Reply With Quote
Great work. I imagine you have to have a strong stomach for doing this statistical work with this team.
Have had this idea in mind for awhile, taking a look at the Spurs record (and point differential) in games where Wemby performs.
The Hypothesis: When Wemby plays well, the Spurs are likely to win. When he doesn't, the Spurs are likely to lose.
Methodology:
To do this, I am relying upon John Hollinger's GameScore model, which is computed as follows:
Game Score is an imperfect metric, but it useful in this case because it gives us a game-by-game, holistic "all-in-one" score for how a player performed individually, accounting for "negative plays" such as missed shots, fouls and turnovers. As you might imagine, minutes played and USG% play a big role in potential game score, because you can't rack up a high score if you don't have the opportunity to take shots, but high minutes and high usage are certainly not required to post a good score. For example, Charles Bassey posted a GameScore of 17.3 in 18 minutes and only 9 FGA on 12/21 versus POR. You are, however, highly unlikely to get into the upper levels of GameScore without being one of the team's top scorers, and you certainly won't do it regularly.
- GmSc - Game Score; the formula is PTS + 0.4 * FG - 0.7 * FGA - 0.4*(FTA - FT) + 0.7 * ORB + 0.3 * DRB + STL + 0.7 * AST + 0.7 * BLK - 0.4 * PF - TOV. Game Score was created by John Hollinger to give a rough measure of a player's productivity for a single game. The scale is similar to that of points scored, (40 is an outstanding performance, 10 is an average performance, etc.).
In that respect, GameScore can be insightful in certain applications for high USG% players.
I compared Wemby's GameScore against the Spurs game-by-game point differential to see if the degree to which Wemby plays well impacts how much the Spurs win or lose by. I also calculated the Spurs record for various levels of Wemby performance.
The Results:
As you can see, there is a pretty decent relationship between the degree to which Wemby performs well and how well we perform as a team (using the game's score differential as a proxy).
And unsurprisingly, the Spurs are 14-5 and when Wemby is "Very Good" or better, but only 5-13 when he's Bad or only Above Average.
Of course, everything should have to fall on Wemby's shoulders, and this tells us what we all already knew... the man needs help. The fact that we can rarely survive Wemby being anything short of Very Good in any given game should reinforce any opinion you may have that our team (outside of Wemby) isn't very good.
I also went ahead and ran the same numbers for our #2: Devin Vassell. The results here told me what my eyes could have already told me... Devin is frequently bad, he's rarely very good, and him being really good doesn't actually impact winning.
Thanks for attending another Sunday Stat Session. Thanks for reading.
Great work. I imagine you have to have a strong stomach for doing this statistical work with this team.
Thanks scott. Can you do that graph for every player on the roster please?
Make sure you haven't just finished a meal when you see most of them.
Thanks Scott. Yeah Vassell scored 20 per last year and they couldn’t win. I think he could roll off the bench and be more impactful. Idk
I might do it for a few guys like Sochan, Barnes, CP3 and Castle, but the full analysis won't really be relevant for a large portion of the roster.
CP3 (2-1), Barnes (3-0), Keldon (1-1), and Champ (1-1) are the only players other players who have broken a game score of 20 all year. Everyone else it might just be more useful for me to make a table of what % of their games they break the 10.0 cut line, and what's the W-L record above and below that cut line. I'll stick with core rotation players because you have to have USG and Playing Time for GameScore to be a relevant stat.
I might also look at what it looks like for SGA, Jokic, Giannis and Tatum to get an idea of how reliant those teams are on their star.
Nice read. The only players worth taking look as well are Sochan and Barnes. Castle and Paul are not impactful. Rest sucks.
It's but obvious that the team is over-reliant on Wemby's excellence. What is interesting to me is that Devin Vassell has had 17 above average performances in comparison to 11 bad ones. Which suggests that box score metrics like Game Score rate him rather okay.Have had this idea in mind for awhile, taking a look at the Spurs record (and point differential) in games where Wemby performs.
The Hypothesis: When Wemby plays well, the Spurs are likely to win. When he doesn't, the Spurs are likely to lose.
