

webspinnre
12-30-2014, 03:16 PM
As cad just pointed out, them having negative numbers doesn't actually mean they played poorly, just that for whatever reason, the differential from when they were on the bench was higher than when they were on the floor.

caduceus
12-30-2014, 07:09 PM
Basketball is a complex sport. The eye test is meaningful, but IMO these statistics give us insight into what we miss while watching the game, whether due to bias, misconceptions, not seeing the whole floor, not understanding the complex offensive or defensive principles employed, etc.

To your example - a post player can have a huge impact on the opposing team's 3 pt shooting. How well he hedges on ball screens, how much help he requires from guards in the paint (thus leaving their man open), how well he gets down the floor in transition, how well he calls out screens even when he is away from the ball -- all of these and more would apply. IMO, this is the specific benefit of RR. Basketball is an incredibly interdependent game and this particular statistic helps "unpack" that a bit. It's not comprehensive but it helps shed light on the subtle ways any given player impacts the team when he is on the floor (or not).

Wow that was a perfect description. Couldn't have said it better myself. Thank you.

RenoZag
12-30-2014, 09:18 PM
RR/40 over multiple games or a season is a much more powerful stat, of course. I'll do that sometime, provided I have the numbers available. Still, I think it's fun to look at these after watching (or before re-watching) a game.

Are the player stats at a site like statsheet.com ( here's K. Pangos LINK (http://statsheet.com/mcb/players/player/gonzaga/kevin-pangos)) the starting point for your charts, cad, or do you have to break it down from the game video yourself ?

Is there a site that tracks the + / - in real time ?

Thanks in advance for your patience. Some of us are noobs.

caduceus
12-30-2014, 09:39 PM
Are the player stats at a site like statsheet.com ( here's K. Pangos LINK (http://statsheet.com/mcb/players/player/gonzaga/kevin-pangos)) the starting point for your charts, cad, or do you have to break it down from the game video yourself ?

Is there a site that tracks the + / - in real time ?

Thanks in advance for your patience. Some of us are noobs.

Statsheet has player stints in detail (usually a couple of hours after each game), as well as rudimentary Roland numbers. However, they're not implemented well, and I don't think their statisticians understand them well. They have national and conference ranks for Roland, and one very important thing about the Roland Rating is that since it's essentially a TEAM stat, you cannot compare players if they are on different teams. They also have a "Roland Avg." which appears to be a completely useless stat since it's not normalized to 40 minutes.

The other issue is that in the past, some games haven't shown up as far as player stints and plus-minus. I'm definitely not gonna sit down to a game and log player stats.

As far as I know, there is no real time plus-minus tracking sites.

NumberCruncher
12-31-2014, 05:07 PM
Some of their numbers were off. Here's the precise data.

http://i.imgur.com/IPGX37e.jpg

caduceus
12-31-2014, 05:49 PM
Some of their numbers were off. Here's the precise data.

Interesting, thanks very much. Like I said, I'm often questioning Statsheet's numbers (and I'm pretty much having to rely on them). I may have to re-evaluate doing this in the future if their plus-minus numbers are that far off. Dranginis' numbers look much more like the performance I saw, and Bell's is more realistic as well.

I'm curious, are these from official stats, or your own numbers? Anyway, awesome job.

NumberCruncher
12-31-2014, 10:24 PM
Interesting, thanks very much. Like I said, I'm often questioning Statsheet's numbers (and I'm pretty much having to rely on them). I may have to re-evaluate doing this in the future if their plus-minus numbers are that far off. Dranginis' numbers look much more like the performance I saw, and Bell's is more realistic as well.

I'm curious, are these from official stats, or your own numbers? Anyway, awesome job.

I start with the play-by-play stats, wherever I can find them. Then I run them through my program looking for errors, which I have to fix manually from video. Most road games have quite a few, while the games at K2 tend to be the most reliable. This game wasn't nearly as bad as some I've seen. Of course, I have the luxury of only doing the Zag stats (and I'm retired).

The official stats always have the score correct and seldom err on crediting the right players with scoring. Assists, steals, rebounds and blocks are usually pretty good. Minutes are rounded off quite crudely. It's not unusual for substitutions to be missed and then thrown in at a later time in the play-by-play. So my numbers are not official, nor are they perfect. But they are more precise.