Statistical Analysis. Humor. Knicks.

Wednesday, August 20, 2014

Trading David Lee for Kobe Bryant Straight-Up: Shrewd Sabermetrics or Laugh Test Flunkie?

In Basketball on Paper, Dean Oliver devoted an entire chapter to comparing the individual rating systems of several NBA analysts. He argued something that I, and most people who do informed analysis, subscribe to: Any system of statistical analysis cannot only be internally consistent, but must also pass the “laugh test.” A statistical model can be built elegantly and beautifully and pass many confidence intervals within its own logical parameters, but if it’s results are absurd, then there’s obviously a need to return to the proverbial drawing board. Oliver thought of the “laugh test” as a litmus. It’s a very broad, absolutely basic determinant of whether a statistic is logical or not. If your rating system projects the best players with the best numbers, then it’s probably onto something. On the other hand, if your rating system argues that Jerome James is a better center than vintage Shaquille O’Neal, then you better recheck your assumptions.

While no single computation can perfectly encompass the entire contribution of a basketball player, John Hollinger developed a system to sum up a player’s boxscore contribution and express them in one number. Player Efficiency Rating (PER) is a sophisticated equation that goes so far as to adjust for the yearly value of possession and the pace a team plays. In Hollinger’s analogy, PER serves as a way of considering players from different positions, allowing an “apples to oranges” comparison. But while PER is a handy little number, what it doesn’t do is convert statistical efficiency into actual wins. That’s where Dave Berri’s Wages of Win (WoW) steps in. WoW takes the same boxscore statistics that PER uses and converts it to a formula that measures how many wins a player produces. This metric can evaluate a player’s total contribution over the course of a season and break it down per minute. Like PER, WoW serves as a way to summarize a player’s contribution in one number.

Now, let’s ask PER who were the most productive basketball players on the planet this past season. PER picks these as its starting five:

1. Dwyane Wade SG 29.2
2. Dirk Nowitzki PF 27.9
3. Yao Ming C 26.7
4. Tim Duncan C 26.4
5. Kobe Bryant SG 26.3

Nothing to laugh at here. In fact, it’s a pretty amazing team. Wade is the best player, slightly ahead of Dirk, who is just a bit ahead of Ming, Duncan, and Bryant, who are in a dead heat for third best. If you were starting a basketball team and were given first pick at any player in the NBA you couldn’t go wrong by picking any of these five players. They’re the best of the best. Granted, PER isn’t intended to be the final word on basketball performance, but it is a good starting point for figuring out relative worth. Would you trade your 15 PER performer for a 29 PER man? Almost certainly. Of course you’d take into account team composition, need, age, defense, contract terms, but all else being equal, you’d be doing your team a service by having the greater PER over the lesser. And if the PER was almost twice greater, like say Dwyane Wade over Jamal Crawford, well, then there’s really no thinking involved. Of course you’d rather have Wade. It’s a no-brainer. In fact, by this measure, you’d rather have Wade than any single player on the Knicks current roster.

Now, WoW gets to pick its own top five. Note that in order to compare WoW to PER we’re using Wins Produced per 48 Minutes (WP/48), since these are both rate stats:

1. David Lee PF .403
2. Jason Kidd PG .403
3. Marcus Camby C .371
4. Shawn Marion F .370
5. Carlos Boozer PF .351

Look at that again. David Lee led the NBA in wins produced rate. Um…really. So according to this sophisticated, statistical model, the most productive professional basketball player on the planet is David Lee. The best. On. The. Planet. Let me say that being a die-hard Knicks fan, I will be the first to argue that Lee is an All-Star caliber forward. He’s cool, he’s great. He’s an out-of-the-box rebounding, ambidextrous-finishing, no-look passing, efficiency machine. He’s awesome! It’s just that, you know, he really doesn’t create much offense. He’s more of a great glue guy than a centerpiece. And that’s why he’s not exactly a superstar.

Now, I really love the guy. Don’t get me wrong. I wouldn’t trade our man for the world. Oh, wait. Yes. Yes, I would. I’d trade David Lee in a heartbeat. For Tim Duncan. Or Yao Ming. Or Dwyane Wade. Or Kobe Bryant. Or Dirk Nowitzki. Or Lebron James. Or Amare Stoudemire. Or…OK, you get the point. I’d trade him for at least a dozen players who aren’t just All-Stars, they’re legitimate championship-level franchise cornerstones. Yet, right there in plain black and white, Wages of Win’s assumptions fail Oliver’s “laugh test.” WoW argues that Lee is the best player in the entire league, and that’s ridiculous.

WoW makes a very big deal about bucking conventional wisdom. And sure enough, statistical analysts are the ones who’re supposed to be bucking said conventional wisdom. At the Wages of Wins Journal, Berri argues that “perceptions of performance in basketball do not match the player’s actual impact on wins” because “less than 15% of wins in the NBA are explained by payroll.” However payroll isn’t a good measuring stick of perception due to the complexities of a closed system like NBA free agency. There are a host of factors on why a player may be overpaid from the talent available to the desperation of the team involved. In other words conventional wisdom thinks Rashard Lewis is overpaid at $126M, too.

So although conventional wisdom has a tendency to be wrong in some areas, figuring out sport superstars is not one of its weaknesses. There usually is a consensus on the league’s best players from both statistical analysis and conventional wisdom. The cream of the crop in the NFL are Peyton Manning, LaDanian Tomlinson, and Larry Johnson whether you go by the numbers or eyes. In MLB it would be Albert Pujols, Ryan Howard, Manny Ramirez, David Ortiz, Alex Rodriguez, and Johan Santana. At the top of the ladder of player evaluation, conventional wisdom is pretty much dead on.

According to WoW, David Lee (.403) is a far more productive player than Kobe Bryant (.242). Since teams with more productive players win more games than other teams, then Lee is better for your basketball team than Bryant. But why stop there? The Knicks could trade Renaldo Balkman (.272) straight up for Dwyane Wade (.255) and lose productivity. That’s right. WoW is arguing that if a Lee for Kobe, and a Balkman for Wade trade went through, then the Knicks would be a worse team for it. They’re arguing that Bryant and Wade, at the cost of our two young, talented forwards will hurt the Knicks’ productivity. You’ve got to be kidding me.