Methodology:
To do this, I am relying upon John Hollinger's GameScore model, which is computed as follows:
Game Score is an imperfect metric, but it useful in this case because it gives us a game-by-game, holistic "all-in-one" score for how a player performed individually, accounting for "negative plays" such as missed shots, fouls and turnovers. As you might imagine, minutes played and USG% play a big role in potential game score, because you can't rack up a high score if you don't have the opportunity to take shots, but high minutes and high usage are certainly not required to post a good score. For example, Charles Bassey posted a GameScore of 17.3 in 18 minutes and only 9 FGA on 12/21 versus POR. You are, however, highly unlikely to get into the upper levels of GameScore without being one of the team's top scorers, and you certainly won't do it regularly.
In that respect, GameScore can be insightful in certain applications for high USG% players.
I compared Wemby's GameScore against the Spurs game-by-game point differential to see if the degree to which Wemby plays well impacts how much the Spurs win or lose by. I also calculated the Spurs record for various levels of Wemby performance.
The Results:
As you can see, there is a pretty decent relationship between the degree to which Wemby performs well and how well we perform as a team (using the game's score differential as a proxy).
And unsurprisingly, the Spurs are 14-5 and when Wemby is "Very Good" or better, but only 5-13 when he's Bad or only Above Average.
Of course, everything should have to fall on Wemby's shoulders, and this tells us what we all already knew... the man needs help. The fact that we can rarely survive Wemby being anything short of Very Good in any given game should reinforce any opinion you may have that our team (outside of Wemby) isn't very good.
I also went ahead and ran the same numbers for our #2: Devin Vassell. The results here told me what my eyes could have already told me... Devin is frequently bad, he's rarely very good, and him being really good doesn't actually impact winning.
Thanks for attending another Sunday Stat Session. Thanks for reading.
If he isn't contributing that much to wins, it either means that Gamescore isn't capturing his performance (or the lack of it) adequately or his impact is muted because of how he is played and with what lineups he is part of.
I think it's a combination of both. Devin Vassell's defense is below par but he is showing effort in recent games and those translate into better box score stats (some stocks). And the lineups he is part of ..particularly the non-Wemby ones are so thoroughly bad that it completely negates whatever positive effects he brings.
If one looks at it carefully...the sure conclusion is that Keldon Johnson is a much bigger liability than any Spur right now relative to minutes/ salary. Vassell at least brings floor spreading via threes, some off ball defense and playmaking. Keldon is currently a black hole - No 3, No D. Replace him with an average bench player wing and Vassell's impact in Non-Wemby minutes will be much better, in my opinion.
Here are two tables, the first I wouldn't put too much stock in but it was requested so I built it.
This first one is every member of the 9 man rotation with their % of games below a GameScore of 10, and the team record when they get below 10, over 10 and over 20. I wouldn't put too much stock in this because GameScore isn't adjusted for minutes or USG or anything like that. So when Julian goes to the bench and only plays 15 minutes, it's unlikely he's going to get to a GameScore of 10 even if he has a good game. Beyond a team's top 2 or 3 players, I don't put too much into this at all.
This next one is far more interesting to me, these are the players I hear talked about as the top MVP candidates (SGA, Joker, Giannis, Tatum, with Wemby often coming in 5th... not Curry, I included him because I wanted another "mid" team that is heavily reliant upon a single star).
One thing you'll note... for guys like SGA, Joker and Giannis... they never get a Game Score below 10. Tatum's 3 games came in games that were all double digit wins (2) or losses (1). The cut line for Stars really is a GameScore of 20, not 10. And you can see that ever team faces a dropoff when they don't get a 20+ game from their star, but the good teams (OKC, DEN, BOS) are able to still win at a very good pace when their star puts in a ho-hum performance by their standards.
I'm glad I included Curry, because it's gives us the same story as Wemby. When he puts in a Very Good or better performance, the Warriors are good (like the Spurs). When he doesn't, the Warriors are bad (also like the Spurs).
Eventually Wemby will get to a place where he never has a below 10 game. Hopefully he also gets like Joker who has only had 3(!) sub 20 games. But, the Spurs need to be able to be better than a .278 team when Wemby has less than a "Very Good" night.
