Welcome Mr. Udoh!

Mr. Udoh going up for a block.

Ekpe Udoh got the first start of his career last night. The Warriors won. While Udoh has garnered quite a few fans with his defensive prowess, what he actually does on defense — aside from blocking shots — is difficult to determine from looking at his box score stats.

And while some appreciate his defense, others criticize his lack of rebounding productivity. This is true and can be measured objectively in several ways. We can look at his personal rebounding percentages: 7.6.% ORR and 12.0% DRR. These are not good at all for someone playing center. We can also look at how the team rebounds when he is on the floor vs. off. Perhaps, someone else picks up the slack, while Udoh is busy doing all that defense. Well, that turns out not to be the case. The net offensive rebounding rate (ORR) is -7.2%, and the net defensive rebounding rate (DRR) is -7.7% (data from 82games.com).

Ok, but rebounding is not everything. Remember the four factors? (If you don't, click on that link.) Besides rebounding, there is shooting efficiency, foul rate, and turnovers. I showed (and others have showed) that shooting efficiency (as measured by eFG%) far outweighs rebounding as a factor for winning. It seems to me as if many of Udoh's critics are unaware of this fact.

So, let me expand on the point with real numbers. I already gave the net rebounding figures for Udoh. The net offensive eFG% when Udoh is on the court is +2.6%. The net defensive eFG% is -4.9% (negative is better in this case, meaning the opponent has a lower shooting percentage when he is on). If you had read my article on the four factors (if you still haven't read it, despite my repeated begging — please do so now), then you would know that we need to take those percentages and plug them into the "winning formula" to determine whether the superior net shooting stats trump the inferior net rebounding rates. Of course, you don't have to do the math  yourself. I have already calculated that (just looking at the shooting and rebounding parts of the formula) the team would win approximately 17 more games during the course of a season, because the shooting stats are so much more important.

If you don't want to use the fancy formula, just look at the net point differential per 100 possessions when Udoh is on the floor vs. off: +5.8 PPP. Taking that and multiplying it by 2.54 (from the wins vs. point differential equation) gives +14.7 wins over the course of a season. The reason it's a bit lower is that I didn't account for turnovers or foul rate above. The point remains: When Udoh is on the floor, the team appears to be vastly improved.

Now, it's perfectly reasonable to argue that it's small sample size, yada, yada. And it's reasonable to argue that until now Udoh has played bench minutes (although with mostly a terrible bench). (But if you're going to make this argument, then you must apply it to his rebounding stats, too, right?) The bottom line is that, if I'm the coach, I look at what the team does when a certain unit is playing. What I want to do is put the best-performing unit on the floor. Box score stats don't really show Udoh's effect on the offense or defense, but his RAPM (+1.7) also suggest a very beneficial effect. EZPM confirms that his rebounding is not good (around -3.0), but also shows his counterpart defense (~ +3) is very, very good. Taken all together, the data (and our eyes!) suggest that the positive effects of Udoh's defense and overall game vastly outweigh his biggest weakness (rebounding). Until the data show me otherwise, I say, welcome Mr. Udoh. Glad to have you on board (no pun intended). Keep doing what you do, Udoh, and don't listen to the nay-sayers.

29 thoughts on “Welcome Mr. Udoh!”

  1. Udoh with Lee must be negative though I don't immediately know how negative. Udoh with Curry, Ellis and Radmanovic are very positive, while he is moderately positive with Wright and Williams. Put 5 man lineups together with either SF and they are both very positive, albeit in 22 minutes for the season in each case.

    I guess you got to experiment more with Udoh & Lee but I'd use Udoh with Rad more too and see if it continues to be good. That could affect off-season considerations with Rad. You can either go with pairs with good numbers or bad ones. I know my preference. The team has a bit more than 1000 minutes left to experiment with.

    Lee is pretty good with Rad too. Too bad few others are.

