Audio for this transcript available
Transcription Services Provided by Verbal Ink
Russ: Welcome back to The BusinessMakers Show, heard on the radio, seen online at thebusinessmakers.com., brought to you by Comcast Business, built for business. Coming to you from the CT Bauer College of Business on the campus of the University of Houston, and my guest is Dr. Wayne Winston, clinical assistant professor of decision information sciences. Wayne, welcome to The BusinessMakers Show.
Wayne: Thanks. Great to be here, Russ.
Russ: You bet. Tell us about this information sciences.
Wayne: Well, the department I guess is made up of four different areas. We have supply chain people; for example, those are people who would try and figure out how when Proctor & Gamble makes diapers they get to Wal-Mart in the most efficient way possible. We have statisticians who deal with data. We have information systems people who deal with how computers work in business. And then we have math modelers, who try and build math models that will help businesses and organizations solve their problems more efficiently.
Russ: But very math-intensive, yet you're in the business school?
Wayne: Well, math is getting more use in business. I think any area of business math can really be helpful. And not just in business; in government, education, healthcare, just about any area of human endeavor. I mean Match.com and eHarmony try to match people up.
Russ: That's right. Absolutely. Okay. Now you got on my radar, though, for sort of a different reason too, and that's because you do go out and do business in the business world. You have this special area where you use your analytics and capability. Tell us about that.
Wayne: Well, I guess you'd call it sports analytics or the math of Moneyball. So from 1999 through 2012, I think with one year off, we worked for Mark Cuban of the Dallas Mavericks as a consultant to them on how to use math to evaluate basketball players in lineups. And the last few years we've worked for the New York Knicks in that area.
Russ: Okay. Now you mentioned Moneyball. We might have viewers that did not see the movie Moneyball, but let me try to describe it and you correct it. Moneyball was about an analytics guy in baseball that looked at the players' ability mathematically, almost only mathematically. They weren't fooled by what the media said, they weren't fooled by the person's personality; it was simply their performance. And this particular person talked the manager of the Oakland A's, right? Into - is that right?
Wayne: Right. It was Billy Beane, I believe, played by Brad Pitt in the movie. Basically and the book was popularized by Michael Lewis and I believe the screenplay was written by Aaron Sorkin of The West Wing. And so basically the idea - I think the subtitle of the book Moneyball was The Art of Winning an Unfair Game. So the Oakland A's maybe spend $50 million a year and the Yankees spend $200 million a year and by using math but not totally math the Oakland A's could pick players that would enable them to compete fairly successfully with the New York Yankees. And the idea is, I think the - Moneyball really caught on in a lot of different businesses. For instance, CVS, the drugstore chain, I believe makes every one of their analytics or math people read the book Moneyball, even if they hate baseball, 'cause the idea is how can I use data to compete better in the arena of business that I'm in.
Russ: Well, and even before then, though, there was lots of data in baseball. I mean the amount of statistics was unbelievable. This guy just figured out a different way and different measurements. Was that not right?
Wayne: Well, I think the father of I guess baseball analytics, called sabermetrics, Society of, I guess, American Baseball Researcher, it's usually credited to this guy, Bill James. There's a great 60 Minutes 20-minute documentary on Bill James, and his life is really interesting. He went to the University of Kansas; wasn't a very good student because all he cared about was baseball. And then he got a job as a night watchman, I believe at Stokely-Van Camp in Lawrence, Kansas, midnight shift. There's not much crime in Lawrence, Kansas, thankfully. So he was looking at baseball stats and he tried to figure out new ways to analyze them. And I think this is a really important lesson for your listeners; companies - organizations always have metrics. You can often have the wrong metric, and trying to do well in the wrong metric can be counterproductive. And Bill James I think was great, even though he really didn't know that much math, at just seeing what mattered and understanding how to come up with a measure that would help you get to what really matters, see the forest from the trees.
Russ: Okay. But you do that in basketball; that's your expertise, right?
Wayne: Yeah, we do that in basketball. We invented something called adjusted +/-. So like if you want to find out who the best basketball player is there's one school of thought, look at the box score. And the box score tells you what points, rebounds, assist block shots, turnovers. The problem with the box score is it's mostly offense, and basketball's half offense and half defense. So I believe the box score often misses what really great defensive players do. Dwight Howard does a lot of that. Patrick Beverley on your Rockets, his main value to the team, and hopefully he'll be back for the playoffs, or else I think you're sunk.