In respect to Devin, I think it's more that GameScore doesn't correlate to wins for Devin because it just isn't a good metric to use for anyone but primary stars in an analysis like this (and honestly I was torn to even include Devin's chart). I think Devin's % of Bad v Good games is relevant, but the team shouldn't really be depending on it's second star for wins or losses. Honestly, I think the "best" scenario for any team is that their other players W/L records relative to GameScore are about in line with their overall W/L record (which Devin's broadly is). Teams don't win or lose because one particular role player had a good game... you just hope that the sum of your role players are good enough to pick up the slack when your star has an off game.
Hope that makes sense, I'm not sure if I explained that well enough.
I didn't intend this to be a Devin slander thread.
Seems to me that we need to find a way to make it easier for Keldon, Champagnie, and Tre to score more so we can win more?Here are two tables, the first I wouldn't put too much stock in but it was requested so I built it.
This first one is every member of the 9 man rotation with their % of games below a GameScore of 10, and the team record when they get below 10, over 10 and over 20. I wouldn't put too much stock in this because GameScore isn't adjusted for minutes or USG or anything like that. So when Julian goes to the bench and only plays 15 minutes, it's unlikely he's going to get to a GameScore of 10 even if he has a good game. Beyond a team's top 2 or 3 players, I don't put too much into this at all.
This next one is far more interesting to me, these are the players I hear talked about as the top MVP candidates (SGA, Joker, Giannis, Tatum, with Wemby often coming in 5th... not Curry, I included him because I wanted another "mid" team that is heavily reliant upon a single star).
One thing you'll note... for guys like SGA, Joker and Giannis... they never get a Game Score below 10. Tatum's 3 games came in games that were all double digit wins (2) or losses (1). The cut line for Stars really is a GameScore of 20, not 10. And you can see that ever team faces a dropoff when they don't get a 20+ game from their star, but the good teams (OKC, DEN, BOS) are able to still win at a very good pace when their star puts in a ho-hum performance by their standards.
I'm glad I included Curry, because it's gives us the same story as Wemby. When he puts in a Very Good or better performance, the Warriors are good (like the Spurs). When he doesn't, the Warriors are bad (also like the Spurs).
Eventually Wemby will get to a place where he never has a below 10 game. Hopefully he also gets like Joker who has only had 3(!) sub 20 games. But, the Spurs need to be able to be better than a .278 team when Wemby has less than a "Very Good" night.
edit: ah i skipped past your paragraph before the table
Thanks Scott. Awesome stuff as usual.
I’ve been thinking about something along the same lines, but more around how the peripheral players perform when wemby plays well, and how wemby performs when these players play well. I’ve already jumped to the conclusion that the team swims or sinks depending on wemby, but wonder what is required for wemby to do well.
Of course these players may play well when wemby does well because wemby drew so much attention, or they may do poorly because the opposition puts too much attention on these other players and didn’t put enough attention on wemby, or they do well thus opening things up for wemby. Given the team will build on wemby, might as well find the fitting pieces and jettison the rest.
This is a good idea to look into in the future - thanks for that!
I have a feeling that Tre Jones will rate very highly by this metric, but it's a hunch and not actually based on data.
Zach LaVine has a game score 20 or higher in 16 games this season along with a 19.8 and a 19.3.
Lauri has a game score of 20 or higher in 8 games including three games above 25 game score.
Cool now do the residuals on the first plot and let us know what the other factors are.Have had this idea in mind for awhile, taking a look at the Spurs record (and point differential) in games where Wemby performs.
The Hypothesis: When Wemby plays well, the Spurs are likely to win. When he doesn't, the Spurs are likely to lose.
Methodology:
To do this, I am relying upon John Hollinger's GameScore model, which is computed as follows:
Game Score is an imperfect metric, but it useful in this case because it gives us a game-by-game, holistic "all-in-one" score for how a player performed individually, accounting for "negative plays" such as missed shots, fouls and turnovers. As you might imagine, minutes played and USG% play a big role in potential game score, because you can't rack up a high score if you don't have the opportunity to take shots, but high minutes and high usage are certainly not required to post a good score. For example, Charles Bassey posted a GameScore of 17.3 in 18 minutes and only 9 FGA on 12/21 versus POR. You are, however, highly unlikely to get into the upper levels of GameScore without being one of the team's top scorers, and you certainly won't do it regularly.
