The Statistical Wizardry of Bob Bellotti & The Washington Basketball Team | Truth About It.net

The Statistical Wizardry of Bob Bellotti & The Washington Basketball Team

By
Updated: October 28, 2010

{flickr/draggin}

The Intro.

- by Arish Narayen

Over the past several years, NBA organizations have increasingly integrated advanced statistics into their decision-making. But exactly how teams employ these statistics in personnel decisions — that is privileged information. The NBA’s trend towards quantitative analysis is seemingly personified by Houston Rockets GM Daryl Morey. It is easy to see why Morey is a darling to stat nerds everywhere: he never played in the NBA, and he got his bachelor’s degree (in Computer Science) at Northwestern, and an MBA from MIT. Morey also serves as chair of the MIT Sloan Sports Analytic Conference. As NBA teams, and media (bloggers) seek new ways to evaluate players, attendance at the Sloan Conference has grown.

In Michael Lewis’ NY Times profile on Shane Battier in February 2009, Morey and Lewis had the following exchange:

“Someone created the box score … and he should be shot.”

I tend to share Morey’s disdain for the traditional, counting stats of the NBA box score. Props to ESPN for adding player plus/minus into their box scores last year, as I’m sure at least a few NBA GMs (I’m looking at you, David Kahn) saw that stat on ESPN for the first time. But plus/minus is just the tip of the mathematical iceberg. While organizations like Dallas, Houston, Denver and others funnel more and more resources into their advanced stats departments, other teams are content with the informational gap, seemingly resting on their traditional scouting laurels. What do we know about the Wizards’ use of new stats? Next to nothing.

Wizards’ team president Ernie Grunfeld has stated that the team does not employ a “stat guy,” but does use “some services.” Let’s hope they have a Synergy Sports account, at least. Sources have indicated that Wizards’ VP of basketball administration Tommy Sheppard was in attendance at the 2010 MIT Sloan Sports Analytic Conference. What Sheppard learned from that experience remains to be seen. However, what we do know about Sheppard, presumably the guy in charge of stats: that he consults with a basketball statistician by the name of Bob Bellotti.

Bellotti got his first break as a statistician in 1988 after he sent copies of his book, Basketball’s Hidden Game: Points Created, Box Score Defense, and Other Revelations, to every NBA GM and coach. In other words, Bellotti is the NBA’s Mel Kiper, Jr. Del Harris — who was coaching the Milwaukee Bucks at the time — contacted Bellotti, and he has been working in the industry ever since, becoming the “stats guy” Grunfeld inherited while GM of the Bucks, who in turn brought the services to Bellotti to Washington.

Bellotti’s principal contribution to the field is his “Points Created” stat, which looks something like this:

Points Created =
PTS + AST * (2-VBP) + (REB + STL + BLK) * VBP -
(FGMiss + FTMiss + TOV) * VBP – 0.5 * VBP * PF

…where VBP = the value of ball possession, which Bellotti defines as the League’s average points per 100 possessions in the previous year.

By nature, I’m wary of statistics that try to boil a player’s value down to a single number. At first glance, I can think of at least two flaws with Points Created (flaws that I am sure Bellotti is aware of — see the interview below):  1) no attempt is made to ascertain the relative value of each part of the statistic, i.e. is it true that an assist is always worth more than a rebound?; 2) Points Created does not account for a player’s defense beyond counting stats: fouls, steals, and blocks. However, Points Created is good at measuring a player’s offensive value; Bellotti cites that “the NBA’s MVP has ranked in the top 4 in Points Created in 25 of the past 26 years” as indicative of this fact.

Clearly, Bellotti is an industry vet who has contributed his work to multiple teams. While he was unable to talk about his consulting with the Wizards, John Townsend’s interview below sheds a lot of light on the mind of Bob Bellotti — the man who, to some degree, the Wizards trust with their statistical analysis.

The Interview.

- by John Townsend

Thanks for the wonderful intro, Arish. Let’s jump right into it, shall we? (After all, we have the Wizards season opener to get ready for.)

On this interview:

Bellotti: “I do work for the Wizards, but I don’t want to tell you what I do for them, or how I do it. I don’t want to give that stuff away.”

Okay. Now, my questions …

(More) On the emergence of Bob Bellotti, Advanced Stats Pioneer:

Bellotti: “Back in the 1980s, I wrote a book called Basketball’s Hidden Game: Points Created, Box Score Defense, and Other Revelations, released in 1988. I published it myself, but I didn’t really have a way to sell the book when I finished writing it.  So what I did was send copies to every NBA coach and every NBA general manager. A lot of them got back to me and wrote me nice letters, but Del Harris — who was the coach of the Milwaukee Bucks at the time — actually called me and said he read the book and had liked the way I evaluated. He asked if I would evaluate all the players in the league for him. That’s how I got started and I went on to work with the Bucks for 19 years. The toughest part was gathering all of the data to use and coming up with a final number for each player. The data gathering was so tough because, at the time, computers were rudimentary and there wasn’t really an internet to speak of, so everything had to be typed in.”