As the Knicks GM, would I pull the trigger on a Lee for Bryant deal? Is there even a debate? Who wouldn’t? Oh, right, WoW wouldn’t. WoW doesn’t even think it’s close. We can all disagree on which player is the very best (or the most productive), but WoW’s results are “laughable.” Dave Berri has criticized PER in the past, but before people can begin to take WoW as seriously as a tool for evaluating player performance as PER, it’s obviously going to have to address what caused this terrible absurdity in its rating process.

51 comments on “Trading David Lee for Kobe Bryant Straight-Up: Shrewd Sabermetrics or Laugh Test Flunkie?

  1. Owen

    “the most productive professional basketball player on the planet is David Lee. The best. On. The. Planet.”

    Basketball reference offers a variety of Dean Oliver numbers for both David Lee and Kobe. The main statistic is PW%, or player winning percentage, which is expressed as ORtg14 / (ORtg14 + DRtg14). Lee scored a .931 in 06-07. Kobe Bryant scored a .686. So Dean Oliver is failing your laugh test also. His numbers also show David Lee was a much better basketball player last year than Kobe.

    The numbers you used are from NBA Babble numbers and aren’t fully correct. They are slightly different from Berri’s, and also are a composite of regular season and playoff numbers, which generally are much lower. The best post for assessing which players are the best is:

    http://www.wagesofwins.com/RookieStar.html

    You fail to consider sample size. Lee has played about 2800 minutes in his NBA career so far. Basically, one full season’s worth. He posted a .197 his first year, and a .378 last year. That’s enough to indicate that trading Lee for Kobe wouldn’t be a great idea, but it’s a bit early for a coronation. If he plays 2800 minutes next year at the same high level he showed this year it will be difficult to argue he is one of the best players in the NBA.

    “The Knicks could trade Renaldo Balkman (.272) straight up for Dwyane Wade (.255) and lose productivity. That?s right. WoW is arguing that…”

    Dwyane Wade through the first 52 games had posted a .318 WP48.

    http://www.wagesofwins.com/Miami52.html

    After returning from injury too early, he played very poorly, which affected his WOW rating dramatically. Renaldo Balkman posted a .252 in 1064 minutes, which is a very small sample. So basically, no “laugh test” problems here. Dwyane Wade when healthy is one of the best shooting guards in the NBA according to the WOW, and better than Renaldo Balkman.

    Re: Lineups – You have to take position into account when constructing a WOW lineup. But off of raw Wins Produced last year the starters and backups would be:

    Kidd/Nash
    Bryant/Ginobili
    Marion/James (Marion is rated as a combo forward)
    Garnett/Duncan
    Howard/Camby

    I think that’s a pretty reasonable list which should pass any laugh test.

    IMHO, your post is off base, and there is a lot to respond to. But I am going to stop there for the moment as I have other matters to attend to. I will post later with some thoughts on the WOW/PER debate, the laugh test, the salary cap and conventional wisdom, and try to respond to what you posted.

  2. retropkid

    These models often use sample sizes that are not large enough…Lee had one year, relatively low minutes within it.

    If the Knicks win while David sits in the stands nursing an injury, you’ll re-visit his impact.

    And since he didn’t help the Knicks even get to .500 — perhaps a factor should be attached lowering his (or any under .500 player) rating. Making a bad team be mediocre more often is easier than making a mediocre team excellent more often.

  3. jay g.

    I might be missing something here but, having watched a lot of David Lee’s knicks games, it does not sound ridiculous to say that he is the most productive player in the league while he is on the court. To me, that means that he does more things to help his team win when he’s playing than anyone else, which includes many things other than scoring (and which also includes NOT doing things that encourage his team to lose).

    Is that or is that not what WOW is “measuring?” Obviously the sample size is a factor, along with things like usage rate (the Kobe vs. David Lee).

    I know that a criticism of these models in the past is that they “overrate” those who are efficient and/or use the ball less, while undervaluing those with high usage rate. Could that be the case here?

  4. jay g.

    I’ve had a second to think this over, and I don’t agree with the premise of your article Michael; nothing personal. PER is much less reliable than WOW and yet you left out the same arguement that you made against WOW.

    Based strictly on memory, John Hollinger’s PER rankings going into this past 2007 season had listed Chuck Hayes ahead of Dwight Howard on his power forward rankings.

    One might then propose the same “laugh test” to this scenario and say that no one in their right mind would trade Howard for Hayes, but the point is not to rank players by overall worth, but by per minute productivity, or efficiency, or some other measure that helps measure contribution. This example would not render Hollingers method completely inneffective, but would just serve to point out its flaws or shortcomings.

  5. Graham

    Isn’t it well-known (or widely argued) that WoW over-values defensive rebounds? The top 5 certainly doesn’t discourage that notion . . .

  6. Adam F

    Keeping in mind that I have only kept up with this blog for 2-3 months, I just want to point out a few things.

    Owen, you have been, as far as i recall, the biggest supporter of Berri who is a frequent poster. I agree 100% with your point, that we need to take sample size into account when viewing Berri’s conclusions. But I sometimes fear that when making your points, you fail to do that exact thing. For example, when arguing about our favorite topic – Eddy Curry, you consistently look at per minute statistics, which are generally a good tool, but as jon abbey i believe, pointed out “your world scoring 1 point in 1 minute is the same as scoring 35 points in 35 minutes.

    Obviously you don’t believe that, but when considering Eddy’s improvement, its important to look at the fact that he was able to MAINTAIN most of his solid offensive statistics while increasing minutes. Now, presumable when talking about sample size with David Lee, you believe that had he played more minutes he would fall more in line with the pack, and pass the “laugh test”. So the fact that Curry, played a larger sample size and maintained his statistics is a form of improvement.

    I really find a lot of what you have to say, Owen, fascinating, but I just think that sample size as you mentioned is crucial in evaluating stats and players, and that you should remain consistent in doing so.

