Is it possible in principle for someone to do better than the guy on the barstool?
With your training, I think you'll understand that yes, it's possible to do better than the guy on the barstool, but the guy on the barstool will never be wrong.
Three points:
1. I haven't seen much of May, but he looks like another Corliss Williamson to me, which means he's good for $6 million/yr down the line. Given that by the time you hit the 21st pick of the 1st round, you're taking a flyer, I don't think a first round selection would be inappropriate. It's a constrained labor market - you draft who you can, and pay them what you're told to pay.
2. Yeah, with a few exceptions (West, Petrie, Grunfeld, Colangelo), GMs appear to do an unspectacular job of using their picks. If anything, they seem to do a better job now than they did in the past (google LaRue Martin). I think this track record is a function of (a) their willingness to buy into hype, (b) their use of hype to hedge risks against job loss, (c) their inability to make good judgments about young players, and (d) a mistaken belief that players' abilities exist as something purely natural, and unlikely to be effected by their ability to train them.
3. Really, though, you could make the same point about any labor market. I've yet to find the person who doesn't think someone he works with or under is a complete moron who must be blowing the higher-ups in trade for advancement. The base of the interestingness of Moneyball is that a highly competitive, well-defined, well-studied field like baseball is so piss-poor at making labor judgments; the (usually) subtext is that the judgments in your field are almost certainly worse.
are, somehow, natural
Those are good points.
1. Williamson, without the toughness.
2. (d) leads to something I wonder about a lot: the extent to which what people call "fit"--the interaction between a particular player and particluar situation--determines how the player will do. Would Gerald Wallace be a star if he had been a Bull rather than a King? Would Tyson Chandler have been a solid, exciting player if he'd gone to a contender that didn't expect him to score? Probably, in both cases. It's not that the GMs misudged the ability of the players to play; it's just very difficult to predict how a player's skills and (more important) personality, will reveal themselves.
3. Yeah, I read that and thought, "but that's what I just said," but that was the draft I didn't publish. Because pro sports are such a small world, and its members usually lifelong obsessives, you'd think there'd be a wealth of knowledge and expertise, but it seems to work just like other organizations, with people fililing positions more out of happenstance than suitability.
I think it's more than just the sort of fit you're talking about. Part of it is social. Jerry West used to say that you looked for the good atheletes from good programs; I think the "good programs" part spoke both to training and also to the socialization (a) to the world of basketball, and (b) to society at large, so that you could focus on basketball. I think part of Carolina's success in producing great pro players is that they were often really good at both (a) and (b). So, for example, I still wonder if Amare Stodemaire (sp) is going to turn into Shawn Kemp/fat Elvis edition down the line.
Have you noticed how crap most of the writing about the NBA is these days? It makes me want to weep sometimes.
Absolutely. In the pros, the difference between the really good players and everyone is almost entirely psychological. The Bulls put their prospective picks through a battery of psychological tests, which is the right idea, but those tests don't tell you anything useful. A "good program" is a decent proxy; probably as good as any, actually. Of course, players can have it together without having gone to a good program--there's Garnett, McGrady, Ben Wallace, to name a few--but if you go down a list of fantasy rankings, what's surprising is how many of the players near the top did go to name schools. There's the issue of whether good programs recruit or train good kids, but from the perspective of the pro drafter, it doesn't make a difference.
And yes, there's no NBA column I read regularly anymore. Lord knows there are enough psychodramas and soap opera plots to make for interesting writing, but it's just not there.
As an aside, the Seattle Supersonics are an interesting organization as they've been regularly mentioned by John Hollinger as the team that's doing something closest to "moneyball" in bsketball and this year they've hired NBA stats gure Dean Oliver as an advisor.
There were some interesting articles about this Part I, Part II, Part III, and Part IV
The person who wrote those articles, Kevin Pelton, is a young guy who's managed to get a job with the Sonics but has also done an enormous amount to build a fan community that's interested in stat based analysis of hoops. If anyone's interested I can provide more links.
Wow, thanks Nick, I'll check those out.
Glad you're interested. I suppose I should go ahead and mention the top spots on the web for NBA stats:
82games.com for detailed player stats and articles. Information is based off a database containing play by play data for every game from the last couple season.
popcornmachine for game flows (this is a little too time consuming and geeky even for me.
basketball-reference.com for historical stats. It includes most of the "cutting-edge" derived statistical measures on the player pages (like offensive and defensive ratings, player win %, etc . . .).
APBRMetrics and knickerblogger are good sources of discussion. APBRMetrics is where all the big name NBA stat's people talk shop.
err, "stats" not "stat's" on the last line. I should know better than to make typos commenting at unfogged . . .
It's ok, w-lfs-n's on vacation. Thanks for all the links.
