Before I get to my thoughts on the 2008 Finals, let’s talk predictions for a moment.
In 2006 I failed to predict the NBA champion before the playoffs started. Actually I am not sure I offered a prediction. Nevertheless, the Miami Heat did not lead the league in efficiency differential that season so if I had made a pick, it would not have been Miami.
In 2007, I predicted the Bulls and Suns would face off for the title before the season started (obviously wrong). But before the playoffs started I said that since the Spurs led the NBA in efficiency differential that they should be the favorite to win the title.
Before the 2007-08 season – based on the past performances of Kevin Garnett, Ray Allen, Paul Pierce, James Posey, Rajon Rondo, etc… — I predicted the Celtics would win the title.
And now for the 2008-2009 season, I am going to predict the Lakers will win the title. Just for the record, I made the prediction before the 2008 Finals had concluded.
What do all these predictions mean? One could argue that I am getting better at predicting. But that’s not what’s going on.
As I noted a few days ago, a team’s efficiency differential is an excellent way to evaluate the quality of a team. If two teams meet in the playoffs, the team with the better efficiency differential tends to win. This doesn’t always happen (see 2006), but it often does. In sum, making predictions in the NBA – relative to baseball, football, and hockey – is relatively easy.
The predictability of the NBA is something I want to talk about in more detail. For today, though, I want to offer my thoughts on the 2008 NBA Finals.
Pierce for MVP?
We know Paul Pierce was named MVP of the Finals. But was he the M2P (Most Productive Player)?
In Games One and Two, Paul Pierce was the top Celtic in terms of PAWS (Position Adjusted Win Score). In Games Three and Four, Ray Allen was the top Boston player. And then in Game Five, Pierce took top honors again.
On Tuesday the Celtics closed out the series. As Table One indicates, Ray Allen – just as he did in Games Three and Four – led the Celtics in PAWS. So in the contest between Pierce and Allen, the final score was 3-3.
Table One: Analyzing Game Six of the 2008 NBA Finals
What if we looked at the performance by each player across all six games?
Table Two: Analyzing the NBA Finals
As Table Two indicates, the M2P — when we look at the entire NBA Finals — is Ray Allen. And the contest isn’t really that close. This is because when Allen was good, he was really, really good. And when Pierce was bad, he was really, really bad. Consequently, if PAWS was our metric of choice (and of course here it is), Ray Allen would have been named MVP of the Finals.
The Greatness of KG
And second to Allen would have been Kevin Garnett.
We already know (at least we should know) that KG was clearly the M2P (Most Productive Player) of the Celtics in the regular season. What about the entire post-season (not just the Finals)?
Table Three: Analyzing the 2008 Post-Season for the Celtics and Lakers
As Table Three indicates – and this should not surprise – Garnett was the M2P of the Celtics in the 2008 post-season. In other words, if we consider the fact you have to win the first three rounds before you even get a chance to play for a championship, then KG should be considered the biggest reason why the Celtics won their 17th title.
My sense – having listened to the announcers call Game Six – is that the Celtics taking this title changes people’s evaluation of Garnett. In reality, though, Garnett in Boston was very much the same player he was in Minnesota. In Boston, though, he got to play with better teammates. Unfortunately TV analysts – unlike player statistics – have trouble separating the player from his teammates. Consequently, analysts now think KG has gotten better.
What about Kobe?
And I think, some analysts might think just the opposite of Kobe.
The media declared Kobe was the league’s MVP and throughout the playoffs we kept hearing that Kobe was the best player in the game. And since we know the team with the best player must win the title (do we know this?), the Lakers were declared the favorites in the NBA Finals.
In the Finals – as Tables One and Two indicate – Kobe was not quite “like Mike.” Nor was he “like Mike” in the regular season or the 2008 playoffs. But Kobe was certainly the M2P of the Lakers in the regular season. And he also was the M2P of the Lakers in the entire 2008 post-season.
What if we compare KG and Kobe? Although Kobe was very good in the regular season, Garnett was even better. And although Kobe played well in the post-season, again, KG was better. And obviously in the Finals, Garnett was better. In sum, the team with the best player did indeed win the NBA Championship (not that this has to happen).
My sense is that many people, having seen the Lakers get blown out in Game Six, can buy into this storyline. But I think these very same people will tell a very different story in 12 months.
