Thursday, November 11, 2010

Team Stats for 2010-11

New readers, please check out the basics before you leave any comments.

Thanks to dougstats (and of course Andres) I have been able to compile advanced stats for every team in the league this season. Along with the Raptor individual player stats, I will continue to update this on a weekly basis. Because I'm having trouble getting Google gadgets to work, here is the non-sortable spreadsheet (spreadsheet here):

Here are my thoughts about the how the Raptors compare to the rest of the league so far:
  • The team is 9th in FGA, but 27th in AdjFG%. The Raptors need to take better shots
  • More specifically, better 3pt shots, as they are 29th in 3pt% - although their 2P% (24th) isn't so hot either
  • At least they are shooting better than the T-Wolves (2nd in FGA, 30th in AdjFG%)!
  • I'm happy to report that the Raptors are 13th in rebounding with 8.3 (per player per 48 minutes, or ppp48), just behind the league average of 8.4
  • Doing a good job of getting offensive rebounds (3rd overall) in particular...actually pretty weak at defensive rebounding (27th). On the offensive side, I'd attribute this to a combination of poor FG%, Reggie Evans, and Amir Johnson; on the defensive side, I'd attribute this to poor FG% defense
  • They are also 9th in steals...but unfortunately 7th in TOs, leaving them at 20th for Net Possessions
  • Interestingly - despite their reputation as "Phoenix North" - the Raps are only 28th in 3Pa/FGa
  • They also do a good job on free throws too, good enough for 8th in the league, 8th in FT%, and 12th in FTa/FGa
  • Assists: 24th, blocks: 27th, and fouls: 9th. None of these rankings are positive
  • Put it all together, and the Raptors are 25th in WP48 - 12th place in the Eastern Conference standings

I'd like to say a few things to the commenters on my recent posts: thank you for reading, and thank you for leaving some comments! Kudos to most of you for being respectful and adding to the discussions. Unfortunately, I'm not necessarily going to be able to respond to all the comments very promptly over the next few days, but that doesn't mean that I didn't read what you wrote.

Coming up next: a post on whether or not top 3 draft picks lead to NBA championships.

 - Devin


    1. Thanks for providing this data. It's also relevant to the earlier discussion about rebounds. Compare the best and worst teams in WP48, and note how small the difference is in Reb48. The top 8 teams are 8.7, the bottom teams are 8.2, a difference of half a rebound per game. In contrast, the difference in AdjFG% is about 4%, or about 1.5 points per game -- 3x as big.

      And Reb48 actually exaggerates the difference in rebounding ability quite a bit, because good teams have far more DRB opportunities than bad teams (they make more of their own shots, and force their opponents to miss, while bad teams do the reverse). And DRBs are much easier to secure. If you adjust for that, about half of the good teams' reb48 advantage disappears, so the players on good teams are really only grabbing about .25 rebounds per game more than players on bad teams. That's a trivial part of the huge difference between these teams.

    2. Kevin Love provided a dramatic example of diminishing returns on rebounds the other day with his 31 Reb game. According to WP, Love added 21 rebounds for Minn (compared to an average PF with the same minutes). But Minn only had 6.5 rebounds more than average. Where did the other 14 rebounds go? (The answer is that Love took most of these boards away from his teammates, not the Knicks).

    3. You're right - I don't fully know how to explain the fact that rebounding doesn't correlate very much with winning (might have something to do with the fact that I'm not a statistician and didn't create the metric). All I can keep saying is that rebounds are valuable, as they add extra possession. Maybe if I looked at possessions/game that would explain the differences between good and bad teams.

      But I would definitely disagree with your example of Kevin Love - of his 31 rebounds, 12 were offensive, 1 more than the entire Knicks team. He wasn't just stealing rebounds from teammates, he was also stealing from opponents. In any case, of the rebounds he "stole", he added to the team rebounding totals. The more rebounds the team gets the better, and the fact that he "stole" a bit from teammates simply means that he gets a little more credit for his rebounding, which I have no problem with. He was the one who went out there and got the rebounds - not someone else.