  7. Mike K. (KnickerBlogger)

    Owen – Wins produced is a function of minutes played. But minutes played is a function of conventional wisdom (coaches). Shouldn’t we be comparing per-minute stats to eliminate any conventional wisdom creeping into those numbers?

    For instance if the Knicks played Balkman more than he would have more Wins Produced, and consequently rank higher than LeBron James.

  8. Mike K. (KnickerBlogger)

    Re: Chuck Hayes – yes PER had him higher than Dwight Howard last year, but this year Hayes’s PER dropped to 13.9 (below average). According to the Win Score stat page, Wow has Chuck Hayes 26th, above Kobe, LeBron, and Yao.

    It’s one thing if there is a statistical anomaly here & there, especially when we’re dealing with an incomplete stat record like NBA box scores. However to consistently rate sub-par players over better ones, fails the laugh test (the point of Mike Z’s article).

  9. Mike K. (KnickerBlogger)

    Adam F – I just want to point out that per-minute stats are highly reliable and much more consistent that per-game stats. It’s a great way to compare players who play different minutes, and evaluate which players deserve more (or less) playing time.

  10. Dylan

    Don’t foget that WoW also claims Dennis Rodman produced wins at a higher rate than Michael Jordan.

    http://dberri.wordpress.com/2006/12/04/on-jordan-and-rodman-again/

    Yes, I understand his argument that he does rank Jordan higher relative to others at his position. Nonetheless, the WoW metric claims that replacing Rodman with an average player would have hurt the Bulls more than replacing Jordan with an average player.

  11. Z

    To me it seems both systems are very good at proving what anybody who watches the games already knows. Even if no stats were kept and at the end of the game the only numbers that were written down were wins and losses people would know that Kobe, Dirk, Wade, Nash, LeBron, et al would be the best players and the ones to build a team around. At the same time, everyone would know that role players who do their job make winning possible and would appreciate guys like Lee.

    Like Adam, I am relatively new to this blog, and I?d never heard of either system before stumbling upon it. Stats make the game more fun for fans, and help the coaching staffs of teams to a degree, but in a world without them I feel like we would still know what we see with our own eyes to be true.

    I guess there are computer games where stats=something meaningful, but does either PER or WoW really teach us anything new? The only people who explore these numbers so deeply are the same people who watch the games religiously, either as a fan or as a professional in the league.

    I don?t want to devalue stats, or the time people spend looking at them. 6 months out of the year it is all fans have to go on, as they represent the memory of a season gone by. I just wonder if debating the two systems is ultimately worth making Owen really angry?

  12. Mike K. (KnickerBlogger)

    “in a world without them [stats] I feel like we would still know what we see with our own eyes to be true.

    I guess there are computer games where stats=something meaningful, but does either PER or WoW really teach us anything new?”

    Z – Stats do reveal a lot that the eyes don’t see on many levels from individual players (statheads knew Jermaine O’Neal and Michael Redd would become quality players long before NBA coaches/GMs did) to the team level (slow teams that are thought to have bad offenses due to their low per/g averages).

    “I just wonder if debating the two systems is ultimately worth making Owen really angry?”

    I want to make this clear: This article wasn’t published with the purpose to make anyone angry. I have to say, Owen is definitely one of my favorite commenters here and I have a lot of respect of Dave Berri who has had nothing but nice things to say about me and my blog. It’s certainly not the first article to criticize WoW, and I doubt it’ll be the last.

    However I view the NBA statistical field like any other evolving science. We’re all trying to make sense of what is happening, and without criticism the field would never advance.

  13. Ben

    Both PER and Wages of Wins have problems, both overvalue players in opposite ways. PER overvalues high usage players by rewarding players with higher usages regardless of efficiency, simply taking more shots will raise your PER even if they are inefficient shots.

    For example if I take an inefficient player like Jamal Crawford and have him shoot ten more shots a game his PER will increase even if his efficiency stays exactly the same. Since his efficiency is below average him taking more shots is most likely hurting his team but according to PER he is a better player simply by shooting more.

    On the other hand WOW tends to overvalue players who have a low usage. Unlike PER that directly rewards high usage, WOW does not directly reward low usage, instead it rewards it indirectly. In WOW players are punished by how many turnovers they have per 48 minutes. Usage is not taken into account so players who very rarely handle the ball like Diop are not hurt by their turnovers even though Diop turns the ball over at a rate of 22.4 which means he turns the ball over 22.4 times per 100 possessions in which he has the ball. On the other hand his teammate Dirk who according to WOW is more turnover prone than Diop turns the ball over only 8.3 times per 100 possessions. The reason Dirk averages more turnovers than Diop is not becasue he is more turnover prone but because he has the ball in his hands almost four times more often so has many more opportunities to turn over the ball.

    If Berri is going to look at points from a per shot perspective than it is only fair that he looks at turnovers from a per possession perspective. He should not look at one from an efficiency perspective and not the other.

    With that said, other than the turnover bias, as a pure efficiency rating WOW is a pretty solid stat and one could argue that David Lee was the most efficient player in the league last year. But by being low usage even with his incredibly high efficiency he does not help his team as often as a player like Kobe Bryant, so while each time David has the ball he is more useful than Kobe he has the ball many less times in a game so he has many less opportunities to help his team.

    With all that being said both stats are useful and while it is not always clear if high usage or low usage is better, most of the time all things being equal high usage is better than low usage but high usage should never excuse low efficiency.

  14. DS

    Lee will fair much better on the “laughability test” after next season. He will have worked on his dribble and mid range jumper with the same intensity as he did with his rebounding and FT shooting.

    If memory serves, Boozer looked like more of a “lunch pail” player in his first 2 seasons and then developed some offensive moves. I realize Lee being a 20/10 guy may be a stretch esp. with Z-Bo and Curry as teammates.

  15. jon abbey

    “Nonetheless, the WoW metric claims that replacing Rodman with an average player would have hurt the Bulls more than replacing Jordan with an average player.”

    talk about failing the laugh test, yowza.

    I love David Lee as much as anyone, but it’s pretty clear that Berri values rebounds too highly.

