Interview with Roland Beech
By Kevin Pelton
Jan. 29, 2004
As long as I've been discussing basketball on the internet, I've heard fans occasionally mention how if they only had the plus-minus ratings the league tracks but rarely releases, they'd really be able to evaluate players. When it appeared before this season, 82games.com made that possible. (Yes, I know that Harvey Pollack's Statistical Yearbooks have tracked this for nearly a decade, but they don't come out until two months into the new season and provide only a raw rating.)
82games is much more than plus-minus ratings, however. Articles using its unique data have addressed questions like the value of blocked shots, which players pad their stats the most in garbage time, and which players are the most clutch.
Page 23 sat down (via e-mail) with the founder of 82games.com, Roland Beech, to ask him about how the site came about, what the reaction to it has been like, and where the site is going in the future.
Page 23: You started out with a football site, TwoMinuteWarning.com. How did that come about?
Roland Beech: I had worked for a company that provided downloadable stats for horse racing and sports (football, basketball, baseball), and realized that there were some stats-oriented things I thought would be interesting to explore, so I went out and built a group of people to watch NFL games and chart games for new detail and the site kind of emerged around that effort.
How successful has that site been?
RB: We've been very popular with certain groups of football fans -- particularly those with a gambling/office pool inclination or fantasy interest -- as well as attracting some attention from people involved with the NFL. In general though the more casual fan probably doesn't care a lot about what we do. There's much more of a heritage for looking at stats in baseball than in football, and the fan base apparently differs enough that there are fewer NFL followers wondering about how to rate offensive linemen individually, etc.
Was the transition to the NBA inevitable? Have you always been an NBA fan?
RB: I have always enjoyed the NBA, in part I suppose because it is a pretty stat-intensive sport that combines the real essence of team play like football, without being so hard to get at the individual contributions. I've followed the sport from about the late 70s, so I have seen all the Magic/Bird/Jordan evolution.
It was daunting to tackle the NBA in that the scale of things is so much different from the NFL (82 games versus 16, seven day a week activity, months of playoffs, etc), but there were certainly many fertile areas for research that interested me and so I'm finding now I'm much more excited about the NBA side of things since it feels like we can do really groundbreaking stuff whereas with the NFL unless we can get our hands on the game film there's severe limitations to how far we can get.
Can you explain the process you use to get the numbers we see on 82games.com?
RB: We look at a number of different play-by-play logs, all of which have their issues, in conjunction with our own charting to build databases. From there it's easy enough to extract the kind of information we want to evaluate, say getting at how often teams score after an offensive rebound, the difference in expected points per possession when the defense is in the penalty situation, etc.
How confident are you in the accuracy of the numbers?
RB: I feel good about the 2003-04 data, and am very comfortable drawing the kind of conclusions we make from the articles we write up. It would obviously be great to have more historical data of the same kind to understand the numbers and their implications better. Last year's stats (we've got numbers for the 2002-03 season on site too) have some known issues that we may eventually get around to fixing.
What kind of response have you gotten from fans?
RB: Much like with the football, different "types" of fans have different reactions. The stat junkies all seem to react very positively, the more casual fan may find something like the raw plus/minus stats or "clutch shooting" numbers interesting but not care for some of the more obscure stuff we do. A lot of fans get turned off if our numbers conflict with their own opinions about players, teams, etc.
I can certainly relate to that feeling. Are you aware of any NBA teams that are using your data?
RB: A lot of NBA people are starting to take notice of the site and I've talked with a number of scouts, coaches, and execs about the numbers. Indeed I am actively seeking out these people for their feedback so I can understand better what kind of statistical tools they currently use, and what kind of things they would like to have access to but currently don't.
How about the media? (Besides this column, of course.)
RB: We haven't put much effort into the PR work yet, but the media people we have talked to are generally engaged by what we are up to. In some ways I expected more enthusiasm from the beat writers, but if you look at baseball again where there's this huge legacy of statistical analysis, you still don't see a whole lot of reference to the more innovative stats work in the daily coverage.