    4. Devin: possessions per team are the same in each game, by definition, so looking at possessions will change anything. The fact is that the best rebounding teams just don't win that much more often than the worst rebounding teams. And that's NOT because rebounds don't matter -- as you correctly say, rebounds are valuable. The reason is that the "best" rebounding teams just don't get many more rebounds than the "worst" rebounding teams. The difference in rebounding ability is just very small. And so rebounding can't explain success in the NBA except to a small degree.

      As for Love, Minn. would typically get 13 ORBs in that game (49 opportunities), and they got 17. If you want to say Love added 4 boards there, I have no problem with that as an approximation. My point is that WP credits Love with adding about 20 rebounds for his team, and that's simply not true. Or even close to being true. And this is just one example of why team rebound totals aren't that different: if one player grabs a lot of boards, it mostly means fewer rebounds for his teammates. If he really took them from opponents, then we would see much larger differences at the team level. Do you see the connection?

    5. I know possessions/team are the same for each team during any one game - I was thinking of something along the lines of 'possessions that lead to a shot/free throws per game'.

      It's not that WP credits Love with "adding about 20 rebounds for his team" - the number of rebounds Love grabbed helps to explain his team's rebounding production. His team grabbed x amount of rebounds, and he had 31 of them. Other players on the team may have rebounded less than they normally do, but so what? On that night, they simply didn't rebound. How else should we evaluate their productivity? We can only say they didn't rebound that game, and give Love the credit for the rebounds he got.

      You've hit on a common criticism of WP. My answer is that, as far as I am aware, the entire WP calculation - which includes rebounding - explains wins in the NBA. I have seen the results time and time again. Rebounding by itself will not do the job, because - by itself - it doesn't tell the whole story, as you have said. With my level of statistical expertise, that is the best answer I can offer.

      There is a difference between 'rebounding ability' and 'rebounding production'. WP is more of a descriptive metric than a predictive one, although I think that given more research, WP will do a better job of predicting.

    6. Yes, WP at the team level predicts wins very well. But that totally supports my argument. As you show above, rebounds account for only a very small part of good teams' WP advantage. If you accept WP as valid, then it shows clearly that rebounds are NOT a very important part of winning in the NBA.

      And let me rewrite your statement, but make it about Kobe and his scoring: "the number of points Bryant scored helps to explain his team's points scored. His team grabbed 93 points, and he had 31 of them. Other players on the team may have scored less than they normally do, but so what? On that night, they simply didn't score. How else should we evaluate their productivity? We can only say they didn't score that game, and give Kobe the credit for the points he got."

      This is the reasoning that you reject (correctly) when applied to scoring. But then why use the same flawed logic on rebounds? The only difference is there is no record of "rebound attempts" or "missed rebounds." But they do occur. And a vast body of statistical evidence proves that individual rebounds do not represent marginal additions to the team total.

    7. Obtaining just any old "Top 3 Draft Picks", in and of themselves, does not correlate with winning the NBA Championship, since there have been more NBA teams with "Top 3 Draft Picks" on their rosters that have failed to win the NBA Championship than there have been Championship-winning teams with Top 3 Draft Picks on their roster.

      This fact alone, however, does not mean that having "Top 3 Draft Picks" on a roster is not an important factor toward the winning of a league championship in the NBA.

      Instead of comparing the number of:

      i. Teams that have won the NBA Championship with 1 or more "Top 3 Draft Picks" on their roster;


      ii. Teams that have failed to win the NBA Championship with 1 or more "Top 3 Draft Picks" on their roster;

      what you should be comparing is the number of:

      A. Teams that have won the NBA Championship with 1 or more "Top 3 Draft Picks" on their roster;


      B. Teams that have won the NBA Championship WITHOUT having 1 or more "Top 3 Draft Picks" on their roster.


    8. khandor:

      I'm already on it using that line of thinking :)


      Again, I really think you are close to something, but it still doesn't feel right. We care about shooting efficiency because a missed shot is essentially a turnover; but a "missed rebound" or a "rebound attempt"? There is no negative associated with not securing a rebound, simply a missed opportunity to gain a positive. Missing a shot on the other hand is clearly negative. That's why the two stats are treated differently.