On the origins of advanced basketball statistics:

Bellotti: “It started a long time ago. I’m not sure exactly how it got started, but you go back to the early parts of the game and depending on what types of records were kept at that time, some sort of very basic statistical analysis was done very early on. It really didn’t start growing until the mid- to late-80s, has really grown quite a bit over the last five to 10 years with the introduction of various elements that have pushed things along. There is a tremendous amount of information available to people. It’s a matter of having the knowledge of how to use that information to best help teams and understand how teams work and how players work with each other. There is probably still some growth in the gathering of this type of information and the ways that people can evaluate players, but it’s getting to the point where the analysis now becomes the more important thing, as opposed to the methods that we use to arrive at various conclusions.”

On what Bellotti pores over to complement film study:

Bellotti: “One of the things I try to do is look at player’s or a team’s weakness — where are they weak and how can they get stronger?  You can identify weaknesses fairly easily. The methods that can be used to do this are the linear weight-type of stats [like Win Shares and PER], and then there are things like plus/minus and using the proper efficiency ratings for offense and defense. Once you have all that information together there’s no really cohesive way to pull that all together into one number, so you kind of have to weigh each of those factors and come to a conclusion about a player or team based on that. There’s no real magical number to do that. The other important thing is watching the games and gathering information as you’re watching the games. For instance, defensive ratings. There are very few defensive ratings available these days, but it’s really half of the game. So, that’s kind of a weakness in the analysis of the sport and it’s hard to gather data on that unless you watch the games and keep track of various things.”

On how to quantify defense beyond the box score:

Bellotti: “Well, I think watching the games is the best way. You can’t really get much from a box score. The only stats you can pick out are blocked shots, steals, defensive rebounds … Personal fouls, too, can be considered defensive stats because most fouls are committed by the defense. But there’s so much more that goes into playing defense. If you would watch a game, if you would watch closely enough on the defensive end, you would find that with almost every shot that is taken by the offense, there’s one or two defensive players that are involved in trying to stop that shot — the other players on the floor may have contributed to the situation in which the player takes the shot, but the actual taking of the shot is usually just guarded by one or two people. So, if you look at it from that point of view then you could gather data that way, but it would have to be watching the games. That type of information is not generally available.”

On predicting NBA careers:

Bellotti: “All of my methods are applicable to basketball at any level, whether it high school, college, or pros. Stats for international players translate somewhat to the NBA. They key there is equating what league you’re looking at internationally and how that league compares to the NBA. I have various methods do to that. When I evaluate draft players — the draft each year for instance — I’ll take into account where different international prospects are playing, weight that league compared to the NBA, and then consequently pull the final numbers. You can easily relate an international player to an NCAA player here.”

On John Wall:

Bellotti: “We don’t know yet, do we? We don’t know whats going to happen now. I can’t really talk much about what I do with the Wizards, and part of that was helping them in that area, but he’s a great player and I’m sure he’ll be a great pro.”

On Gilbert Arenas, without talking about Gilbert Arenas, because he cannot talk about Gilbert Arenas:

Bellotti: “One of the things you have to take into consideration is a player’s salary. How much can a team sign a guy for? What’s his expected value? Is it worth it at whatever salary you expect to pay him? Or is it not? Some guys, particularly younger players, may not be very heralded. In other words, they may not have a great reputation for being productive players, but if they get on a team and are put in a certain role which they may be very good at, they can be very valuable for a team — especially because younger generally make less money.”

On why we kept Cartier Martin:

Bellotti: “You have to look at various measures, overall performance, plus/minus, and defensive play. You can put all that stuff together and say, ‘We have enough scorers on this team and this guy is a really good defender.’ It many not show up in his Points Created score, it may not show up in his PER, but given the amount of money that we are paying him, we can use this guy. This guy will be of more value than, say, bringing in a veteran who costs a lot more and can do the same things. A guy like Shane Battier is known for his ability to contribute to the team without scoring. Whenever he’s on the floor the team generally does better than when he’s not on the floor. So, if you get a guy whose plus/minus like that is high year, after year, after year … you look at a guy for five or six years and notice that whatever team he’s on, regardless of how many points he scores, he’s really a valuable player.”

