Saturday, October 16th, 2010 at 6:57 pm  |  20 responses

Dean Oliver on Quantitative Analysis

Oliver talks about the role his statistical analysis plays in the NBA.

SLAM: Have you seen more intrigue from players in quantitative analysis?

DO: Since I work through the coaching staff, I don’t ask about that. If I had a player like that, I would ask what he thinks. What they have to digest is not small. There’s a mentality for scoring, for defending man-to-man, for defending the paint. Certain people can adopt a personality at times to change their normal minds. Shane, on the defensive side, he knows for him to do what his mental mindset wants to do, those scouting reports help. Maybe the next guy who wants to digest all this stuff is a completely different player from Shane. I don’t want to dismiss that this happens already, except they do it with tape. If you do that right, that can be very representative of what a player does.

SLAM: I want to go into a couple stats and what value they bring to the game. How about possession-based analysis? What value does that have?

DO: What you’re trying to do is separate out what happens in a game. So, you see the final score. Win and loss is the most coarse measure of what happens in a game. There’s the score, 129-121, like those guys scored a lot of points. The next stage from that is, whether it was because they were shooting the lights out, or because they were running a lot. Possession-based analysis is taking it to that one level beyond the score to understand whether they were running a lot or whether they were just shooting the lights out. Sometimes both teams have that kind of night.

I think that’s a theme of what I do. You’re trying to break everything down from a top level, which is win or loss, to the details. With possession-based analysis, you have the four factors (effective field goal percentage, offensive rebounding percentage, turnovers per possession, free throws made per field goal attempt). Then each of those factors can be broken down. All these things tell a better story. Or they can tell a very detailed story. It may not necessarily be the story you need to tell to make a trade, but it may be the story you need to tell your coach. You tell whatever story you need to tell to solve the problem you have.

SLAM: Good rebounders are certainly skilled in that area. But you don’t always know where and how the ball will bounce off the rim. How much do you incorporate luck when you do at rebounding analysis?

DO: It’s a little more complicated than what you said, because of course you have guys who are, by the nature of their position, in the right place at the right time more often. It’s a little bit convoluted. There is certainly some luck. There are players who’ve figured out the skill and other players who haven’t. They tend to rebound their area, but they won’t rebound outside of that. The gradation between the ones that don’t and the ones who are perfectly good at it isn’t perfectly smooth. You have guys who definitely play the same position and who are better than others.

SLAM: Do you do referee analysis?

DO: I stay away from it. I like looking at the things that we can control. Frankly, there are a lot of people who are courtside referees. A lot of fans’ favorite pastime is criticizing referees. Those guys probably get booed more than the road team. I stay away from analyzing them too much.

SLAM: What does the future hold for quantitative analysis in the NBA?

DO: I think as long as there is a demonstrable edge to it, it’s going to grow. Any divide between qualitative and quantitative people will diminish fairly quickly. As I was saying before, a lot of times those divides aren’t that real when you listen to the conversation. The danger is when you start off by saying you like a guy, and the other guy says he’s okay. Are you really on the same page? In what way?

Wall Street — this is my understanding more than my own study — for a long time was made up of people who guessed at the market. Guess is probably too strong of a word. They analyzed in whatever way they did. But that they had a feel. Then people came in with high-level math and analyzed Wall Street. You have 40, 50 years of quantification of what goes on in Wall Street, and the early adopters made a lot of money and of course a lot of people continue to make money using quantifiable methods on Wall Street.

SLAM: Could you coach in the NBA? Do you have a desire to do that?

DO: I love the coaches. I think it’s more of a coach’s game sometimes than people realize. I like being around the coaches. I like hearing their stories about what matters on their level. Certainly I like being there. What I want is I want to win a championship. If that’s working closely with a coaching staff and being part of a coaching staff, that’s fine. If it’s not, that’s fine too. I realize how basketball is played. I’ve played, I’ve coached, I’ve scouted, I’ve done all this numbers stuff and I realize I like all of it. There’s frustrating aspects of all of it, too. But if you ask me to do something and I think it’ll help us win, shoot, I’ll do it.

SLAM: How many games do you watch per night?

DO: [Laughs] Usually not more than one. Because realistically, in many ways the numbers are watching a lot of games. Depending on what I’m doing, I may be watching a college game, a minor league game, an international game or an NBA game. It depends on our time of season and our needs, that sort of thing.