    9. khandor: I'm not looking at correlations here. In your example, what you want to do is ask: how many wins does a champion typically get from top 3 picks, vs. how many wins an average team gets from top 3 picks. That will tell you how important they are. And that's what I'm doing with rebounds: how many extra points/wins do playoff teams get from their rebounds? (The answer: not that many.) That doesn't mean they don't matter at all, or that championship teams can be lousy rebounders (which I assume is rare), it just means that they are a relatively small factor compared to shooting efficiency and defense.

      Devin: You say "There is no negative associated with not securing a rebound, simply a missed opportunity to gain a positive. Missing a shot on the other hand is clearly negative." But both have the exact same consequence: zero points for me and the other team gets the ball. The two are perfectly analogous. The reason we don't know "missed rebounds" is not because they don't matter, it's because there's no obvious way for scorers to assign responsibility. We know who took a FGA, we don't (usually) know who failed to get a rebound. We can determine statistically that high-rebound players take a lot of rebounds away from teammates -- this is undeniable in a mathematical sense -- but we can't say that player X missed a particular rebound.

    10. Guy,

      The whole concept of attributing a specific "number of wins produced", per se, to a particular player, derived from his individual performance according to simple [or complex] set of "game stats" is non-sensical, from a basketball standpoint. Therefore, any attempt to do the same for a sub-set of "Top 3 Draft Picks", exclusively, is also an invalid method of assessing how the game of basketball actually works.

      Individual players only have a role to play as they fit within the Team Concept - e.g. 5-on-5, 9-on-9, etc. - against a specific opponent.

      In the game of basketball, "averages" ... and, therefore, "average performances", as well ... are essentially meaningless, in spite of what different "stats gurus" might try to tell you about how the game actually works at the highest levels.

      PS. It's a game of Individual Match-ups, within the Team Concept; not a game of "averages".

    11. khandor:

      I'm sorry, but that's just silly. According to that logic, Kevin Garnett wasn't very good until he was traded to the Celtics; Kobe Bryant was an elite player from 1996 (his rookie season) until 2004, magically became a bad player (the Lakers went 34-48), and then magically became an elite player again starting in 2007. Kevin Love is a bad player right now because his team sucks. Sorry; I'm not buying that argument. Adam Morrison is a two-time NBA champion, but Karl Malone, John Stockton, and Charles Barkley "never won anything". I think we've all heard enough of that.

      Is is a game of individual matchups, or is it a team game? You can't have it both ways. Pick one.


      Sure, they end up with the same result, but they didn't start from the same place. If you miss a shot, you had possession and put it up for grabs. If you fail to secure a rebound, a possession was up for grabs and you failed to capitalize.

      If you're going to count up "missed rebounds", you'd have to charge everyone on the court with a missed rebound every time another player on the court grabbed a rebound...which is silly.

    12. Guy:

      I had an epiphany while doing the dishes. To answer your original question/point, scoring efficiency explain wins better than rebounds because...*drumroll*...scoring efficiency is the outcome of a possession. Even if your team manages to secure more possessions than your opponent (through fewer TOs and more rebounds), those extra possessions mean nothing if you can't capitalize on them.

      To give an extreme example: imagine a team that grabs every rebound and doesn't turn the ball over, but can't score any points at all. They can't turn their possessions into points, and so they will end up losing all of their games (a regular team still has the chance of scoring after the tip-off or at the beginning of a quarter, but they only have two chances to score in regulation. They'd eventually win after enough overtimes, though).

      Thanks for the conversation - you've further solidified my belief in WP.