(b) their use of hype to hedge risks against job loss
I think this is the most crucial point. I work in the finance industry (as a programmer), and one thing that's been drummed into my head by pretty much everyone who does stock analysis for a living is that by far the worst thing in the world for an analyst is to be wrong when everyone else is right, especially in the Internet age. Thomas Friedman calls this the "electronic herd", and I think the same effect often applies to the people who make decisions about which players to draft, sign, trade for, etc.
In the aggregate, the quality of decision-making by GMs almost can't rise too far above that of the guy on the barstool, because such an increase in quality would necessarily require too much risk-taking on the part of GMs as a group. They'd rather just take the player that everyone expects them to take, even if their own internal evaluations lead them to a different conclusion, because if they're wrong, well, who can blame them? Everyone else was wrong too.
Walter:
I think the phenomenon you note is fairly widespread. Once upon a time, people said, "No one ever got fired for buying from IBM. Now I think it's Dell.
That said, I think part of it is also that we simply don't understand what goes into basketball (or finance) nearly well enough to discriminate between our own judgments and those of the herd. There are those that can, and they do phenomenally well, but they are the exception. But I think the best of them (in any field) have a felt understanding for what is well-understood, and then start taking risks in ancillary areas. They know what areas are likely to be undervalued, and so what areas are likely to be productive even if they make choices in that area in a completely random fashion. So Petrie, for example, trades Webber for three journeymen b/c he think that three journeymen will be significantly easier to trade down the line, or will be better (b/c undervalued) than expected. Or West takes a bunch of upper-middling players with the intent of shopping them for newer risks down the line. But the key, I think, is an acknowledgment of how little they can know, and a willingness to trust their own judgment once that acknowledgment is in place. And getting that kind of space (to make mistakes and not have to change b/c of them) is probably an accident of fortuitous fate.
At the end of the day, I think lots of people succeed, where not sheer luck, b/c of simple competence - it turns out that 80% of it really is just showing up.
A couple more links worth mentioning (assuming that I haven't overloaded you).
hoopsanalyst has nice, short analysis pieces but doesn't update frequently.
There are two articles by Kevin Pelton on hoopsworld that are worth reading. The Economically Efficient NBA Team in which he talks a great deal about the value of rookie contracts and the importance of drafting well late in the draft (see also this analysis of the value of draft picks).
Kevin Pelton also did worked on trying to predict the NBA potential of college players in the 2004 draft Part I and Part II. Looking back you'd have to say that his predictions are reasonably accurate. Presumably he will do a similar analysis for the upcoming draft at some point.
There are links to all of Kevin's Hoopsworld columns here (unfortunately the search function on hoopworld only turns up recent columns, so this list is very nice to have)
I would also recommend that if you want a good intro to NBA analysis I recommend the first Pro Basketball Prospectus by John Hollinger. The later ones are good as well, but the first on benefits from "first book syndrome" -- he's had years to think about what he wants to say, he wants to be as persuasive as possible, and he packs everything that he can get into it. And it looks like half.com has one copy selling for $1.50 + shipping.
But, really, the first 4 articles that I linked to about the Supersonics playing moneyball are a great place to start. Post a comment if you do get around to reading them.
It's also interesting that Dean Olliver has frequently commented that most NBA teams don't use these sorts of statistics very much. There are some, Mark Cuban's a big fan and I get the sense that Indiana and San Antonio use some statistics but Seattle is really ground zero for the use of statistical analysis in the NBA.
Nick -
I really hope you're not Belton or a sibling, but I just read a number of the articles you cited, and he's gawd-awful. Maybe that's too harsh; what analysis I saw added no new useful information to conventional wisdom, and, where it ran contra cw, it was just wrong.
For example, he looked at the 2004 draft and restricted his analysis to college players. But the best pool of players are probably either entering high-schoolers or foreign players, so he's really looking at a watered down pool that will mostly fall into the latter half of the first round - complete flyers and bench warmers. So who cares, unless he unearths a diamond. Now look at his rankings of college "point guards": Nelson, Harris, West, Gordon. The only reason that list isn't completely wrong is because Harris and West are roughly equivalent and roughly in the same places they'd be anyway. But if you were drafting today, the list would clearly be closer to Gordon, Harris/West, Nelson. Actual draft order? Gordon, Harris, Nelson, West. We'll see, but I bet West has a better NBA career than Nelson (though both are second-unit).
In some ways, it's not his fault. I'm not sure basketball yields itself to statistical analysis nearly as easily as baseball. Baseball you can break into 1-1 competitions; it's a lot harder in basketball.
SCMT,
You miss the point of the article. He wasn't posting his personal mock draft. He was saying, "I have tried to develop a formula that predicts how college stats will translate into NBA stats. Here are the results of the formula as applied to college players entering this year's draft."
He's only working with college players because that's the only group of players for which there's sufficient statistics to do that sort of study (though I head Hollinger has tried to work on something similar for foreign players).
Nobody claims that in Basketball statistical analysis will replace scouting (if you read some of the other links, everyone acknowledges that it's much more difficult in basketball than in baseball) but they claim that statistics can be a useful tool.