In that respect, GameScore can be insightful in certain applications for high USG% players.
I compared Wemby's GameScore against the Spurs game-by-game point differential to see if the degree to which Wemby plays well impacts how much the Spurs win or lose by. I also calculated the Spurs record for various levels of Wemby performance.
The Results:
As you can see, there is a pretty decent relationship between the degree to which Wemby performs well and how well we perform as a team (using the game's score differential as a proxy).
And unsurprisingly, the Spurs are 14-5 and when Wemby is "Very Good" or better, but only 5-13 when he's Bad or only Above Average.
Of course, everything should have to fall on Wemby's shoulders, and this tells us what we all already knew... the man needs help. The fact that we can rarely survive Wemby being anything short of Very Good in any given game should reinforce any opinion you may have that our team (outside of Wemby) isn't very good.
I also went ahead and ran the same numbers for our #2: Devin Vassell. The results here told me what my eyes could have already told me... Devin is frequently bad, he's rarely very good, and him being really good doesn't actually impact winning.
Thanks for attending another Sunday Stat Session. Thanks for reading.
Rereading this scott
correct me if im wrong
it’s when keldon is heavily involved in the offense AND does well is when the win/loss is high. Right?
It's not just offensive. GameScore is intended to be (though by no means perfect) a total "all-in-one" metric based on BoxScore. I posted the formula in the original post. Bassey has games where he has posted a high GameScore with only 9 FGA, because he pumped in rebounds and blocks.
The metric is harder to apply to guys getting limited minutes, because obviously they aren't getting as many opportunities so the GameScore needs to be scaled. The Scale I provided is really more applicable to starters in higher USG roles.
But basically... when Keldon plays well, the team usually does well. Compared to Devin, where whether or not he plays well doesn't really seem to correlate to when the team plays well or plays poorly. The next question is - does Keldon play well because the team is playing well, or does the team play well because Keldon is playing well? (My guess is it's the latter).
Thanks for reiterating it
so yeah, if the answer is the latter then he may be one of the keys to the future
Sorry, I meant to say the former
The problem is that Keldon doesn't play well that often, but GameScore isn't necessarily the best way to measure that for him.
I don't think he's one of the keys to the future personally. As I've shown in other posts, however, the big problem is that he and Vassell should never play together.
I think your final conclusion is wrong. Keldon is one of the few on this team that can reliably get to the bucket. When he's doing that well, it makes the Spurs - especially the bench unit - so much harder to stop. I think that's what the relationship is showing.
EDIT: Meant Keldon.
I actually don't think there is much of a relationship here to show anything, because the measure isn't very applicable to Keldon. Basically, we're just looking at the cutoff of a GameScore of 10 as whether a player had a good game or a bad game. This is a fairly rudimentary way of looking at it, but it is especially not applicable for guys who play bench roles and have limited minutes. The majority of Keldon's games have a GameScore of under 10, but I don't think you can definitively say that he was bad in those games. On the other hand, it's easy to say Wemby had a bad game when his GameScore is under 10, because of his role.
If we want to use GameScore as a proxy for guys like Keldon, I'd need to scale it appropriately, which is not something I'm prepared to do now and probably don't have a high degree of interest in digging into to do properly (just being honest about expectations here).
I wouldn't draw too much of any conclusions here, other than I'd say that Keldon is generally not the key driver in our success. That shouldn't be much of a controversial statement, since a team's 6th/7th man usually isn't the key driver in a team's success.
I like a player that goes hard to the basket. KJ does that. His problem is his tunnel vision when he’s driving. He seldom passes out to a teammate. As many times as he goes to the hoop but he has just 1 assist per game. Not sure how long it will take to fix that.
I kind of view the rest of this season as a referendum on Keldon vs Devin. I have another thread somewhere that basically highlights how these two can't share the floor together. If you look at two-man pairings with over 100 minutes, Keldon-Vassell is the 4th worst in NetRating at -18.5 (and unfortunately for Devin, his name litters this list... ouch). The two are just wholly incompatible with one another. I do believe Keldon can add value, but I don't think there is room for Keldon and Devin on the roster. Let the best man win the rest of the year, and trade the other one this offseason. Based on the last two nights, Keldon has shown a of a lot more heart that Dev.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