    13. Devin: you seem to be suggesting that a rebound is good if my team gets it, but is not good for the other team if THEY get it. But that's obviously impossible. And anything good for my opponent is bad for me -- basketball is a zero sum game. So failing to get a rebound must be a negative, just like missing a shot.

      But the fact we can't measure (yet) "missed rebounds" is a relatively small factor compared to the fact we don't know whether a rebound would have been grabbed by a teammate. That's the big analytic challenge. We know on average that each extra rebound recorded by a player adds about 1/3 of a rebound for a team (higher for OReb, lower for DReb). That's why the difference in rebounds among teams is generally so small. But that's only an average -- for some players it's likely higher, others lower. It would obviously be great to find a way to more specifically measure the individual contribution of players. But until then, crediting players with a fraction of a rebound will be much, much more accurate than pretending there's a 1-to-1 payoff which we know isn't real.

    14. Guy:

      I don't see where/how I said that rebounds are only good for my team. Anything good for your opponent is bad for you, yes; but it's simply not fair to call "missing" a rebound a negative.

      An example: you and I are in some service industry, competing against each other for sales. Every dollar earned is positive, and every dollar I earn is a dollar you can't earn, and vice versa. Should we assign you a "negative score" every time I earn a dollar that you don't? No. I earn money (points) by making sales (grabbing rebounds), but my success does not take away from the success you have already achieved. Every rebound I grab is a rebound you didn't grab, but it doesn't take away from the rebounds you have already secured. Punishing you for failing to grab that rebound would be punishing you twice; awarding me for grabbing the rebound is punishment enough.

    15. Devin: Basketball is zero-sum, service industries are not. I may boost my sales without impacting yours at all. That can't happen in basketball: either I grab the rebound or you do -- there is no third option. And if my opponent getting a rebound on a missed shot is good for him -- which you agree is true -- it must also be bad for me. But again, this really isn't worth arguing about. Let's focus on analysis and facts, and not worry about subjective moral categories like "hurting the team" that are just a distraction.

      The question I'm raising is simply how important are rebounds for winning in the NBA. Not "how valuable could they theoretically be?" but how much do they explain real wins and losses. This isn't hard to answer. Look at the teams with high WP, and compare them to league average (or to bad teams, doesn't matter). What you find is that rebounds account for a small share of good teams' success. Ask Arturo to run some regressions if you want -- he can confirm this. And the reason rebounds don't account for many wins is simple: there just isn't that much difference in rebounding performance among NBA teams. They are mainly clustered very closely around the league average (c. 73% of DReb and 27% of OReb). Some are a bit higher on one or the other, though that often offset by being average or worse at other end of the court.

      When something matters in a sport, we see it among the best players/teams: great tennis players have high 1st serve %. Great baseball teams hit a lot of homeruns. Great NFL teams have a high passing yard efficiency. etc. etc. But good NBA teams get only a few more rebounds than the average team (less than one rebound last season). And that tells us rebounds just aren't that important most of the time.

    16. Guy:

      Yes, it hurts the team. But you have to look at it this way: two teams are playing a game and have 40 rebounds each. A shot goes up, and one team gets it. One team now has 41 rebounds; the other has 40. It doesn't matter if securing one rebound has a two rebound "swing". The team that grabbed the rebound is not taking away from any of the rebounds that the other team already had; all they are doing is taking away a hypothetical rebound. What you are doing is similar to what happens when multiple choice tests are graded with incorrect answers being penalized - it's not fair.

      I've seen Arturo's regressions on rebounds - by itself, rebounds didn't do a very good job of predicting anything, as you've noted. But WP was very, very high. And as I said, I figured out the obvious reason why: possessions - by themselves - are meaningless. A team that does a poor job of capitalizing on their possessions will need a substantial amount of extra possessions to make up for it, which is why we see the points/possession explains wins more than rebounding.

    17. This comment has been removed by a blog administrator.