Pod Insights StarTalk Podcast #ICYMI - Moneyball 2.0 – Soccer Edition About this episode In case you missed this episode on the Playing with Science channel… Hosts Gary O’Reilly and Chuck Nice enter the world of big-data analytics and its rising influence on the game of soccer on the world stage alongside Dan Altman from North Yard Analytics and Howard Hamilton from Soccermetrics.
Image Credit: Soccermetrics, with data by DataFactory LA.
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:09 Gary O'Reilly? I'm Gary O'Reilly.
0:10 Chuck Nice? And I'm Chuck Nice.
0:11 Gary O'Reilly? And this is Playing with Science. Today's a show about football, or soccer if you prefer, a look not through the eyes of a player or a fan, but through the cascading waterfall of zeros and ones that data analysts love to hide behind.
0:27 Chuck Nice? Yes, and soccer, or football if you prefer, has all been considered Moneyball 2.0, a sport at the elite level that's ripe for a data dump and seismic shift into thinking on how all sport is viewed, conceived and enjoyed.
0:43 Gary O'Reilly? Yes, and joining us to pull back the curtain and reveal all is one of, if not one, of the world's leading experts, analytics, Dan Altman, the founder of North Yard Analytics, a man whose opinion and thoughts are sought far and wide.
1:01 Chuck Nice? Yes, and later we'll also hear from Dr. Howard Hamilton, the founder and CEO of Soccer Metrics Research.
1:07 Gary O'Reilly? Yes, but first, Dan Altman. Dan, welcome to Playing with Science.
1:11 Speaker 02 Thanks very much, great to be here.
1:13 Gary O'Reilly? Thank you. Well, let's so as people listening know exactly why you're here. Sports analyst and professor of economics at NYU Stern School of Business, the list goes on. Author of Outrageous Fortunes: The Twelve Surprising Trends That Will Reshape the Global Economy, fascinating, I think I might need that book.
1:31 Chuck Nice? Oh yeah, we all need that book.
1:33 Gary O'Reilly? Founder of North Yard Analytics and leading sports data consultancy. So have I done you justice, sir?
1:38 Speaker 02 Yeah, you've even mentioned things from my former life as an economist, so more than justice.
1:43 Gary O'Reilly? Good man, all right, so here we go. Let's take a little trip back down memory lane. When did data analytics, and I use this pun intentionally, kick in for football and who drove it?
1:55 Chuck Nice? Oh, I see what you did there.
1:56 Gary O'Reilly? Thank you very much.
1:57 Speaker 02 Well, some people would argue that it was Charles Reap after the Second World War, who was standing on the sidelines out in the cold, taking things down on a notepad, who really got things going. But I don't think that it had the same sort of traction that it did, let's say, in baseball a few decades ago with the emergence of Bill James and his acolytes. It's really just been in the past decade, I would say, that sports analytics as we know it in other sports, especially in the United States, has started to percolate into football in the great reaches of Europe and possibly even South America.
2:32 Gary O'Reilly? What is the typical kind of data? I mean, we know, we can all look at who won the game because of the number of goals scored. But what is the typical type of data that is being used and who actually is capturing this?
2:46 Speaker 02 So there are two major types of data. One is match data that comes from the events that happen on the ball. And you can get several companies that will provide you with a feed live or after the match of up to, let's say, 2,000 events that occur, every pass, every tackle, every shot. And they'll tell you a little bit about where it happened as well as when it happened and some specifics of how it happened, but really just on-the-ball data. So with that kind of data, you know the narrative of the match, but you don't know what all 22 players and referees and the ball were doing at every moment. For that, you need tracking data, which is provided by another set of companies quite often. And they will be tracking every moving object on the pitch for, let's say, 30 frames per
3:31 Speaker 02 second. So that generates a huge amount of data just for one match, more perhaps for one match than you would get for a whole season of the event data. And the key is to try and meld those things together so that you know what everybody's doing and you know the context as well.
3:47 Chuck Nice? Wow, that sounds insane. I mean, you know, when you talk about that much data, which by the way, could only exist in this day and age, if you think about it. I mean, you need a huge amount of computing power to make that happen. But let me ask you this, how is it used? I mean, how can you even make heads or tails of that much data? And does it do anything for player evaluation, game strategy? How is it used most commonly?
4:17 Speaker 02 There's a ton of things that you can do with this kind of data. And I would say that even as the things we can do and implement within a soccer club are growing, because there's greater understanding of these things among the coaching staff, the scouts, et cetera, the amount of things that we can do is growing even faster. So the gap between what we can implement and what we can do is actually widening, even as we get better at using the data, which is a weird thing. But we use most commonly the match data for something like scouting, because we can get that same type of match data from dozens of leagues around the world. The tracking data are much more difficult to come by, and they're often recorded in different formats in different leagues.