Andrew Bynum is supposed to come back. As I have said before, if Bynum can produce as he did earlier in the year, the Lakers are the favorites to win the title next year (and the year after that as well). When that happens, we will most certainly hear the following stories (from the very same people criticizing Kobe today).
1. Kobe is the greatest player in the game. We heard that this year without the Lakers winning a title.
2. Kobe is as good as Michael Jordan. Again, we started to hear that in the playoffs this year.
3. Kobe has willed his team to another championship.
Although I fully expect to hear these three statements (assuming the Lakers win in 2009), I also think that Kobe – who is a very good shooting guard – will still be the same Kobe. In other words, I suspect that the very same writers that are a bit down on Kobe today; will be just as high on the same Kobe in 12 months. Remember, many of the same analysts who were down on KG 12 months ago have apparently changed their perspective on Garnett.
Again, we should expect basketball analysts to separate players from teammates. But again and again, we see that this doesn’t happen. With that in mind, keep track of the people writing negative stories about Kobe and the Lakers right now. If Bynum comes back as we expect, those same people will write a different story – about Kobe — in just a few short months.
– DJ
The WoW Journal Comments Policy
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
Wages of Wins Journal 
Ap
June 19, 2008
Amen to those comments about journalists changing their tune. I think, honestly, most writers seek to verbalize the majority of fans opinions (since, people theoretically enjoy hearing what they already believe due to confirmation bias–isn’t that why there is bias in the news media?), and not what is actually true. And since most fans see the scoring and such as the only determinants, its only logical that writers will do the same.
C
June 19, 2008
Maybe I’m just being optimistic (as a Celtics fan), but I suspect that Kobe’s failure to win in 2008 has permanently shaped the way that he’s viewed. For starters, I suspect the Kobe = MJ comments will just about vanish. And even if the Lakers do win it all next year thanks to a boost from Bynum, I think most commentators will see the cause and effect, and will recognize that Kobe needs a very good supporting cast to beat a team as good as the current Celtics.
Nick
June 19, 2008
I understand the scale of PAWS/minute easily. I am sticking with that.
Logic
June 19, 2008
I laud the predictive success of the system.
I still think that separating the individual’s production from the team’s is an exercise in futility, because of the large interdependency of basketball statistics.
porteno
June 19, 2008
Ignorance is bliss, “logic.”
Chris S
June 19, 2008
I’m curious as to how the addition of Bynum (for a full healthy season) alone would make up the difference in efficiency differential between Boston and the Lakers. Further, it seems that even nothwithstanding a prediction that the Lakers will have a better efficiency differential, the Celtics might still be favored since they would be likely to have an easier road through to the finals in the eastern conference side of the playoffs.
Logic
June 19, 2008
Please share your experience, porteno.
Another Pete
June 19, 2008
I still think that separating the individual’s production from the team’s is an exercise in futility, because of the large interdependency of basketball statistics.
There’s some good evidence to suggest that it’s not an exercise in futility (in the WoW book and elsewhere). You claim a “large interdependency of basketball statistics”, but you don’t present any evidence that for why that’s true or important. You’ve given anecdotes, but anecdotes aren’t evidence, and are often misleading (Brian Scalabrine dropped 29 points against Golden State…).
I think it’s ok to be skeptical about the whole WoW approach, but some some reasons for being skeptical are more honest than others. To me it looks like most posters here who don’t like it don’t like it because it doesn’t agree with their preconceptions… not a big surprise. It’s tough to reexamine preconceptions.
I know that I’m probably innately biased towards being convinced by a system like WoW before even looking at the data — I like the idea of a different approach stirring things up and showing up the “experts”. It’s a big reason why I enjoy books like Moneyball and WoW (and Freakonomics and…), as well as reading this blog. Plus, I think it gives a different kind of insight into the game… ie I feel like I have a better understanding of the value of a rebound after all the discussions about it here.
Kop-TV
June 19, 2008
Independent Liverpool FC Website
http://kop-tv.com
Logic
June 19, 2008
Interestingly enough, I read the Freakonomics blog as well, and I enjoy it.
I just don’t enjoy WoW, probably because I play basketball frequently, which may lead to my skepticism toward a statistically-based system being able to describe the game. This may explain my innate bias not to believe in its conclusions, I admit, and should not be relevant, but we are humans here, at least most of us are.