And so getting back to what we did. So most people analyze basketball through the box score. So my best friend from college at MIT and my colleague, Jeff Sagarin, who is at BCS Computer, gets lots of hate mail when he doesn't have a team rated highly, basically he rates teams. So I went to a Mavericks game in 1999. Mark Cuban was in my class in 1981, and so I saw him, and he said, "Do you have any way to make the Mavericks better?" So I was swimming in the pool and I said, "Well, we should rate players, not - Jeff, you rate teams." And so how do you rate players, and so the idea is what data do you need; this is a very important business.
So we needed the database that said every minute of every game who's on the court for each team, 'cause basically our theory is a good player will change the score in his team's favor, like Bill Russell, never had really great offensive stats, but he always beat Wilt Chamberlain, maybe 'cause he had better players. So we look at how the score moves when you're in and out of the game and we adjust - +/- means - a +/- of +5 means if you win by 5 points per 48 minutes your +/- is +5, but you need to adjust for who you played with and who you played against, and that's what Jeff and I invented. And it sort of shows a great player like a LeBron James or Dirk Nowitzki basically or a Kevin Durant is worth about ten points more than an average NBA player.
Russ: Okay. So you would look at all players on a team or all players in the league and just look what happened when they were in the game.
Wayne: Right. Every minute of every game. I believe there's about 33,000 substitutions during an NBA season, so the data looks like, okay, for five minutes the Rockets had their starting lineup on the court, maybe against Denver's starting lineup last night, and they won by three points. So every row of the spreadsheet looks like that, and there are 33,000 rows. And the idea is to tease out of that how the score moves - how did you contribute to the score moving, after adjusting for the nine guys on the court. So it's really a measure tied to team success rather than individual statistics. And Patrick Beverley shines there.
Russ: Okay. But did you - have you boiled it down to that? Is that the essence of what you do, in what you're contributing?
Wayne: That's really the essence of what we do. There's a lot of complex math involved in doing this right, which I probably can't talk to to your viewers- Unless they had a PhD.
Russ: Right. Okay. But I mean that's sort of interesting. You know, I mean I take it back to Moneyball again once again, and once you sort of have the formula anybody could do it from any team, couldn't they?
Wayne: I think there's something about the mathematical problem of teasing out the individual player abilities that most people who do it miss, that we don't - we think we don't miss. But I mean you're right, a lot of times an algorithm - you can boil down a process to an algorithm and then anybody can do it, and I mean do you copyright algorithms? I'm not a lawyer; I don't know. But I mean - but that's in a lot of different fields, like I teach optimization, finding the best way to do something. I'm going to teach that at 6:00 tonight to the NBA stats class here at Bauer College. And basically lots of people can solve an optimization problem; I think the key is thinking what the problem is to solve, what the data is and building the model so it incorporates the essential aspects of the business problem.
Russ: Okay. It's really interesting, though, that you figured that out in basketball, 'cause basketball is probably more difficult to assign numbers to than is baseball and football, because everybody plays offense and defense while they're in the game, and essentially everybody gets the ball, everybody shoots the ball, everybody plays defensive guy that has the ball, right?
Wayne: Well, yeah, you're right on, Russ. So when I'm going to teach - I'm going to teach, I think, a sports analytics class as a guest speaker tomorrow at Rice, and whenever I do this guest gig, and I taught this at Bauer College to the undergrad marketing grad classes last week, one of the first questions I'll ask is, "Baseball, football, and basketball, for what sport does the math work best?" And so the answer is baseball. And so most people, when I ask them why they'll say there's more data, more games. Well, that's not the point; it's an individual score, basically it's hitter versus pitcher, and so with fielders lay or not. And for your listeners, there's a great site, FanGraphs, that if you really mastered that you would know really probably 90-percent of what the general managers need to know in basically running a baseball team.
The problem in basketball is how you do depends on who you played with; if the coach makes bad lineup decisions the team will do badly. And I think the key thing we do, I mentioned the player ratings, but probably the most important thing we do is talk about how well each lineup plays and try and understand basically substitution patterns that may help the team. Like I got called by somebody from New Orleans Pelicans. God, why are they called the Pelicans? Okay. But I know why, 'cause there's pelicans in New Orleans. But actually just graduated Rice and he was doing the lineup analysis for the Pelicans, and I said, "The key thing you can do for the Pelicans is determine how do you rest your star players, like Anthony Davis." Okay. Basically the key for your team is like when you rest him can you come up with the combinations that will rest him better, because basically in the Mavericks key, in the 2011 defeat of the Heat, which nobody saw coming, although I think I did, basically was how do you rest Dirk Nowitzki. You know, it's when your star goes out you hope to break even, and it basically - so Houston, when Arden's out and Dwight Howard's out, if you guys can break even in the playoff series you should win the game. Okay, but it's not that easy to do, 'cause Harden's hard to replace, and Beverley out I think is just, maybe he'll be back with that torn cartilage, but if he's out I think that's a loss you can't make up in a playoff series.