  • Add a Comment
  • Share
  • RSS

Tags: , , , ,

  • http://slamonline.com Bryan Crawford

    Nice job, Kyle. But for a guy like myself who feels that the game is played on the court and not with a calculator, I’m till not convinced of the value of analytics. Especially not when Mr. Oliver stated that half the teams in the NBA “dabble” in it and 7-10 teams use it in their decision making process. That’s just more supporting evidence for me that despite what “stat guys” would have you believe, advanced stats are just not that popular. Sure, you have a community of people that support analytics, but they represent such a small percentage to the point that they’re insignificant in the grand scheme although they’re certainly loud and boisterous enough to have you think that they represent a large majority.

  • http://PSD Chronz

    Well Bryan, would it change your mind if the half that did “dabble” were among the most successful franchises? Now it doesnt take a genius to realize building around a player like Kobe can lead to a competitive squad, the true value in statistics lay in building the core around that special player or in building a team sans superstar. Take Daryl Morey for example;

    The Houston Rockets were able to field a competitive team despite being crippled by the absence of 2 MAX players, think about that, thats like the Lakers losing Kobe and Pau and still going .500. The Rockets were able to accomplish something no other team has ever done due to the fact that they were continuously able to find diamonds in the rough in the draft and free agency in large part because of their statistical markers.
    Now this can be accomplished without the use of stats but the fact of the matter is this, if we were to replace all the decisions made in the draft as a result of “intuition” with the selections a computer model projects as the superior prospect, chances are your team will be better off.

    I dont know what gives you the impression that stat geeks are “loud” or that they represent a majority movement when the article you just read contradicts that very statement.
    I understand Dean O doesnt represent every stat head out there but in my experience we are well aware of where we stand among the masses. All you have to do is visit truehoop to see how often they refer to themselves as outsiders or geeks. Still the movement is happening, you can either get with the program and optimize your decision making process with tons of useful information, or you could be like the Clippers and be left behind.

  • shu

    Did you ask how his fantasy basketball team did last year? Who’s his “sleeper pick”?

  • Nathan

    The interviewer worked but I don’t think Oliver really gave him that much back with any depth or precision. Has he even opened the door for any interns? If not, that would be disappointing.

  • http://PSD Chronz

    PS The Kobe mentioning was in no way related to the Lakers being among those teams, they are actually one of those teams that totally disregards stats from what Ive heard

  • permaculture james

    I like the section on possession based analysis (p3). Overall though, at times he seems almost evasive — not a slam really, maybe just the nature of being a quantitative guy — or maybe coy and protecting his data? Would love to watch a game with him and learn more. Hard to quantify how much the stats reveal the game or conversely, whether the true nature of the game gets obscured by the chess match of coaches in real time. For example, does someone like Phil Jackson has an ace up his sleeve that will appear during the playoffs in something as small as resting a guard a few minutes earlier in the game yet keeping total minutes the same.

  • permaculture james

    i meant p4

  • Jerome

    Playing and coaching Cal Tech was probably about the same level as a run of the mill high school team.

  • Jerome

    To get inside the league you got to be pretty good but also got to know how to pimp it. Pimp it to the right guys for the sale. The best pimps win.

  • http://slamonline.com Bryan Crawford

    @Chronz: If by “successful” you mean that half of the teams that “dabble” in analytics in some form or fashion finishes a season .500 or better, does that really equate success? No matter how you slice it, the end game in professional sports is a championship. Fans (people who buy tickets) are only going to accept the notion of, “Our team had a great season” for so long. As of a year ago, only 9 teams have an analytics department: Boston, Cleveland, Dallas, Denver, Houston, New Jersey, Oklahoma City, Portland, and San Antonio. Of those 9, only Boston and San Antonio have won championships and it can be argued that in the case of the Celtics, analytics had almost nothing to do with it. As for the Cavs, they had LeBron James. The real test (and value) of their analytics department will come this season. Will they be like Morey’s Rockets and not have their ship sink totally with their marquee and franchise guy gone? That remains to be seen. And what about the Nets? Analytics didn’t help them. They won 12 games last season. OKC’s analytics department seems to be doing a great job in constructing their team, but even with a great player like Kevin Durant, can they really win a title? And the rest of the other teams are really just middle of the pack. They have great regular seasons, but are no real threat to win a title. Not even the Spurs anymore. I just think that analytics in basketball makes mediocrity acceptable and teams that integrate it into their decision making process will be just that and the teams that rely more on traditional talent evaluation of players and assess value that way will continue to outshine them. The notion that a computer model can build a better team than good, old-fashioned basketball “intuition” is silly, IMO and I ask you, are these computer modeled teams being built to win something tangible — like an NBA Championship trophy — or just a good amount of regular season games?