  16. Z

    “Stats do reveal a lot that the eyes don?t see on many levels from individual players (statheads knew Jermaine O?Neal and Michael Redd would become quality players long before NBA coaches/GMs did)”

    Okay, there may be a few cases, and without looking at Michael Redd’s stats, I’d say he is an excellent shooter without doing much else and makes max money on a team that is consistently bad (kind of like our own A. Houston not long ago). He seems overrated to a non-stathound.

    Certainly stats tell a story, and with only this post to go by, it seem that the two methods can work together to tell a better story.

    But we could see from watching Knick games this year the team was better with Lee on the court and Balkman showed that he has potential to help win a lot of games and that neither are likely to be superstars, but rather good role players. Don’t the stats say the same thing, and, if not, are they really telling an accurate story?

  17. Mike K. (KnickerBlogger)

    Z – We’re talking the 2002 Redd, that only averaged 11ppg.

    “Certainly stats tell a story, and with only this post to go by, it seem that the two methods can work together to tell a better story.”

    Exactly. Stats are hardly the answer in basketball on their own. Largely this is due to the stats kept (or rather the stats that aren’t kept). But there is a fair amount of information we know at the individual level (a statistically minded Blazers GM would have never traded Jermaine O’Neal who averaged 11 boards & nearly 3 blocks per-40 for AD) and even more at the team level (Detroit was a strong offensive team, despite being 21st in ppg last year) that we might not have known otherwise.

  18. Owen

    This was not the best day or week for me to do battle here. I am quite busy unfortunately, just when the opportunity to have a spirited knock down drag out WOW debate arises. Let me just throw out a few thoughts and responses.

    First, there seems to ba a lot of skepticism about box score data and what it tells us. I don’t know how to quantify it exactly, but I sense that for instance Mike would give basketball stats a 6 or 6.5 out of 10, in terms of their ability to evaluate players. I would give them about a 8.75 out of ten. They don’t capture everything, but they capture most of it. I dont think they are absolute, and I could pull back to an 8, but generally I think the data is pretty solid, even more so if garnished with insights from +/- methods.

    Instead of correcting some errors in Mike Z’s post, and suggesting that I think the WOW actually does pass the laugh test, which it does, what I should have said from the start was that I think the “laugh test” is sort of bizarre. Is that really a legitimate test of any kind in this case? It’s actually seems very weird to me. Usually, confirmation bias is a danger to be avoided, but here it is cited approvingly as if it should actually be the major goal of statistical analysis. That seems very backwards to me.

    Compared to the data used in many areas of life to make decisions, basketball box score data actually seems quite good. There are a variety of ways to slice and dice it to be sure (PER/WOW/Kubatko’s Winshares etc), some better than others, but basically, we have fairly good tools at our disposal. Not quite as good as a sabrmetrician perhaps, but still, good enough to make very well reasoned decisions about how good various players are at winning basketball games. And we have an added advantage in that basketball players are much more consistent than baseball players, and much less dependent on their teammates than football players. IMHO, a basketball statistician should be very happy with his subject. We can have an honest debate about limitations of the data, but it’s not iike we are navigating through fog here.

    Vis a vis Lee and Bryant, there are a few ways to look at it even with the WOW data in hand. But really, I feel entirely comfortable with saying that if Lee plays as well as he did last year, and as much as Bryant does next year, that he will be the better player.

    Mike Z – Here is my question. Allen Iverson had a PER of 26 in 2005-6. Andre Miller was at 16.4. They switched teams, and to all outward appearances, the Sixers were more successful with Miller than Iverson. At the very least, they were just as good. How do you explain this? Was this some sort of freak anomaly? Or perhaps does PER have an imperfect relationship to wins, as you would expect given that it doesn’t actually try to explain them?

    Dylan – Berri doesnt think Rodman is better than Jordan. He never said anything to that effect. In fact, the post you linked is devoted to combating that misconception.

    Retropkid – The Knicks were almost a .500 team until Lee went down. They probably would have been I bet if he had played 36 minutes per.

    Adam F – Re Curry, i dont know what to say that I haven’t said a million times before. He is the worst defending, worst rebounding, nearly worst passing, second most turnover committing center in the NBA. I grant that he matched his poor historical level of productivity this year in increased minutes, but any way you slice it he isn’t as good as Isaiah wants you to think he is. He scores well, and is worse in category in almost everything else. Unless you consider scoring the only skill that matters, which I suspect you do, you simply can’t say Curry is a dominant, or even good basketball player.

    Ben – I always like your comments. That was a good one. But I think we have hit an impasse with the rate thing. I understand your point. I guess my response would be that it is possible for Diop to be out there, not touch the ball very much, and still be extremely effective. That’s one of the quirks in usage that I sse. If you are 7 feet tall, it doesn’t really matter if you don’t touch the ball that much. You can still be incredibly valuable to your team. Every team needs both high usage and low usage players to win and be effective. And while the best players are often scorers, there is more parity between the two categories than generally supposed. A great role player produces more wins than a good scorer. And a team suffers just as much from having mediocre role players as it does from having mediocre scorers.

    Jon – You love David Lee as much as anyone? Not nearly as much as me.

    Alright. I am going to sign off. That was a bit of a hasty post, and really I am not all that satisfied with it, I can already hear it being picked to pieces but some sort of response was needed from the only WOW partisan on this board.

  19. Brian M

    “If you are 7 feet tall, it doesn?t really matter if you don?t touch the ball that much. You can still be incredibly valuable to your team.”

    Fair enough, but should a statistical evaluation system *reward* a player because he never touches the ball? It is not a bad thing to use possessions, after all, because *somebody* has to use a possession each time down the floor. But in effect WoW punishes players for using possessions which is nonsensical for the very reason that by definition a possession event must be attributed to a player each time down the court.

    Really it comes down to different ideas on usage. If there is some sense in which David Lee is better than Kobe Bryant, it is this– he makes more of his opportunities than Bryant does, in an absolute sense. But most people will not be comfortable leaving it at that, because there are clear differences in context and usage in particular. That is to say, there are differences in the *nature* of their respective opportunities.