As far as why the interdependency would matter in assessing an individual’s productivity, it affects WHO or WHAT is responsible for their increase or decrease in productivity – is it another player? Is it the team’s system? Is it the opponents? Is it the player himself? It affects what the conclusion is if player A’s “productivity” is higher than player B’s. Does that mean player A is a better player than player B? Or are there other factors at work?
Baseball seems to work better for this type of analysis (yes “anectdotally”) because the statistics are discrete: you have the batter, by himself, against the pitcher, with some variabilty in fielders. It’s not so simple with basketball.
Take for example rebounds.
I’ve played on many teams where one player got the majority of uncontested rebounds because that was their “role”, while his teammates ran down the other end of the court to play offense. He could collect as many as 10 of these in a singe game. If his teammates fought him for those rebounds instead of going down court, he would have fewer rebounds, and they could have more. So, how can you tell from the statistics how good a rebounder he is? You can’t. You have to look at his technique on contested rebounds, because the statistics would be muddled by all of the extra rebounds he collected simply because his teammates conceded the rebound to him.
This is only one example, but I assure you there are many more examples where the statistics do not reflect the actual skill of the player.
Jeremy
June 19, 2008
Berri, I love how you are pointing out that KG is better than Kobe. I would imagine if this poll was posted on ESPN, America would respond 92% that Kobe was better than KG. This is not the case today, and has never been the case. I cant emphasize how much I appreciate this point being hammered home. We saw the TRUTH in the NBA Finals. Kobe was the 5th best player on the court alot of the time.
ilikeflowers
June 19, 2008
So the rebound is worth less because of the role? That’s the entire point behind position (role) adjustments. Or if you put in someone who rebounds prettier into that role and they grab fewer rebounds then they’re still better? Rebounds are worth whatever gaining a possession is worth, period. Some of the credit indeed goes to the position played. No one expects a point guard to rebound like a center.
I assure you that there are many more examples where your eyeball evaluation of the skill of a player does not reflect the actual skill of the player. Especially, when the skill being measured is the ability to produce wins and not the ability to impress eyeballs.
Logic
June 19, 2008
Depends on whose eyeballs are doing the looking.
ilikeflowers
June 19, 2008
Exactly, that’s the problem.
Logic
June 19, 2008
Yep.
porteno
June 19, 2008
In other words, Logic has prejudices about the value of regressions.
Now, I might lambaste him for being so close-minded. But I have areas in which I reject scientific orthodoxy, so I don’t want to be too harsh. That said, I come up with better reasons for my departures from scientific orthodoxy than, “I play basketball, so my prejudices are accurate.”
Mike
June 19, 2008
Anecdotally, Kobe had a real tough time in the series, and Pierce was a fierce defender on him. Surely there’s some possibility of causality here.
But as far as I can tell, PAWS doesn’t account for things like “good defense that forced the best player on a team to pass it instead of shooting himself”.
If I was on the Celtics and I put Kobe in a bear hug for 48 minutes (literally) and did not otherwise help out on offense or defense, I think we can all agree I’d be the best player on the court (as far as the Celtics were concerned). But my Win Score would be 0.
I guess I’m trying to say… how do you account for the defense that’s not a part of what you’re accounting for with Win Score? Or do you not believe it exists? Or the things not measured in WS are just of negligible value?
porteno
June 20, 2008
_If I was on the Celtics and I put Kobe in a bear hug for 48 minutes (literally) and did not otherwise help out on offense or defense, I think we can all agree I’d be the best player on the court (as far as the Celtics were concerned)._
This is far from correct. Do you see why?
Another Pete
June 20, 2008
It affects what the conclusion is if player A’s “productivity” is higher than player B’s. Does that mean player A is a better player than player B? Or are there other factors at work?
I think this is a really interesting area to think about: “more productive” vs “better”.
I think it’s reasonable to argue that Kobe is the BEST PLAYER IN THE WORLD ™ at finishing tough, contested shots… (anecdotally :-).
Lets compare two scenarios:
Kobe drives into the lane, help comes, Kobe leaps and twists and flips up a crazy shot with two defenders draped all over him… score! Two points!
Paul Pierce has a defender on him, 18 ft. from the basket. Jab step. Jab step. Jumper. Swish. Two points.
