Alexi Lalas was kind enough to sit down at the Sloan Sports Analytics Conference and give me a half hour interview. The first half of the interview is presented in this post, and focuses on the role of analytics in Lalas' own career and how they're being used at ESPN during live soccer matches. The interview was on the Friday night of the conference, which was a day ahead of the soccer analytics panel in which he participated. It was after the Thursday night Soccer Analysts meet up that he also attended. At that meet up I got about a half hour of his time, during which we talked about our backgrounds and he quizzed me on some of my statistical theories. At one point during our meal he put me on the spot regarding the odds of Arsenal winning at Anfield that Saturday. I put the odds around 20% to 30% based upon TPI numbers. Simon Gleave was also at the table, and Infostrada's Euro Club Index had the odds of an Arsenal win at 24%. Given the even play in the match and the late winner by Arsenal, we probably weren't too far off. Regardless, Alexi's direct questioning served as a good reference point to the challenges that management will make towards analysts when they claim to have the data needed to predict outcomes. It's a point Alexi and I would return to during the interview.
Herewith is the first half of my interview with Alexi Lalas for the Howler.
Zach Slaton: Since we're here at the conference, I wanted to understand what kinds of soccer analytics or statistics material are you watching or reading? Blogs, magazines, TV?
Alex Lalas: My introduction to analytics has basically occurred my post-executive career. It's been much more involved at ESPN and how I can apply it to my current job.
But I am also constantly recollecting times with different teams as a general manager when I feel it could have come in handy. That's the long way of saying that right now we have incredible statistician and analytics folks in-house at ESPN and they are constantly feeding us information. My job is to sort through all of it and make sure that I pull out the pertinent ones from an entertainment perspective. There's plenty of stats out there, whether you're on or off air, but they might not lead you to a decision.
When I put my GM hat back on and think of opportunities that either weren't there just because of the timing or I probably didn't recognize at the time were there, I think, to be fair, was probably a wasted opportunity. That's okay, that's the nature of the business and how time changes things.
ZS: Kinda like the evolution of the sport here?
AL: Without a doubt. Without a doubt...
ZS: You mention not having the stats when you were a GM, but you also have a lot of player experience. You've been competing at the top level of the game in various roles for 20-plus years now. How has player evaluation changed throughout your career? What are some of the benefits you see today, and where are some of the areas it's total overkill?
AL: When I look back at my career there's the traditional stats. You know... the 40 yeard dash, the Cooper test, all the physical fitness type tests you did. Once those were done, we'd say "soccer is a little different" because of the stops and starts, and the change of direction. So it became much more about agility and beep tests and those types of things. That's used to measure physical type of stuff, and once you get those stats back you'd say, "Okay, that guy's good at running, but can he trap the ball and cross the ball?". There wasn't a whole lot of that.
I am going to give away a lot of the stuff I am going to talk about tomorrow, but I will say that in a very rudimentary form I recognized very early how stats could be beneficial to me individually. From an early age, once I began watching myself and being able to watch myself on video, and to be able to watch games and go back and replay things visually I started keeping track. Especially as a defender, where the times that you lose the ball should be very minimal if you're doing your job. From an early point in my career I started tracking the loss of possession that I had. I'm not talking about a 50/50 tackle where I save a goal line ball and I kick it up in the stands. I'm talking about when I made a decision when I had possession of the ball to make a certain pass within that I lost the ball. I knew within each half that it it was more than one, maybe two, I knew something was wrong. Regardless of who we were playing.
When I got to a point later in my career I would co-opt rookies and have them sit on the sideline and stat it out for me. At halftime I would ask them, "How many times did I lose it?" I wouldn't know in my mind, but I would always check it. I didn't know it was called analytics, and it was a very basic statistic but I recognized the value then.
Fast forward to later in my career. It was never presented as, "Hey, this is what you're doing and this is how it's affecting you." Now was it being done behind scenes? Very minimal. Rarely if at all. We were never presented as players with a ratio. Now it's changed over the years, and I haven't played since the early 2000's.
As a GM my task was two fold. One was the product on the field, and one was the product off the field... the business off the field. Ironically, I immersed myself in the business, and that's where I came in contact with analytics. I knew everything about everyone in my front office. What they were selling. Their trends. What they had done before, what I predicted they would do next, and how that reflected their continued involvement with our team. All of that stuff. I knew trends from other teams and what they were doing. Season tickets, sponsorship, everything that was involved.