Russ: Okay. Really interesting. Now you mentioned going to - did you say you went to school with Mark Cuban?
Wayne: I taught him at Indiana University in the Kelly School of Business in 1981. So it's an interesting story. So it was an NBA statistics class and Mark was a sophomore. And so Mark jiggered the computer system at registration so he could get in the class; I don't know why. But I mean it's - you wonder about these random things in life, like how you met your wife or your husband or things like that, 'cause I mean it's just random that you met them.
I mean, and so it was random Mark was in my class and we would have never thought about the basketball stuff. And Mark owned the best bar in town in 1981, Motley's Bar, terrific bar. And I think that's really-
Russ: And this was where?
Wayne: In Bloomington, Indiana. The home of the Breaking Away movie; I don't know if your listeners have seen that. Okay. And basically Mark owned that best bar in town; I think that's where he learned how to be a great - provide greater entertainment. 'Cause I haven't been to a Rockets game yet, 'cause usually they're playing the nights I teach, but Mark I think really has learned how to be a great entertainer and he's fantastic. I guess his greatest role now is Shark Tank.
Russ: Okay, yeah. Absolutely. So, but so you crossed paths with him in 1999. You were just happened, what, to be at a game of his?
Wayne: Yeah. Yeah.
Russ: And since you knew him you could get - you could visit with him?
Wayne: Well, this is a really interesting story. So my son's a really big Pacer fan, so - was at the time, so at Spring Break we went down to Dallas and we'd go to the Pacer/Mavericks game. We bought the tickets. But Mark roamed the stands, and so Mark saw me, he said, "Hey, Wayne how are you?" He said, "I remember, 'cause you were on Jeopardy." I was on Jeopardy in 1992; I won two games and then I got slaughtered. So his last words to me at the game - but he'd roam the stands to see what the fans wanted; almost no one ever does that. And so Mark said, his last words to me - not that he's dead - but his last words at the game were basically, "Hey, Wayne, do you have any way to make the Mavericks better? Let me know." So I worked with my friend Jeff on what we talked about. And then the next season we got a contract with Mark to supply him daily with our numbers from a website which teams can subscribe to. And so we worked with him for 11 or 12 years and then last season he wanted us back, but he wouldn't pay us enough money, so we didn't take the offer.
Russ: Well, shouldn't he now be able to do the formula that you do without you?
Wayne: No, because there's a trick in the - lots of people try and do this adjusted +/-, but I feel they miss something important. I mean they miss something that can mess up their numbers. And I mean we understood right away after the first season what it was that could mess up the numbers, and I'm positive, although there have been improvements made, nobody really understands what messes up the numbers like we do.
I mean I think Jeff and I, not that we're geniuses, but we're - Jeff calls it "little boy methods." Basically everything we do is based on high school algebra. I mean Einstein said if you can't make it simple then basically it's not worth doing.
And so basically the trick we use to sort of fix this problem is fairly simple; it's not easy to think of it. But I think Jeff and I are good at thinking of things - we're good at problem-solvers, we're good at looking at little math problems and figuring out what we need to do to get it done.
But football, I never mentioned why that's hard, because football, unlike basketball, where everybody does everything; in football you have the - here's a great question. So Peyton Manning's a great quarterback, except in the Super Bowl. Okay, and post-season. But then the question is would Peyton Manning be a great quarterback if he didn't have a good offensive line? 'Cause he's always had a great offensive line. If he had the Washington Redskins' offensive line or last season's Texans offensive line my guess is he would've been a terrible quarterback.
So here's a great unsolved problem in football analytics, how do you partition the ability of a team's passing game between the receivers, the quarterback and the offensive line, and perhaps the running? A great - for your listeners, a great site on football is AdvancedNFLStats.com.
Russ: Okay. Real interesting, for sure. So but today you're back in the business, right, and you're working for the New York Knickerbockers, right?