The first scenario is more impressive, more exciting, gets the adrenaline pumping, and might even be harder to accomplish… but both scenarios are worth the same two points.
So, “better” really depends on what you’re interested in measuring… and it’s useful to realize that there are different levels of correlation with winning for different areas of skill. The WoW approach is an attempt to figure out, using math, which skills are most closely correlated with winning. It turns out that many people are frustrated to learn that “finishing hard shots” is perhaps not as valuable as its difficulty and “awesomeness” might suggest.
Another Pete
June 20, 2008
how do you account for the defense that’s not a part of what you’re accounting for with Win Score? Or do you not believe it exists? Or the things not measured in WS are just of negligible value?
It could be that the things not measured in WS are strongly enough correlated with the things that are measured that adding them to the equation is of negligible value… that doesn’t mean that the things themselves are of negligible value.
Mike
June 20, 2008
Porteno, I don’t. Maybe you’d be kind enough to explain? Or is that too much to ask?
Another Pete, interesting point. Still, just for clarity, I’d like to see it broken out if it’s at all possible – if for no other reason than to see the coefficients. And, while your explanation is possible, it just doesn’t make sense to me. How would something like FG% be related to defense?
Vince Gagliano
June 20, 2008
By the way, I doubt that Ray Allen is the “true” MVP of the Finals.
Even though both led the Celtics in PAWS 3 times, and both had records of 2-1 in that stretch, Pierce led Boston in PAWS when it set the tone for the series in the first 2 games.
Also, one of Allen’s wins was in a Game 6 that, ultimately, wasn’t really that much of a contest anyway.
Besides, giving the longest tenured member of the Celtics the Finals MVP is a magnanimous gesture on David Stern’s part; by honoring the team’s senior member and team captain, it symbolizes the step Pierce has taken from one of Boston’s really good players to one of its legends.
That’s not to say that Allen was unimportant. He was. However, I think that Pierce deserves the award for playing his best when his best was needed (to quote John Wooden)
Logic
June 20, 2008
Porteno thinks perhaps that I have a problem with the scientific method. I don’t.
Hopefully I’m stating the obvious when I say that basketball analysis is not “science”, especially when the data the analysis is based on is not scientific.
As far as Another Pete goes, I would agree in your example that both are equally productive. What I mean to say is that given that Player B is more productive than Player A, would Player B still be more productive if he were on another team, with a different role? Surely players’ productivity is not the same every year, only changing as they improve or regress in thier skills. So what are these other factors besides
each player’s skill that make them more or less productive? This is significant in evaluating a player’s worth. Without experience (or “eyeballs” as some dismiss it here) you can only say that Player B is productive in that particular context. This would be important if you were to use WoW as a tool to evaluate players, or, say to create a team as a General Manager. You could end up choosing Player B over Player A, only to find out that Player B was less productive in his new environment. Or you may find he is more productive. Or he may indeed be just as productive. But how would you project that from the numbers?
Put another way: if the value of a player’s skills is measured in his productivity, then why would his productivity ever change, except for a change in his skills? This is what makes separating the individual from the team nebulous. Some part of any player’s individual stats is due to the context in which he is playing, and a large part is due to the player’s skill. How much of the productivity is attributable to the player only, and what part is due to other factors that may change outside of him?
ilikeflowers
June 20, 2008
Good eyeballs are very useful and not to be dismissed. The question is how to test the predictions of eyeballs.
ilikeflowers
June 20, 2008
Logic, what is your preferred method of determining a player’s value and predicting the performance of a team when players are added and subtracted or minutes are changed? How accurate are your predictions? If you can’t demonstrate that eyeball valuations are better than w48 or PER or some other statistical method then what’s your point? We all know that statistical models have their limits (as do eyeballs) but you seem to claim that eyeballs are better without providing the slightest bit of evidence.
Another Pete
June 20, 2008
And, while your explanation is possible, it just doesn’t make sense to me. How would something like FG% be related to defense?
It doesn’t have to be FG%. If my hypothesis is true, I’d imagine quality of overall defense is more closely correlated to steals, blocks, fouls given and, to a lesser extent, rebounds.
Yes, it’s easy to find a counterexample (ie there are players who get lots of blocks but aren’t overall great defenders), but that’s ok… a system doesn’t have to be a perfect predictor to be useful, it just needs to be a “good enough” predictor.