BTW, I see that I messed up the link to Part II of the draft preview.
See this discussion about how the value (or lack) of the current models. One person does say, "ANY of us statheads would do a SIGNIFICANTLY better job in drafting and especially free agent pickups than many GMs." But most of the people are relatively modest in their claims.
SCMT,
Two questions. What would an article have to contain to count as interesting information to you? You say that most of what was in the articles was simple cw. I'm curious what you think is included in the conventional wisdom.
I would look at statements like the following and say that they're pretty basic statments of an educated Fans perspective but not yet part of the general cw:
"It's a pretty simple concept, but one that has largely escaped most NBA front offices: The idea that what a player does on a per-minute basis is far more important than his per-game stats."
"In the old days - I know, my first year here - you start talking about a player in a trade and you start breaking out the old NBA Guides," says Walker. "'Here were his stats the last three years.' That's just the way it was always done. To be able to see, 'Well, maybe there's a trend on a per-minute basis over the last four years, the player's productivity is going down or going up,' is a little better than just dragging out the old books."
"While Oliver calls his system the Four Factors, it could just as easily be named Eight Factors, as each category is applicable on both offense and defense. Research by Ed Kupfer has shed light on the relative value of each factor (Kupfer did do a slightly different study, using free throws attempted instead of free throws made):
Off Factor Def
22.4% Shooting 19.5%
9.2% Rebounding 7.8%
6.5% Free Throws 8.7%
13.2% Turnovers 12.6%"
Nick:
1. In the species of articles that Belton writes, I'm looking for interesting analysis, which means counterintuitive analysis that is convincing. If you write an article arguing that the best point guard in the NBA is Jamaal Tinsley, I'm going to discount the rest of what you say because you're just on-the-face-of-it wrong. If you make an argument that McGrady is one of the five best players in the league - who cares, I knew that already, and all your pretty graphs are for naught. An interesting article might be one that argues that signing J. Kidd to the Spurs would have been disasterous for the spurs, b/c his value is highest in an open game, and Duncan's value is highest in a slower game that allows him to use his footwork around the blocks. (That actually seems pretty straightforward to me, so it's not that interesting).
The problem is that Belton is looking for a quick and dirty solution/metric to a game that isn't easily quantified. Baseball is really a series of well-understood games stitched together: pitcher/batter, race, catch. Basketball is much, much more complicated; a great scorer might (like Marbury) be a detriment to the team b/c he dominates the ball too much. So, at this juncture, interesting analyses are to be found in the smaller questions. What are the classifications of team styles? How efficient are each at getting X wins? How cheap are each at getting wins? Are there styles that will get you 40 wins cheaply, but never past the first round (i.e., the Grizzlies)?
2. Some of the metrics you mention seem obvious to anyone who loves the NBA. Who hasn't argued that his favorite player should be getting more PT b/c look at what he does in such a ltd time? But more importantly, I think those metrics suck at predicting who has most value. For example, per min. stats, for example, only matter if they also include the effect on the efficiency of others that using that player has. So, again, I don't think such stats reveal much counterintuitive useful information unless situated in a broader model that hasn't yet been fleshed out. I guess my question is - who do the stats indicate has been undervalued, and is your prediction borne out?
Okay, here are some things I would pick out as surprising results that are revealed by statistical analysis.
(1) "The Sonics play one of the slowest-paced games in the NBA, and the way to beat them is to go small and outrun them."
(2) The Jazz are the second worst defensive team in the league and commit by far the most fouls of any team.
(3) Manu Ginobili is, conservatively one of the top 20 players in the NBA. Note, Ginobili shows up as the 3rd best player in the NBA in this study.
(4) In general when a player that has been playing limited minutes is given more minutes their per/minute production rises (this is somewhat controversial, but the evidence for it is far stronger than I would have thought)
(5) The top two rebounders in the league are Reggie Evans and Dan Gadzuric
All of those statements are things that you might know if you follow the game sufficiently closely, but I think it would be difficult to consider any of them "conventional wisdom"
I haven't been following the NBA that closely this year, but one thing I've been wondering about is how the (relatively) recent rule changes have played into personnel decisions. I would assume that back when only the NCAA allowed the zone it was even harder to predict how a player would turn out, but it could be that it's so hard to identify talent that the specifics of the rules don't matter at all.
Unrelated note: Either all of my clocks are wrong, or the timestamps don't do daylight savings.
the timestamps don't do daylight savings.
Apparently, they take the system time, so if my host changes the time on their servers, the timestamps will change; if not, not.
I dunno, Ogged. He's got the skills and the hands to play on a pro level and one more year really hitting the weights (the change in his physique this year versus last year is striking) will make him faster and stronger. Where his pro game is suspect, though, is that in NBA terms, he doesn't have the body of a center, but of a power forward. Nothing wrong with that except that he's spent his whole life learning to play with his back to the basket, rather than facing it, and that's a lot to unlearn.
All the same, go Heels! One more win, fellas, one more win.