    As a consequence, not all of the efficiency difference between Kobe and Lee can be chalked up to differences in quality of the individual players. Much of the difference is a function of different roles they play on their respective teams. For instance, if Lee and Kobe had to swap their roles in an offense and the defensive pressures they face respectively, Kobe would be the more efficient player in an absolute sense. And in fact the difference between Kobe role player and Lee superstar would most likely be bigger than the difference between Lee role player and Kobe superstar. Actually Kobe role player might still be lower efficiency than Lee role player, but Lee superstar would be in the basement compared to Kobe superstar.

    Another way to think about it would be that ideally, we would be able to weight the value of each player’s possessions used by the average degree of difficulty of his possessions used. Lee is more efficient but the degree of difficulty of his opportunities is far lower than Kobe’s. What is more is that Lee’s possessions are easier in large part because he is more selective than most players in what opportunities he takes, rather than because he creates those easier opportunities himself. And again, just as with possessions, it is the case that *someone* is going to have to use those more difficult possessions. That someone is going to be a guy like Kobe rather than a guy like Lee. A statistical evaluation system should take that into account somehow.

  20. Ben

    Owen – I actually like WOW and think it is just as valuable as PER but I think both are flawed and only show part of the picture.

    The point about the rate stats is; since Berri breaks down scoring into a rate stat looking at points per shot wouldn’t it make sense if he also broke down turnovers, rebounds and assists the same way. Why does he look at a total per 48 minutes for assists, rebounds and turnovers but at points per shot for scoring. At the very least shouldn’t he adjust rebounds, assists and turnovers for pace. I mean a player for the Warriors had more opportunities for turnovers, assists and rebounds per minute than a player for the Pistons. I am just saying to really look at efficiency, everything or nothing, should be broken into a per attempt/possession stat not some per attempt/possession and some per minute.

    Also I agree that Diop is valuable without touching the ball but wouldn’t he be a more valuable role player if the Mavs did not have to worry about him turning it over every time he touched it. If you adjusted his win score using rate stats for turnovers, assists and rebounds I still think it would be good it would just be more in line with his real value.

    I totally agree with your assessment of role players I think they are extremely valuable and would agree that a great role player is more valuable than a good scorer. I would have to think long and hard before trading Lee for anyone short of a top ten player in this league.

    Also to defend WOW you have to look at players by position, it is unfair to compair Lee to Kobe because SG’s have much lower win scores than power forwards, and Berri has said as much.

  21. DontLetsStart

    The laugh test seems like a ridiculous idea to me. What’s the point of having stats if you’re only going to accept the ones that give results you already agree with? Why not just rank the players off the top of your head?

    There are many valid reasons to criticise WoW, but the fact that it doesn’t put superstars at the top of its rankings isn’t one of them.

  22. Z

    “A statistically minded Blazers GM would have never traded Jermaine O?Neal who averaged 11 boards & nearly 3 blocks per-40 for [Dale Davis]”

    Stats don’t account for all aspects of a team game. Personalities, chemistry, and human relationships are as valuable/destructive to a team as defensive rebounds. I don’t know anything about the Blazers at the time of the JO trade but there are a host of reasons that that trade could have been made even with full knowledge of his statistical progression. Someone who only cared about stats and retained players destined to be good could easily be hurting a team (take Artest in Sac; Wallace in Det; Francis everywhere; etc…).

    “and even more at the team level (Detroit was a strong offensive team, despite being 21st in ppg last year) that we might not have known otherwise.”

    Why do we need to know this? If we watch the Pistons, we see they are a good team with players that can score the ball. They are not the Fratello Cavs of the mid-90′s.

    For players there is a lot of ambiguity surrounding their performance and value, but teams are judged by just 1 thing: wins. I guess well used stats can teach us why underachieving teams fail and can help make them better, but it’s probably pretty obvious to people who watch the games (the Fratello Cavs of the mid 90′s couldn’t score, etc…)

    It’s not that I don’t like stats. These discussions just make my head spin with numbers, rates, percentages, and things that remind me way too much of calculus (which is strange because I never made it even close to stepping into a calculus classroom, and can’t even tell you what calculus is (or if I’m even spelling it correctly…))– just to end up with the knowledge that Lee helped the Knicks win games and Curry hurts the Knicks at least as much as he helps them.

  23. Graham

    “The laugh test seems like a ridiculous idea to me. What?s the point of having stats if you?re only going to accept the ones that give results you already agree with? Why not just rank the players off the top of your head?”

    The point of the laugh test is so that we don’t just make up statistical measures off of the top of our heads and trust them blindly. The laugh test leaves plenty of room for the acceptance of counterintuitive results.

  24. DontLetsStart

    “The point of the laugh test is so that we don?t just make up statistical measures off of the top of our heads and trust them blindly. The laugh test leaves plenty of room for the acceptance of counterintuitive results.”

    But we already have a much better test for stats than that. Good stats should have predictive power. We should be able to use stats to predict the outcome of the upcoming season for example. The closer to accurate our predictions using a certain stat, the better that stat is.

    This is an area where I feel WoW falls down, if you read Berri’s blog he’s quick to claim credit when he’s right (eg the Iverson trade), but when he’s wrong (eg the Golden State/Indiana trade), he explains it away by saying players magically improved.

  25. MattinDC

    Regarding Predictive Power: Using PER and WoW together should give us some decent estimates about a player’s future performance, but it will never be PERFECT. Too many human factors come into play: team chemistry, becoming the focus of opposing defenses, and confidence can affect a player’s performance, and with it the sucess of a statistical formula’s predictive power.

    Can’t we just use both statistical evaluations and find a middle ground? I personally like PER.

  26. Mike K. (KnickerBlogger)

    I think this is exactly Dean’s point about the holy grail of basketball stats. According to WoW trading Wallace for Chandler was a downgrade for Chicago, and according to PER trading Iverson for Miller should have been an upgrade for Denver. But Chicago won more games, and Denver lost more games. The problem? Neither person fit the team’s needs. Chicago needed another defensive center that can’t carry an offense like Denver needed another inefficient scorer.