Another Pete
June 20, 2008
Hopefully I’m stating the obvious when I say that basketball analysis is not “science”, especially when the data the analysis is based on is not scientific.
This seems like a very odd statement. Science is a method for looking at the world and the data it contains. The scientific method can be used to analyze anything where data is available. The result may be “no conclusions can be drawn”, or “this hypothesis cannot be proven”, but you get there by following a process, not by trusting hunches or gut feel.
What I mean to say is that given that Player B is more productive than Player A, would Player B still be more productive if he were on another team, with a different role?
If you troll back through some of the earlier posts on this blog, you’ll see that that’s a big part of what’s been looked at: it turns out that players produce pretty consistently, both across different seasons and when changing teams. Not perfectly consistently, of course, but consistent enough to be useful.
There are, of course, other factors that do contribute to how well players play. For example I think dberri has done research on the impact of coaches, and it looks like that some coaches have a positive influence on player performance. It’s good to know about these, but it’s still useful to understand that there’s consistency in performance even when these factors are, well, factored out.
porteno
June 20, 2008
Logic —
I don’t even need to give you a shovel to let you keep digging down the hole of ignorance, do I? ;)
Logic
June 20, 2008
Porteno – you seem to be ignorant of your own ignorance.
Logic
June 20, 2008
And quite smug, I should add.
Logic
June 20, 2008
Another Pete – thanks for your responses. They are appreciated.
And I think you all agree there are limitations to the system in question (“a system doesn’t have to be a perfect predictor to be useful, it just needs to be a “good enough” predictor), so those limits must be considered when evaluating Player A versus Player B, on the basis of their statistical production.
Another Pete
June 20, 2008
And I think you all agree there are limitations to the system in question
Absolutely. However, I believe that the “eyeballs” system, such as it is, can have even more limitations, depending on the eyeballs in question. As ilikeflowers has been suggesting, we currently have no evidence in support of the skill of your specific eyeballs. We do have evidence in support of Win Score etc.
Given that, and the evidence that even expert eyeballs are suspect (look at the lack of a relationship between payroll and team performance!), our skepticism shouldn’t be a surprise.
so those limits must be considered when evaluating Player A versus Player B, on the basis of their statistical production.
Of course. If one player has a wp48 of 0.250 across a season, and another has a wp48 of 0.230… well, I don’t know what the error level is with the model, but I’d guess that these players are pretty similar and it’s quite possible that the second player actually contributes more.
However, when we see one player with a wp48 of 0.406 (Chris Paul) and another has a wp48 of 0.247 (Kobe) I feel pretty confident that the first player in this comparison has contributed more.
merl
June 20, 2008
The Dallas Miami series was the only finals in the last 25 years when the team with the worse regular season record and the worst efficiency differential won.
Only the officials could have predicted the outcome of that series.
ilikeflowers
June 21, 2008
Mike, a player with that sort of theoretical ability would indeed break some models. +/- would need to be used in that case. Don’t forget to account for the negative effects that this player will have on his team’s offense however.
I imagine that good team offense can much more easily counteract individual defensive excellence than it can team wide defensive excellence and vise-versa.
Bruce Bowen for example has a reputation as a lock-down defender, but his net plus-minus is negative over the last six seasons and his best (only positive) contribution over that span was +4.7 (which may have been due to better offense that year or a particularly poor defensive backup or just random fluctuation). Even though his defense may win you that close game that was decided in the last few possessions, it can’t be overlooked that if he had been more consistent on defense or more productive on offense then perhaps his team wouldn’t have required his defensive greatness in the closing possessions in order to win the game.
mrparker
June 22, 2008
The question of man defense keeps rearing its ugly head. This time it regards Pierce’s defense of Bryant in the finals and an anecdote involving Bruce Bowen.
Making a guy take a bad shot does not constitute a defensive stop. Grabbing a defensive rebound, a steal, or taking a charge does. So anecdotally Pierce’s great defense on Kobe does not become a success for Boston until Kevin Garnett grabs a defensive rebound or Rondo gets a steal or Pierce blocks one of his shots and then grabs that rebound.
Logic,
Hang in there man. While I don’t agree with any of your points at least you are presenting a dissenting opinion without being a jerk.