We are looking to get deals to have our content in newspapers, on web sites, broadcasts, etc. and hope to have something going there shortly.
Put yourself in an NBA front office. How would you use your data?
RB: Well, that's a question I should be able to answer much better in another six months when a lot of the research projects have been completed and we have a better sense of the importance of various areas. I know a lot of teams do use +/- stats and look at how 5-man units perform, etc, but while those are nice, I think we will move beyond those quickly into more refined analysis where we can assess, say, optimal 5-man unit components such that you can take a specific team and quickly see what changes could have positive benefits, or how to substitute against an opposing 5-man unit.
In terms of being a GM type, I think we have a lot of stats that can be useful for evaluating players and their real value to the current team/roster situation. What we publish on the site, even at over 2500 pages of content, doesn't really display the full capabilities. For instance we can take a guy and see how the rest of the players on the roster perform in detail with that player (e.g. does a big center impact the number of times they take shots close to the hoop, does the big man get them more open looks from outside, do they end up taking more shots in the waning seconds of the shot clock due to passes inside that come back, etc). We can filter the stats both player specific, team specific, and on court/off court by almost any imaginable parameter.
So I would use the stats to understand the strengths and weaknesses of the players currently on the roster, to seek out the guys who are perhaps underrated in the rest of the league and grab them at "bargain rates", and generally try to build a stronger team. It's a thornier issue in terms of when a GM tries to influence the coach into changing player minutes, rotations, etc.
Everyone knows about the plus-minus data on 82games.com. What are some of the other features you think are underrated?
RB: Right now I put a lot of stock in the on court versus off court comparisons for each player, since I believe this is an effective way to tell the player's true "team contribution" to offense and defense -- you can find players with nice personal stats who, for whatever reason, bring the team's offense down while on court, or, more commonly, guys with some decent numbers who must be leaky on defense since the team's defense is horrible when they are out there. Then you have the guys with mundane stats who must be doing the hustle things well since they have great +/- tallies.
You seem very well-versed in NBA statistical analysis. Was that something you followed before starting 82games.com?
RB: I have tried to keep in touch with what's going on in regards to the NBA "sabermetricians" and have made it a point to get in contact with people like Dean Oliver, Bob Chaikin, John Hollinger, Bob Bellotti, etc.
I balance the perspective from these people with the things I'm hearing from people closely affiliated with teams and the league.
It does seem like there's a lot of room for creative statistical analysis with pro hoops, and I hope that 82games is going some way towards making that happen. We are very open to people approaching us to do their own studies. We have, for instance, two different college professors creating their own independent versions of the Sagarin/Winston ratings that the Mavs use.
Where do you want to take the site? What can we expect in the future?
RB: My outlook is that this is a learning season to find out what the general response is, to do tons of research to understand the numbers we have now, and in particular to see what kind of holes there are in the data teams would like to have their hands on. As we get more feedback we can integrate it into our plans going forward. A simple example of this is our clutch stats -- we initially defined the "clutch" part of a game as fourth quarter on when neither team was ahead by more than five points. Rick Barry and others pressed us to make it a tighter cutoff -- the last five minutes of a close game, and so we've made that switch. When the people who play and coach the game have strong beliefs about something, we're inclined to listen!
Then if it appears like the opportunity is there to provide some kind of valuable service, I would expect that next season (04-05) we will be charting all kinds of "never before seen" stats that get even deeper into accurately assessing teams, players and strategies. There's just so many things I can think of off the top of my head that I would like to see (say a player's offensive possessions when double teamed versus single teamed, individual one-on-one defense, energy getting up and down the court, passing patterns, etc. etc. etc.).
Again a large part of the exercise this year is figuring out what exactly we want to pull out from charting the games that isn't currently there, and so hopefully by combining the wisdom of the NBA folks, the bright minds like Dean Oliver, and some of our own ideas and experience with tracking gameplay we'll develop something special.
Kevin Pelton is an intern for the Seattle SuperSonics and is responsible for original content on Supersonics.com. He writes "Page 23" for Hoopsworld.com on a semi-regular basis. He can be reached via e-mail at email@example.com.