  • Preston

    Most of the stat geeks were hired by already good teams. So you’ve got selection bias.

  • Tom

    I agree with you to a certain extent Bryan. I think the real value in quantitative analysis in hoops is not so much in determining who’s good and who’s not. But in determining which groups play well together, and which players would play well in a certain system. For example there are some players in the league that can only be successful in an uptempo system and if you put this in a half court system they would be far less effective. And to an extent you see that with OKC, and houston. All the players on those squads really fit the systems being used there

  • Tom

    PS…..great interview.

  • http://slamonline.com Bryan Crawford

    @Tom: Sometimes, it’s not always about the best fit that gets it done. Even though they won, was Ron Artest really a better “fit” in LA than Trevor Ariza? Sure, it works now, but are the guys surrounding Kevin Durant really championship caliber pieces? Sometimes, you have to go away from what the numbers say that works and put a squad together that you know is going to be capable of winning it all when the time comes. That’s all I’m saying. Sure, certain systems fit certain players better, but if you’re always looking for the “right fit,” your team is not going to be very good. Sometimes, you have to be unconventional if you want to compete for something other than just regular season victories and whether it suits a player or not, these guys are pros. They’ll adjust.

  • DW

    There are lots of analytic ways you can learn about “fit” and apply that knowledge. Including playoffs and clutch.

    Boston has analytic help.

    Lakers, there were rumors Rudy T’s son did some computer work for them. Not sure how much analysis.

    But their system guys for offense and defense, I’d call them analytic, general speaking, but traditional tactical analysts. Focused on plays and getting stats out of them rather than analyzing the game in bigger bites. That form of analytic certainly pays off. Other kinds can too.

  • Ronald

    @Chronz: I hate it how the writer of Truehoop calls himself a geek despite not having the remote qualifications or brainpower to be considered one. He just uses the word to indirectly call himself “smart”. At least, the one use of quantitative analysis is to prevent GM’s from handing out bloated contracts largely based on scoring alone.

  • http://PSD Chronz

    —– If by “successful” you mean that half of the teams that “dabble” in analytics in some form or fashion finishes a season .500 or better, does that really equate success?
    Yes, like I said from the start. It doesnt take a genius or any sort of intuition to know building around Kobe will lead to success. But to try and say there is only 1 successful franchise per year is nonsensical IMO. Only 1 franchise obtains the ultimate level of success sure, but there are various levels of success. I guess this comes down to how you want to evaluate success.

    —– Fans (people who buy tickets) are only going to accept the notion of, “Our team had a great season” for so long.

    Based on? People buy tickets for a variety of reasons, just look at the Clippers. Theyve been turning a profit for decades and it has nothing to do with their success so a blanket statement like this is meaningless.

    —– it can be argued that in the case of the Celtics, analytics had almost nothing to do with it.
    Not really, Rondo was a stud by draft measurements

    —– As for the Cavs, they had LeBron James. The real test (and value) of their analytics department will come this season. Will they be like Morey’s Rockets and not have their ship sink totally with their marquee and franchise guy gone? That remains to be seen. And what about the Nets? Analytics didn’t help them. They won 12 games last season.
    So let me get this straight, you can credit the success to players but all the blame gos on stats for them losing? Thats pretty poor system man. I based success on whats expected of the team and what they control. For example, if a team drafts a solid prospect late in the draft due to analytics then its made a successful decision ala Ty Lawson. I dont look at raw W-Ls because there are alot of factors that go into that.

    —– OKC’s analytics department seems to be doing a great job in constructing their team, but even with a great player like Kevin Durant, can they really win a title?
    I dont know, nor do I care. All I know is that awhile back they were nowhere near where they are now. Thats a SUCCESSFUL TURNAROUND. One that actually only happened because Portland actually went against what the stats and models said to do and went with conventional wisdom of taking big over small when every metric out there said to take Durant.