    The problem with assigning a single number to a player is that basketball is too complex for such simplicity. As Dean Oliver has often said you have to understand the team first. If a team has a need and you’re just adding to that need, then the law of diminishing returns comes in to play. They don’t have that in baseball. A 1.000 OPS guy will help any lineup. But in basketball a 25 PER guy or a .250 Win Shares guy may not make some teams better.

    To answer Owen’s assumption above, I would say stats explain 9 out of 10 on the team level (maybe more), but translating that to the individual level is a dicey proposition. For some players on some teams it could be a 9 out of 10. For other players it might only be a 1 out of 10. Basketball is almost like football in that sometimes it’s the RB (Barry Sanders), sometimes it’s the OL (Denver). In basketball it’s hard to distinguish the player from his teammates & the system they play in. That’s why I shy away from using PER (or WoW) as an absolute reference, I often site the individual stats (eFG, per-40, etc.)

  27. Frank O.

    Okay. I too am new here, but I think the laugh test is important because statistics, as we see in politics and elsewhere every day, can be manipulated to make whatever point a person is trying to make. I mean, statistically, one could say the U.S. won the war in Vietnam, but the actual outcome was significantly different…
    However, when Owen said that if Lee played as much as Bryant he has no doubt Lee would be the better player, it caused me to snicker.
    The reason is that you are extrapolating from data you have and projecting, in your mind accurately, Lee’s potential.
    But in reality, were Lee to get the ball as often as Bryant, teams would orient their defense to stop him. He would face double and triple teams. Bryant has the ball-handling skills, speed, athletic ability and basketball acumen to overcome those obstacles and still score 81 points in a game. Whereas Lee achieved his admittedly productive output last year as a sixth man, in most cases, producing often outside the opponents focus. His baskets largely were put backs, layups and dunks.
    I do not diminish his importance and feel he is very valuable to the Knicks.
    But if he were to attract the kind of attention Bryant attracts, and thrives in, Lee would be a single digit scorer and rebounder. He simply would be shut down.
    To say Lee would be the better player, if playing time and possessions were equal, is simply ludicrous.
    Hence the value of the laugh test…

  28. Ben

    Mike K. – Great point about the individual stats. I totally agree, every individual stat has other factors, does that player rebound well because he is really helping his team get more rebounds or is it because he is taking rebounds from another teammate, does that player shoot a really high percentage because he is really good at scoring or is it because he is playing with a great PG.

    It is because of these relationships I think you have to go really in depth to truly figure out a players value, comining box scores with +/- rating, looking at other players performances when the player was on the court and off. The more I try to find broad numbers and stats to rate a player the more I find it is all an illusion.

  29. KD

    Remember, Berri’s not in it to inform. He’s in it to have people talking about him, and to get paying gigs that inspire people to talk about him. His non-basketball gigs won’t have people talking about him on blogs and such — so that’s why he does what he does. I’m sure he knows that his findings didn’t pass the laugh test, so he just stuck with it and decided to be the contrarian’s contrarian; champion of rebounders everywhere.

  30. Dylan

    “Dylan – Berri doesnt think Rodman is better than Jordan. He never said anything to that effect.”

    And I never said anything to that effect either. I didn’t use the word better. I said Berri claims that Rodman produced wins at a higher rate than Jordan and, therefore, replacing Rodman with an average player harm the bulls more than replacing Jordan with an average player. Is that not what the wins produced metric means?

    My problem with Berri is that when his results don’t pass the laugh test he doesn’t use that as an opportunity to improve his analysis. He is so sure he has figured out the one metric to rule them all that he is blind to it’s flaws. It’s a fatal flaw for an academic.

  31. Frank

    Gotta agree with my namesake Frank O. — I’m relatively new to this boards as well and already Owen and I have gone back and forth a few times about the relative merits of David Lee. But Frank O. has got it absolutely right — the only reason David Lee seems as good as he does is because no one is really paying attention to him. Sure, the opposing coach says “try to box him out” but I guarantee you no coach goes into a Knicks game and game plans against David Lee. He’s only good because he has players around him who attract more attention. Does that make him a great player? It makes him a good player since there are so many players in the league that get ignored by the other team and still stink. But saying he’s anywhere in the same universe as Kobe Bryant as a player is so ridiculous that failing the laugh test is not even a strong enough description for it. That’d be like saying the flea-flicker is the best play in football because it accounts for a lot of big plays — yes, but it’s only because you’ve been pounding the other team up the middle for 5 yards/carry and the other team is committed to stopping the run, thereby allowing a trick play to beat them. I liken David Lee to the play-action pass — very effective but only when the clear threat of something else has occupied the other team’s attention. Sure, it’s a staple of any offense, but is completely dependent on a much more integral part of any football offense — a strong running game.

    And the trouble with arguing against David Lee’s greatness is that the better the Knicks are, the better he will be and the more WoW believers will trumpet their statistics. The more attention Zach and Eddy get in the post, the more David Lee benefits. And that’s great– very symbiotic. As a Knicks fan, I love what David Lee does. But I don’t think it should be ignored that he’s the play action and Eddy/Zach will be the running game.

    Can you imagine how good he’d be if our running game was Kobe? Or D-Wade?

  32. KD

    I do have to take on the revisionist history regarding the O’Neal trade to the Pacers.

    Portland had already traded for Shawn Kemp. They had Arvydas Sabonis at center, they had Rasheed, and they had the ability to stick Scottie or Bonzi in the post. Detlef Schrempf may have come back for another year. They weren’t hurting for what O’Neal was back then — a burgeoning scorer who could rebound and block shots. That team was loaded.

    What they did need was a solid body that would get away with things with the refs (unlike Arvydas and Rasheed, and Kemp probably), and O’Neal was an absolute fouling machine against Shaq the year before. No way that Jermaine, coming off the bench for those minutes, was going to get any respect from the refs and have a fair chance guarding Shaq with that frame.

    They knew what they had in O’Neal, knew he’d be a 20 and 10 guy down the road, and made a trade that they knew would be for 2000-01 and little beyond that. They wanted Davis to sop up minutes, not O’Neal to put up big numbers alongside Rasheed in 2002-03 and beyond.