Wayne: Yeah, we've worked with the Knicks. Glen Grunwald, who was the general manager till he got an untimely fired at the beginning of the season, hired us for that. And so another lesson for your listeners, experimentation is important to business. If you're selling stuff like candy bars at a grocery store you want to juggle the price, you want to juggle the products on display to figure out what's going to make you the most money.
So in basketball the experimentation you can do as a team is the lineups. So you really should spend the first 30 or 40 games trying the - basically try a lot of lineups and try and get some understanding from the data of what really works. Just like the people selling at supermarkets should get an understanding of the price, display, and advertising combinations that will maximize their profit.
Russ: Right. Absolutely. So in basketball, when you're working for a basketball team, do you help a lot both in who they select and draft and trade as well as who they put in the lineup?
Wayne: Okay. But basically we are one input. Any analytics person who says they know it all is just plain wrong and very-
Wayne: Arrogant. Arrogant. Because the scouts have information that's valuable. Now on football, like should Texans go with Johnny Manziel; I love the guy, 'cause I hate Alabama, and the day he beat Alabama I was jumping up and down on my coach, saying, "This guy's my hero and my savior." But one thing I can tell you about analytics, it really has no luck predicting NFL quarterback success. The proof of that is half of all first-round draft picks that are quarterbacks are failures and they ruin the team. But I guess you have to look at which of these four quarterbacks I guess in contention for being picked has the maximum upside, 'cause you're not going to win a Super Bowl, I think these days without a great quarterback. And so Daryl Morey had a good point. What is your goal as a sports team? Is it to be competitive or is to maximize the probable you win the championship? They're different. 'Cause if you want to maximize the probable you win the championship it's like you have to take a flyer out of startup; you need something with upside, 'cause you have to be the best out of 30 teams, and so basically, no offense to Fitzpatrick, who is actually a decent quarterback, better than I thought looking at the metrics, but basically if you guys - if the Texans want to contend for the championship you need somebody in that position, I believe, has upside.
Russ: Right. Interesting. Okay, so before I let you go I've got to ask about Jeopardy.
Wayne: Yeah, sure.
Russ: So you've been on Jeopardy twice or three times?
Wayne: Okay, one of my life's ambitions - I grew up watching the show with Art Fleming, coming home from lunch, and so I basically I always wanted to be on Jeopardy. And so one of my college roommates at Yale, he got on Jeopardy, won one game. And so I kept trying. So '87 my wife and I flew out to L.A., I took the test. If your listeners want to get on Jeopardy, there's an online test, 50 questions, 15 seconds each, you need to get three-quarters of them right. So I passed the test in '87, but then you have this interview. If you pass the test I think your chances are 1 chance in 6 of getting on. So I didn't get on. So then there was a contestant search in 1992. My wife and I mailed in 300 postcards to get in the contestant search. So I got in the contestant search and basically I passed the test. I don't know what my score was, but I passed.
And then I did the interview, I thought I did a little bit better. And then a month before my eligibility was up they called me up and said, "You're going to be on Jeopardy." So I got on. You do five shows in one day; you need to bring three changes of clothes or they won't let you on, 'cause you get five minutes to change your clothes. And so I was on the third show on a Monday; they tape Monday and Tuesday. I won the third and fourth shows and I got killed. So we broke for lunch after the third show and I kept telling everybody the key to the game is the buzzer.
So what happens with the buzzer, and you don't know when you watch the show, you see people clicking and don't get called on, and the reason for that is if you buzz in too early you're locked out a tenth of a second each time you buzz. So when you don't get called on you don't know if you buzzed too early or too late. So the buzzer is really almost more of a key to the show than knowing the answers. And so basically it drove me nuts the first five questions, basically I rang and I didn't get called on, but then I did real well the first game.
But at lunch, when we sit with the people, we don't know who we're going to play against, and they don't even let you go to the bathroom in Jeopardy, they're afraid of a quiz show scandal; they follow you to the bathroom. You don't know the categories. But at the lunch the guy sitting across from me, I said, "Well, the key this show is the buzzer" and I quickly explained it, so that guy beat the crap out of me. But I mean he would've beat me anyway; he was really good.
Russ: Well, Wayne, I really appreciate you sharing your story.
Wayne: Okay, great.
Russ: Thanks a lot.
Wayne: Thanks. It was great meeting you.
Russ: You bet. All right. And that wraps up my discussion with Dr. Wayne Winston, the clinical assistant professor of decision information sciences here at the CT Bauer College of Business. And this is The BusinessMaker's Show, heard on the radio, seen online at thebusinessmakers.com, brought to you by Comcast Business, built for business.