On whether or not two scorers (Gilbert Arenas and Andray Blatche) are enough to carry the scoring load:

Bellotti: “I would agree with that.”

On players Bellotti’s system has pegged to do well and why Grunfeld needs to bow to the stats:

Bellotti: “The two guys that I like coming into the league this year are Gordon Hayward, Who is with Utah, Damion James, who is with the Nets. Hayward was a pretty high pick … but James was pretty far down the list in the first round and I think he’s going to be a really good player. Hayward just has a great basketball sense and he’s going to fit in very nicely in Utah’s system. He may not get a lot of minutes at first, but I think he’ll be very effective in the minutes he does get. Both Hayward and James did well in Summer League, so those my two sleepers. The guy who did that last year was DeJuan Blair, who fell into the second round and was, on a per minute basis, one of the best rebounders in the league. The methods that I use are good at spotting things like that.”

[Editor's Note/Beating A Horse On Life Support: Clearly the DeJuan Blair analysis fell upon deaf ears. Sure, selecting Jermaine Taylor with the 32nd pick instead and selling him to the Houston Rockets for $2.5 million cash was peddled by management as the right move because the 2009-10 Wizards were set to be a "veteran" team. But also know that rebounding was a glaring need, and Blair was pegged by some to be, statistically, the best rebounder in college basketball history. And sure there were concerns about Blair's knee injury history, but using a second round pick would have been well worth the risk.

Instead, Grunfeld used the money gained from Houston to sign a 34-year old Fabricio Oberto, who is now in Portland. The San Antonio Spurs nabbed DeJuan with pick 37; Blair went on to make the All-Rookie second team and now starts for the Spurs.

Finally, and surely, one could say that a lot of teams passed on Blair ... but on draft night, it was painfully obvious that the Wizards should have selected Blair. Now, you have what is perhaps one of the darkest blemishes on Grunfeld's record, worse than trading the fifth pick for Mike Miller and Randy Foye, especially when the statistical measurements of a trusted consultant prove that Blair should have been ripe for the picking. -K. Weidie]

On whether or not we should keep waiting for Nick Young to become a star(ter):

Bellotti: “The guys who do well are those who make a high percentage of their shots, get a certain percentage of rebounds per minute, and play well coming off the bench. They may not be getting a lot of minutes, but for the minutes that they’re getting, they’re really producing quite well. So there are a number of players each year who do that. Sometimes they go on to become stars; sometimes they don’t. It’s just hard to determine why a certain guy becomes a star or why he doesn’t, even though he may have started for the bulk of his career.”

On injuries:

Bellotti: “Injuries are a huge factor. You have guys on the injured list quite a bit. You look at the kid from Baylor — Ekpe Udoh. He got hurt right off the bat and he looked like he would be a really solid pro, but you just don’t know that. Blake Griffin is another guy. You don’t know what he’ll come back and play like. [Note: He looks better than never. -- Townsend] He was a cinch to be a real impact starter and perhaps an All Star player, but sat out his whole first year. Greg Oden? Same thing. Sat out his first whole year. Injuries can really mess up a player’s career. If they’re on a trajectory where they appear to be really good players, that can be kind of an ‘X-factor’. It also depends on the player who is injured, the kind of work ethic he has, how hard is he going to work in rehab to get back. You get some guys who don’t work as hard as others.”

On whether advanced stats and mathematics have changed basketball:

Bellotti: “They’ve definitely changed the way that teams look at their players. The same way that that happened for baseball. Fifty years ago, you didn’t have baseball teams looking at player stats in such detail, and now you do. That’s really changed the way teams have gone about acquiring players, trading players, drafting players and I think the same is goes for basketball. It probably hasn’t reached its peak, where it’s kind of a given for all 30 teams yet, but it’s getting pretty close. Most teams are heavily into [advanced statistics].”

On the future of the game:

Bellotti: “One of the things that I’ve heard is that in the future, there might be sensors on the court, implanted in the floor, implanted in a player’s sneaker, so that you can track where a player goes on the court and which of those patterns are typically more successful. So. I think as technology advances, the analytical side of the business will advance also.”

Many thanks to Mr. Bellotti for his time.
For more on him, check out BellottiBasketball.com.


  • Paul Storfer

    Good article/post, but I have to say that I think it is only part of the picture.

    Many years ago I was working for the Cavaliers (Before the Lebron era) and we did a competency analysis while building a model for players. It turned out to be highly predictive, as testified to by Wayne Embry, who ws the GM at the time. Interestingly enough, we found in our research that it was not the on-court performance that was most predictive of success but rather some of the behavioral traits. More on that if you have interest…..

    Best,

    Paul