    It seemed to be the fans and media that were (for some reason) surprised at how good O’Neal was with the Pacers, like he was some bust or something with Portland just because he couldn’t get minutes. Meanwhile, Portland had been rumored to not want Chicago’s 4th and 7th picks for O’Neal just a few months before the Davis trade. They didn’t think it was enough.

    Davis was enough — exactly what they needed on paper, for 2000-01. On a sane team, one that doesn’t throw towels at each other, that was a good trade for that year. Portland made heaps of mistakes, and this is one of them, but I’m not going to slam them too much for that trade. They knew what they were giving up.

  33. Brian Cronin

    I get your point, KD, and you’re correct that that WAS what Portland was thinking (O’Neal will be better than Davis, but Davis fit a specific need better), but I disagree that they shouldn’t be slammed too much for that trade.

    Even considering their needs, Dale Davis was not enough to get back for Jermaine O’Neal.

    Then again, Camby for Oakley was a similar trade, so I guess Portland at least has company in foolishly moving young thin players for bigger veterans.

  34. Z

    “Then again, Camby for Oakley was a similar trade, so I guess Portland at least has company in foolishly moving young thin players for bigger veterans.”

    Stat hounds may have known Camby was going to be one of the most valuable players in the league early on, but again, he had personality issues that he overcame. Van Gundy didn’t warm to him until the second round of the playoffs his first year (after asking for a breather in a game V.G. benched him for several weeks). He won over Knick fans, but it took a while.

    I’m sure the Rodman for Sean Elliot trade looked good to someone. Rodman was crazy and Elliot was supposed to be stabilizing. Elliot got depressed, ate McDonalds everyday, got fat, played like shit, and got run out of Mo Town. A few years later, back in S.A. he was an intergral part of a title run.

    There are many other suspect trades in recent history– Rodman for Purdue (did they have an equal WoW or PER??); Webber for Gugliotta; Ewing for Rice; Barkley for Horry and Cassell; etc…

    Stats fail to recognize a slew of reasons these trades occured. Stats probably justify Q for Kurt Thomas, after Q had a few big seasons in LA and Phoenix. But they don’t take into account his back had an expiration date quickly approaching.

    And who knows, maybe Frye for Randolph will be another exaple of: “foolishly moving young thin players for bigger veterans.” What will stat hounds say then?

    “Gotta agree with my namesake Frank O.”

    I thought you two were the same person. Sorry…

  35. Felix

    As a HS basketball coach, i can say there there isnt any one single formula, statistic that can rank players into one figure. There are too many complexities. Basketball is a sport where you can be effective at one aspect without being effiecient in others. scoring, rebounding, perimeter D, man to man D, steals, blocks, charges, fouls etc.
    There are simply too many facets of the game to determine one player better than the other.
    Attempting to determine a player’s effeiciency rate is highly flawed becuase that one players initial effeciency rate is dependant on his teamates. (Ron Artest is a good defender but his team wasnt)
    Its easier to break it down position by position. Or by category, scoring, blocks, etc
    Most formulas mix a bunch of these facets but still leaves other out.
    I love D. Lee but it is upsurd that he would be better than kobe, Kobe and Duncan are the two best 2 way players in this league, niether has weaknesses in the different facets of the game.
    Lee still is not a top scorer, top defender. while Kobe is.
    Kobe is still a good rebounder given that he takes about 30+ shots a game. Kobe still averages the same amount of assists as when he won 3 champoinship rings with SHAQ.
    Lee IMO (and this yr stats do give me some merit) is an elite rebounder. (kobe isnt an elite rebounder, just good)
    Lee is a good passer. Kobe is a very good passer.(when he choses to pass. lol)
    As a coach i always look to a player’s weaknesses to determine how good he is, not his strengths. A good scorer is a good scorer, how good is another debate. but can that good scorer be a good defender, rebounder, team player etc.

  36. JChan

    Not sure who is still reading the comments down to here, but just wanted to drop a few things here for the record. On my Win Score Stats site I am using DBerri’s Win Score formula. Then I take my own interpretation of a position adjustment (I use 82games as a reference, if you’re interested), and use that to calculate a position-adjusted Win Score. Then I use that number to calculate an approximate Wins Produced / 48 minutes, using a shortcut calculation that DBerri provided.

    This leaves me two steps apart from the official Wages of Wins calculations for WP48. Over this offseason I’m working on a revamp of the site, and one of my goals is to work with Dave to try to better represent his formulas. But as of now, my stats are in no way the “official” Wins Produced stats. Hopefully in the future they will be and we can use them for more in depth discussions.

    I really like the debates going on here, and hope they keep going. As for my take on the subject, I believe WP48 does not tell us who is the better basketball player. But I believe it does tell us who contributes more per minute to their team winning. If I were Mitch Kupchak, would I trade Kobe straight up for David Lee? No. But do I think he contributed more to winning per minute he was out on the floor last year than Kobe did? Yes. And would I try like mad to get David Lee on my team to compliment Kobe? Yes.

    Those who think rebounding is so overrated in Dberri’s system should try something. Watch a basketball game, but try to ignore who is shooting the ball. Focus on the rebounding. There are players who just grab the ball if it comes to them. And then there are players who consistently position themselves correctly, box out and put a full effort into getting every possible rebound. Yeah, Eddy Curry can get 7 rebounds a game by being tall. But those that really work at it affect the game immensely. Whether they are taking away the other team’s chance at another shot, or giving their own team’s shooters a mulligan, rebounding is a key part of the game, and one that is just as important as shooting. Everybody wants to score, but not everybody wants to rebound. And that’s what makes those that put in the real effort so important to a winning team.

  37. ASG

    I think it’s important to note that the year Iverson had success, he was surrounded by low turnover, defensive rebounding types that consistently rank high according to WoW. Something to consider, though: if you stick his 4 teammates during that year (Mutombo, Snow, McKie, Lynch) with an average guy replacing Iverson, what’s going to happen? Someone has to score and handle the ball, which is something none of those guys can do consistently. You can grab all the rebounds in the world, and shoot with the highest efficiency, but if you can’t generate a shot and can’t bring the ball upcourt you’re not playing basketball. This is what I think that Hollinger’s Usage Rate is trying to compensate for, even if it is overweighted in his system. In an end of game situation, giving the ball to Allen Iverson may be a better option than giving it to Dikembe Mutombo, shooting percentages be damned, because Mutombo may not even be able to get a shot up in the air. 50% shooting * no shot = 0 points, whereas 40% shooting * 2 pointer = sometimes, you’ll win.