    —— And the rest of the other teams are really just middle of the pack. They have great regular seasons, but are no real threat to win a title. Not even the Spurs anymore. I just think that analytics in basketball makes mediocrity acceptable and teams that integrate it into their decision making process will be just that and the teams that rely more on traditional talent evaluation of players and assess value that way will continue to outshine them.
    I dont see a shred of proof to promote this opinion but ok

    —– The notion that a computer model can build a better team than good, old-fashioned basketball “intuition” is silly, IMO and I ask you, are these computer modeled teams being built to win something tangible — like an NBA Championship trophy — or just a good amount of regular season games?
    They are meant to be an upgrade, whether you win a title or not is dependent on so many external factors that trying to pin them down to a single aspect is laughable, that you find it silly is again irrelevant when you look at the historical results of draft rating models this decade, it is a quantifiable fact that they have done a better job of landing impact talent.

  • http://PSD Chronz

    WOW that totally didnt work out the way I envisioned, Ill just copy my retorts to your claims

    On Success:
    Yes, like I said from the start. It doesnt take a genius or any sort of intuition to know building around Kobe will lead to success. But to try and say there is only 1 successful franchise per year is nonsensical IMO. Only 1 franchise obtains the ultimate level of success sure, but there are various levels of success. I guess this comes down to how you want to evaluate success

    On Consumer Tendencies
    Based on? People buy tickets for a variety of reasons, just look at the Clippers. Theyve been turning a profit for decades and it has nothing to do with their success so a blanket statement like this is meaningless.

    On the C’s title
    Not really, Rondo was a stud by draft measurements and Perks signing screams moneyball tactics

    On your system for evaluating success
    So let me get this straight, you can credit the success to players but all the blame gos on stats for them losing? Thats pretty poor system man. I based success on whats expected of the team and what they control. For example, if a team drafts a solid prospect late in the draft due to analytics then its made a successful decision ala Ty Lawson. I dont look at raw W-Ls because there are alot of factors that go into that

    On OKC Building Plan
    I dont know, nor do I care. All I know is that awhile back they were nowhere near where they are now. Thats a SUCCESSFUL TURNAROUND. One that actually only happened because Portland actually went against what the stats and models said to do and went with conventional wisdom of taking big over small when every metric out there said to take Durant

    On Draft Models being better indicators of talent vs intuition
    They are meant to be an upgrade, whether you win a title or not is dependent on so many external factors that trying to pin them down to a single aspect is laughable, that you find it silly is again irrelevant when you look at the historical results of draft rating models this decade, it is a quantifiable fact that they have done a better job of landing impact talent.

  • http://PSD Chronz

    To Bryan again
    @Tom: Sometimes, it’s not always about the best fit that gets it done. Even though they won, was Ron Artest really a better “fit” in LA than Trevor Ariza? Sure, it works now, but are the guys surrounding Kevin Durant really championship caliber pieces? Sometimes, you have to go away from what the numbers say that works and put a squad together that you know is going to be capable of winning it all when the time comes. That’s all I’m saying. Sure, certain systems fit certain players better, but if you’re always looking for the “right fit,” your team is not going to be very good. Sometimes, you have to be unconventional if you want to compete for something other than just regular season victories and whether it suits a player or not, these guys are pros. They’ll adjust.

    Analytics are what help you decide whats the best fit, intuition isnt going to tell you how often playerA made a shot from zone B when open. Because you are incapable of watching every possession from every player, going by gut instinct instead of factual data is counter productive. Your eyes can lie just as much as the data. So there is no SINGLE WAY of building a team, simply put if you ignore the stats it will eventually catch up to you

  • http://PSD Chronz

    @Chronz: I hate it how the writer of Truehoop calls himself a geek despite not having the remote qualifications or brainpower to be considered one. He just uses the word to indirectly call himself “smart”. At least, the one use of quantitative analysis is to prevent GM’s from handing out bloated contracts largely based on scoring alone.

    This may be true but I consider myself a statgeek and at the risk of tooting my own horn and coming off just as egotistical, I hold a certain amount of respect in the blogoshphere/forums because of this openness to statistics, yet compared to the likes Dean, Kevin and Hollinger, Im a nobody.

    I dont think you should hate someone taking pride in their work, atleast when they admit their flaws, which Ive seen Henry do.

Advertisement