What I didn't do was recognize or obtain the ability to transfer that to the on-the-field product. When I look back I was a real young GM. I made up a lot of it and had to go through that process. When I go back I can see certain mistakes that I made where I was going to make those mistakes anyways, but it would have been nice to have that tool at my disposal. Because we all know that there's limitations to analytics. They shouldn't just be used completely. You have to use your intuition, you have to use your experience, but if I could have then brought in that part of the equation in regards to the on-the-field product I think I would have had a much more valuable opportunity to make a decision.
|I don't think this is the kind of trapping that Alexi was talking about.|
ZS: I think the point you're making is really key. There were a lot of panels this [Friday] morning on the business aspect, and because you're getting sales and you know where people are coming from it makes it a lot easier to quantify the business aspects.
The point you made regarding possession was really interesting. There are critiques where people pay way too much attention to time of possession in soccer, but what you talked about is what we often emphasize in engineering. It's really easy to design a product, but it's much more difficult to design a product that doesn't fail. To go back to the possession analogy you made, the analytics come in trying to understand how do I avoid buying and selling players who introduce certain failure modes when they're on the pitch. They end up impacting things like time of possession, which allows you to potentially have more opportunities to score. I can really relate to the self analysis that to make yourself a better player is more about not making the same mistakes over and over again. Using very basic stats, just tick marks.
AL: Yeah, but it all came about because of video. There was a whole generation before me who
didn't have that at their disposal.
ZS: Do you think video coming along at that time was critical to your self analysis?
AL: It was crucial. The VCR... talk about innovation and evolution! The ability to actually watch yourself doing something was huge in terms of the development of the team aspect and from my development. I have every VCR tape from every one of the games I played, and I would watch them obsessively to find out what I did wrong and what I did well.
We're athletes, and we tend to be revisionists as athletes. We look back and say, "I did this, and I did this because..." When you actually see it in reality... what actually happened... it's amazing to actually present that to a player and ask, "What were you doing here, and why did you do it?" For a long time it was very difficult to do with reel-to-reel tapes, but the VCR and the ability to see it changed everything. Now with the ability to isolate individual plays and to have them in rapid succession, you can tell a player "You made this move six times in this half and these are the decisions that you made." The ability to look back on this stuff is what's lead to the analytical part, where you can actually track and use the data and information as a valuable tool.
Once again, it's just a tool to then reach that ultimate decision.
ZS: Almost like you've got a tool belt with many tools, and analytics are just one of the tools.
AL: Without a doubt.
ZS: You're emphasizing the fact that you did this to improve your own game, but you can't always count on a player to do that. You also talked of confronting players who have somewhat of a revisionist mindset. We hear a lot at these conferences about how it's tough to get the GM's and get the front office to buy into it, because they've got this network of scouts and that's how they've always done it. Do you also thing there's some resistance from the players, because it can be an uncomfortable situation to view yourself?
AL: Oh, sure. Even more so in soccer. Soccer is free flowing, so it lends itself to anomalies. You don't have a start and stop. All of the variables constantly change throughout the game. I think that there is an idea, and some truth to it, that analytics aren't necessarily going to apply. Becuase there are so many different options in any given moment, as opposed to other sports there's a start and stop, you're pitching from the mound, and a bunch of other stuff that stays constant. And soccer doesn't.
Having looked back on it, I know probably as a player if someone had put these things in front of me and said, "This is why we're trading you," or "This is why we need you to do more of this, or this is where you need to work on [your game]," I think that there would be some pushback and some resentment. I still think that it exists.
Look, you guys are ultimately selling something. At a certain point you're going to have to walk into an office of a sports team and say, "This is how I can make you do your job better. This is how I can be of value to this team." You're going to have to sell it. You're going to have to get over the initial question of, "Well, who should I buy? Am I going to win this year?"
ZS: Or what's the odds of Arsenal winning at Anfield tomorrow? [we both laugh at the reference from the previous night's dinner]
AL: Right. Because they're going to want that quick fix. That's just human nature, because it challenges their background, their perspective, their experience, and their tradition that for so long has been used to make decisions. What I think everyone here has to be able to do is go in there and say, "This is just a tool. And at times it's going to confirm that instinct that you had, and make you look even smarter. At times it might get you to think twice,"
Once again, when I look back on being a GM there are times when I acted where had I had more patience and time to assess and evaluate what was going on, I might have made a different decision. Maybe that data would have enabled me to take a step back. I could have said, "I believe this, but this data maybe is pointing me in a different direction. Let's take a step back and really assess it more." When all things point to one answer that's all fine and well, but when they point different directions maybe there is a problem. Either on one side or the other.