    So is Lee more productive per minute than Kobe? Maybe, the stats seem to back it up. But can Lee ever be the focal point of his team’s offense, or realistically, even *A* focal point of his team’s offense? I don’t think so, and at the end of the day, you still need to score more points than the other team. Lee is a complementary player (and a damn fine one), and perhaps what WoW shows us is that a great complementary player can be even more important than a superstar to a team’s wins.

  38. Brian M

    “I believe WP48 does not tell us who is the better basketball player. But I believe it does tell us who contributes more per minute to their team winning. If I were Mitch Kupchak, would I trade Kobe straight up for David Lee? No. But do I think he contributed more to winning per minute he was out on the floor last year than Kobe did? Yes.”

    If the goal is to win games, and David Lee contributes more to winning per minute than Kobe, why would you not trade Kobe for Lee? Refusing to trade Kobe for Lee seems to imply that on the whole you believe he contributes more to winning than Lee, in spite of what the WoW stats might say. Or at the least, that there is some qualification involved in the interpretation of what it means to contribute to winning in the WoW system.

  39. Dan

    Neither Hollinger or Berri seems likely in near term to fundamentally reassess formula and make changes so I am not really up for engaging in that too much today but I will say why not blend the two ratings 50/50 (using z-scores to standardize the level on the different scales) and see the top 5-10 of that? Still won’t be perfect but might be worth seeing.
    I might try it later but can’t today.

  40. dave crockett

    “Basketball is almost like football in that sometimes it?s the RB (Barry Sanders), sometimes it?s the OL (Denver). In basketball it?s hard to distinguish the player from his teammates & the system they play in.”

    That should come on surgeon general style warning label for every basketball (and football) stat.

    I find myself in general agreement with Brian M’s comments about WoW’s underlying intuition. I think WoW is onto something by challenging the conventional thinking that equates usage with quality. Usage may be correlated with quality but that correlation may be quite low. What I struggle with is that WoW *seems* to argue that usage and quality are orthogonal, not related at all. I have a very difficult time accepting that. As Brian M implies, the thinking behind an analytical device that rewards players for creating possessions but then punishes players for using them seems bizarre to me.

  41. Dan W.

    Looking just at top 15 on PER and the guys in top 5 on WP not in top15 on PER,

    and averaging the relative performance on these two metrics within this group,

    the 50/50 approach puts two from PER top 5 on the combined top 5 (Duncan and Nowitski), two from WP top 5 (Lee and Boozer) and one who was on neither (Garnett). Is this top 5 any better? That is up to the reader.

    Marion would get 5th place if the 50/50 rating was based on PER and PAWS/min and Camby 6th, both because of very strong PAWS/min rating; just as Lee’s performance in combined is because of the exceptionally high WP mark.

  42. Dan W.

    This quick combined ranking suggests a 20/10 guy on the block is very valuable. Maybe the most valuable player type over wing scorer or top passer or top defender. I have been disposed to think that already looking at most championships teams since Jordan (which means Shaq and Duncan and Hakeem).

    What about Lee? Could he be close to a 20/10 guy on right team? Depending on what the Knicks want to spend and how Randolph works into the mix we might found out if a team grabs him away and gives him a bigger role. Give him 6 more possessions and 5 more minutes on the court could he have per game pts/rebs close to Dwight Howard? Maybe 6 more possessions is too much and his ceiling will be exposed. He has only taken 10+ shots in 8 NBA games and the max is 13 so the record doesnt answer this yet. And his college role doesnt really answer it that well either. Lee might split the difference and be able to step up half the distance, half way between Bogut and Howard.

  43. Ted Nelson

    I think it’s absolutely fair to bash Portland for trading JO for Dale Davis.
    Even if they understood the short-term/long-term trade off they were making I would say they made the wrong choice (unless you look long-term enough to say they ended up with Greg Oden, LMA, and Roy 7 seasons later). They didn’t even improve in the short-term (obviously not due entirely to this trade): they won 59 games in JO’s last season there, then didn’t top 50 in the next 3 seasons before falling out of the playoffs for the past 4 seasons.
    Every move made by evey GM has an explanation, having a justifiable reason for making a mistake doesn’t excuse making that mistake.

  44. JChan

    “If the goal is to win games, and David Lee contributes more to winning per minute than Kobe, why would you not trade Kobe for Lee? Refusing to trade Kobe for Lee seems to imply that on the whole you believe he contributes more to winning than Lee, in spite of what the WoW stats might say. Or at the least, that there is some qualification involved in the interpretation of what it means to contribute to winning in the WoW system.”

    Why would I not trade Kobe for Lee?

    1. Any GM that did that trade would be run out of town.

    2. Lee only has two years in the league, we cannot say for sure yet if his fantastic year this last year is going to be indicative of future success (although it is likely).

    3. I understand the difference between a star and a complimentary player. You cannot compose your entire roster with complimentary players. It is highly likely that if Lee was asked to take 25 shots a game, that his efficiency would go down.

    4. DBerri says himself that the Wins Produced formula does not absolutely answer all questions, but that it should be used as a starting point to make proper evaluations. He has never said, “Take the top 5 in WP48, that’s your perfect team”. Instead, take the formula and what it tells you about each player and how they are helping their team to win games. Then use your basketball knowledge to try to build the team that will win the most. And that means that you find yourself a star or two who are efficient and can score, like Duncan or Nowitzki or Kobe or Yao. And you avoid those who score but are inefficient, like Iverson and Joe Johnson, Zach Randolph or Ben Gorden. Then you surround your stars with guys like David Lee. Guys that hustle, guys that do more good things than bad.

    Anyway, I always say more than I mean to. I love this discussion.

Comments are closed.