  2. Well, the more I look the more I see.

    The top 5 4 man sets for season +/- all have Lee, in contrast to Lee's overall negative +/-. I guess it might pay to emphasize lineup that are working and avoid dinking around with other variations. Only 1 of the 5 had both Curry and Ellis in the quad (though they might still be on the floor as the 5th man). 4 were Lee with Rad. None of the 5 with Udoh.

    Of Udoh's top 5 quads none had Lee and only 2 had both Curry & Ellis.

    Start who you want for politics or motivation but get to the good lineups.

    1. I totally agree. Smart messes around with so many lineup combos. Udoh's top lineup by minutes (which is now the starting lineup) is +0, which is actually good for this team.

      1. Do you want a change or are you willing to give more time?

        If for change, would you go with an offensively or defensively "biased" coach?

        1. I want a defensive-minded coach and GM. Not sure who that would be...any college coaches fit that description that might be looking to move up?

        2. I am not really up on college coaches who might consider the NBA and might do well.

          Mike Brown or Dwayne Casey might be worth considering.
          Or Elston Turner.

  3. Lee only had 1 positive player pair in NY last season (with a minor player- M. Landry). If one went by the boxscore alone they would be ignoring that information. At their own risk.

  4. The top 2 most used Warrior lineups are negative on both raw and Adjusted +/- and 6 of the top 8 as well. You can't win that way.

    It might be worth looking at moving averages to see if any of the bad ones were improving.

    Some would dismiss the whole enterprise as too small sample. But the samples are as small as they are as a result of the decision to use so many lineups and at least some of the samples could be bigger and more significant if you managed the time with that purpose in mind. Coaches tend to live in the tactical world of this game and what might work at this very moment. Probably too much so, but they are usually focused on their W-L for self preservation reasons. An organization trying to rise is probably wise to use as much of the time as possible to test the best guess lineups.

  5. (I guess one of those 6 lineups I called negative on both raw ad Adjusted was actually even on raw +/-.)

    Wonder if new ownership has availed itself of additional analytic assistance yet or might soon.

  6. I read recently that GSW was one of 13 teams that don't have analytics departments. Although it is known that the new owner (Lacob) hired his son Kirk to be "the stats guy".

    I certainly wouldn't turn down a chance to give them my opinions/data.

    1. I guess I heard something about that but then forgot.

      I think it would probably be good to have several folks working the numbers and talking about them with each other and the rest of the front office and the coaches.

  7. Fantastic write up this perfectly sums up my own observations about rebounding vs. defense. A lot of people try to claim that a rebound is equivalent to a possession and therefore more important than defense, which if you think about it is ridiculous. A rebound does not equate to a possession, it CAN equate to a possession but not always. A rebound is counted even when a player is taking away a rebound from another player on his own team. Often times rebounds are not contested by anyone and it falls directly into the players hand. This is why if we are going to count rebounding as a statistic we have to differentiate between the type of rebound that is being occurring. Likewise if horrid defense is played there will be less rebounding opportunities. This is one of the reasons why the metrics centered around standard statistics like PER and Win Shares while useful in some regards are not indicative of a players actual impact in its entirety. Players like Udoh will be undervalued because of their lack of statistical output yet other players will be overrated. I really like what you have been doing on this Blog and I think basketball statistics are coming a long way due to work done by people like you. Thank you for the very interesting article. If I was Lacob I would actually hire someone like you to do this work that you are currently doing for free for the Warriors I think it could improve the team mightily.

  8. Great writeup, EvanZ, and I buy your argument that Udoh's defense helps in ways that the box score doesn't capture.

    What I don't buy is the argument that you hint at without outright saying: that Udoh's *offense* helps in ways that the box score doesn't capture. Udoh's been better on the offensive glass than on the defensive glass, but he hasn't been good, and everything else about his offense -- low usage, terrible scoring efficiency, high turnover rate, low assist rate -- is downright poisonous. I can't think of a single NBA player with offensive production that poor who's had a beneficial effect on his teams' offenses over the long or even medium term.