ZS: It's never going to be a clear cut answer or decision, it's just a guide of where you might want to look.
AL: Look, you're hedging these big decisions. I think why this is crucial in MLS is that you don't have a lot of money to spend. If you make a mistake it can be incredibly detrimental. So you need all guns blazing and all the bullets in your belt, and if this is just one thing that you can spend a little money on that's going to make you have a more valuable and more informed decision it's a no brainer.
|ESPN will look to build upon their 2010 World Cup coverage via|
increased use of analytics and data-driven graphics at Euro 2012.
ZS: So I want to finish up with one more stats question before we move on to more laid back questions. You guys have the Five Aside blog, you have some great analysts at ESPN that I'm trying to meet here this week. How much are you engaging the analytics side when determining what makes it on the TV? A lot of the analytics right now are going through the web outlets, so how much of this is creeping into your guys' coverage?
AL: For example, this summer we have the European Championships coming up, and I've already received emails from our stats folks and our graphics folks on how they can make it work. From a TV perspective it's very different because you have to integrate it in with what you're covering. It has to be entertaining, it has to be graphically pleasing, and it has to be quick. You can't take a long time to explain it. God forbid it is confusing to the viewer. I'm not saying the viewer is dumb. I'm just saying when you start talking some of these numbers it can at times get heavy, and at times it can get confusing. So the ability to distill it down to its most basic forms so that everyone can understand it, it's quick, and we can put it up on TV. That's the trick.
That's why time of possession is so easy to put up on TV. But what's time of possession? Italy had incredible time of possession [against the US] the other day, and they lost the game.
ZS: Arsenal has this. They just back-pass and side-pass until the cows come home and have high time of possession, but it doesn't matter.
AL: Exactly. If we can start putting that time of possession in context and say, "This is where they had possession. This is where they found joy." And heat maps. That starts to become more valuable, especially for me. When I come back at halftime, I want to be able to give people something to look forward to. "This is what I would like you to look forward to. This is what happened good or bad in the first half, and this is why this change or that change may affect the second half."
ZS: So you guys are crunching numbers on the first half. You might have your stats guys running numbers as the first half is going along, and then you pick a couple of key things to highlight at half time.
AL: Sometimes they'll ask me ahead of time, "What do you want to see?" The other day I wanted to see how many times [Andrea] Pirlo gets the ball. A lot of times he was literally taking the ball off the foot of the centerback. So where's his starting position in a heat map? But we all know he's far more dangerous when he's another ten yards ahead.
These are the types of things at halftime that I am going to say. "Yes, Pirlo had incredible possession, but if you look at this heat map this is where he's getting that possession. If I'm in the locker room, I'm going to say, 'You know what? This is where he's getting it, but I already know he's much more effective if he's getting it ten yards up the field.' So I am going to say, 'Andrea, when you get that ball make sure you're another ten yards upfield.' And I am going to talk to my centerbacks and say, 'Don't give him that ball at his feet for three yards. Force him to get that ten yards, because it's going to be better for him and better for our team.' "
ZS: So it will be something to look forward to on Euro 2012 coverage?
AL: It's up to me to work with those guys and integrate it. They're incredible, the data that they have. We only hit the tip of the iceberg when it comes to the stuff actually on the air. We should try and do a better job without forcing it.
ZS: Is there a way to roll it out supplementary, like things you don't get to? Kind of like Bill Maher, who does an "Overtime" segment. Is their any value in that for you guys?
AL: There might be. I don't know if people are ultimately interested in it once the score is there. How many people want to dive down and find out why?
ZS: You guys have people like [Michael Cox of] Zonal Marking that are writing these great things for the website. I don't know what the production value is versus the cost to produce it.
AL: Exactly. I don't know either...
So the first half to the interview, which was focused on analytics, was finished. The second half focused on a few US Men's National Team and MLS questions, and finished with a few light hearted questions about hair products and ginger traits. I'll post the second half of the interview later this week.