    In other words, I don't buy the idea that the team's improved eFG% with Udoh on the court is anything besides random noise, and I think that over time, Udoh will hurt the Warriors' offense, not help it. Now, it's very possible that there are factors I'm not considering. But if those factors exist, what are they? And do you think that Udoh will continue to be an offensive plus?

    1. Owen, that's a fair point about the offense. If, over time, he doesn't improve his offensive contribution *and* becomes a detriment to the offense (and we can't find someone better on offense), then he will need to justify his presence based on defense alone. Of course, we could have said the same thing about Biedrins, right? I think that Udoh's offense will improve as he gets more minutes and goes through training camp, but even if that weren't true, it is certainly possible that his defense would keep him in there.

      BTW, here's a quick numerical rule of thumb that comes out of the regression: Roughly speaking, for every 5%-point decrease in rebounding percentage, there has to be about a 1%-point increase in eFG% (either on offense or defense) to offset. Regardless of what you believe about Udoh, I would just hope that people take away from this article that *team* shooting efficiency is much more important than rebounding, when we are talking about who should play. If we ever were to get a low post presence that could score with his back to the basket, he better damn well play some D, and not just be a one-dimensional scorer/rebounder (which is kind of how I see Love developing).

      1. Agreed on all counts. I'm not contending that Udoh's lack of offensive contributions make him useless... his defense could still make him a net plus (and, as you say, more of a plus than the current version of Andris). I just don't think he'll continue to post good results on the offensive side. Could be wrong.

        1. "I just don’t think he’ll continue to post good results on the offensive side."

          I assume you mean you think *the team* won't continue to post good results as a result of his offense...considering we agree his own offensive stats are not good even now.

          He certainly doesn't appear to be the next Nene, but given that he is relatively cheap, and that (I think) we both desire the Warriors to focus on re-building, he may be the best guy to put at center right now, regardless of how he plays. In fact, maybe the strategy of putting the best player in right now is counter-productive, if the objective is to lose more games next year. ;)

  9. It would be possible to check how often the team’s eFG% improved over the its average with Udoh on the court game to game. If it is high % (say 70% or more) then it is more likely not noise than if it is a lower % (closer to 50% or less).

  10. Udoh's impact early in a still small amount of minutes may well have been affected by a share of loosely defined "garbage time". If the team’s eFG% improved with him as a starter and stayed that way at the end of the season that would be pretty promising for a rookie.

      1. He has done rebounding... with offensive and defensive rebs together... but not the others yet.

        Would be nice to have that fresh single Factor data to supercede, compare or blend with Joe Sill's run for 2008-9.

        Ultimately a multi-season version would be desirable / trustworthy.

        1. Make that "more" trustworthy.

          Yeah I was rooting around for ways to check further into impact claims short of the Adjusted Factors but I agree it would be better to have the Adjusted Factors.

  11. Hitting 67% of all his inside shots is markedly better than other Warrior big man. Hitting 67% FG on all his shots when at PF is pretty good too. Over 50% better than what he does at C. So position may matter for him and his impact. His time at C may be reducing the size of the impact, though noise may well be exaggerating the early difference. It is somewhat surprising that he has an overall positive impact on some team factors playing more than half his time at center when his at center estimated counterpart stats are so bad. But he does so far.

  12. Udoh-Biedrins has only been used 11 minutes so far. What would Udoh do playing with a real center instead of Lee or a better center than Biedrins?

  13. You can either conceive of a style of play, with goal team factor performance levels, and perhaps also types of players to use together to achieve those goals and do everything to fit "the plan", or you can assess who you have and their best blending and the best factor goals for those lineups. Or dance between them from off-season to in-season.

    Is Udoh more a piece that fits a specific chosen plan where he has a specific and proper role he is likely to succeed at or is he a piece in search of a plan?

Leave a Reply