4:59 Speaker 02 So for scouting, if you want to look at everybody using the same benchmarks all around the world, a player from Brazil or a player from Croatia, you're really confined to using the match data. When it comes to tracking data, you're much more often trying to examine your own players or your own team's performance, because you are collecting that data inside your own stadium. Sometimes you might be able to get it from other clubs in your league, but it's tougher to get it further afield.
5:23 Gary O'Reilly? So if, as we've said, goal scored, the ultimate stat in any data collection for a football game, what are the main areas you find coaches, players, rather than the fans outside, and I want to exclude fans, because there's no professional football without fans, what do you find are the most important metrics that people are always focusing on?
5:49 Speaker 02 Well, it's interesting, because there's just a huge amount of diversity in what the coaches want to hear. With one club I was working with, I went from an assistant coach, because typically I'll be working with the assistant coach on a day-to-day basis, you know, while the manager or the head coach is dealing with the players. An assistant coach, he just wanted to see a plot of all the goals scored, let's say, from corners from the whole league, and I didn't know really what that was meant to describe.
6:13 Gary O'Reilly? Oh, I think I do.
6:14 Speaker 02 That's what we were going to do. Yeah, but I mean, you know, we're not playing the whole league every weekend, we're playing a specific club, but that's what he wanted to see. And then the next guy who came in, because as you know, in European football, the managers and coaches can change quite frequently. The next guy who came in said... I want to see the expected goal networks for every club that we're going to play based on the opponents that were most similar to us in playing style measured by these technical attributes.
6:38 Gary O'Reilly? You've got to be kidding me, that was a request? On a Wednesday.
6:41 Speaker 02 That was a request, yeah.
6:42 Gary O'Reilly? Only, only on a Wednesday. Only on a Wednesday. Yeah. What's interesting me, the assistant coach who wanted all of the goals scored directly from a corner kick.
6:49 Speaker 02 Yeah.
6:51 Gary O'Reilly? My imagination, because I used to be a defender, I'm now thinking, how do I set up to negate vulnerability from the broad spectrum of what I'd be looking to face? I would then be looking on a game by game basis at the opponent and how they set themselves up for set plays and corner kicks, but just as an overarching, should I be going man to man, should I be playing zonal, or should I go hybrid to best solve the most things that I would be facing? That, just when you said that, my little cobweb.
7:27 Speaker 02 So I think that's, that's a totally valid point for two situations. One is when you're trying to set up for the whole season, you know, how should you be training on a day to day basis?
7:35 Gary O'Reilly? Yeah.
7:36 Speaker 02 The second is for scouting, if I'm looking for a centre back, I want the kind of guy who's going to be well placed, well positioned to deal with those threats. But this particular coach wanted to see that chart every week and you know, you're only adding a couple of goals to it every week, so it would be pretty much the same chart every week.
7:53 Gary O'Reilly? Right, exactly. Yeah, in which case I've got no thought to offer on the back of it. Every week indeed.
8:02 Gary O'Reilly? So how, how does it, how productive is this data avalanche when it comes to analyzing individual players and then you say, right, this is our player or this group of players are our intended targets? How successful are they with A, identification and then B, bringing them to say, I'm guessing a European league and then being a success within the clubs that you've worked with?
8:32 Speaker 02 Well, I'll give you a prime example. I worked until recently with DC United, which is part of an ownership group that I was serving as head of strategy for. And we had a signing that made some news over the summer. A guy named Wayne Rooney, England's all-time leading scorer, came over after spending a season at his original hometown club Leicester, sorry, Everton, I should say, another blue team, Everton. And before that, he was with Manchester United for many years. And, you know, we were going to spend a huge amount of money for an MLS team to bring him to the United States. And we were going to just get one player, rather than signing, let's say, a few young prospects from Latin America or
9:17 Speaker 02 something. So it was a risky proposition in some senses. And of course, we wanted to be able to predict how he was going to perform in his new league. Well, one of the things that I've worked on for many years is creating league adjustments, so that we can look at a player's performance in one league and try and predict how well he'll play in another league. And we looked at him at a couple different positions. And as far as his attacking output, he was exactly where we thought he would be coming to MLS. And that prediction was based on thousands of matches, dozens of players who've moved not just between MLS and the Premier League, but also at one remove. You know, a player
9:58 Speaker 02 who might have moved from, let's say, France to the Premier League to MLS. Well, then, now you have a link from France all the way to MLS, right?
10:08 Gary O'Reilly? Yeah.
10:09 Speaker 02 So we can make lots of connections to try and see how players might perform if they go from one league to another. And in this case, it was spot on. And as you saw, there was good reason to sign him.
10:19 Gary O'Reilly? Wow. If you're not familiar with the Wayne Rooney story, Chuck, and I won't give the whole... Born to now, just if you look at his time at DC United in MLS, a big fanfare, a major coup for the club. So congratulations to DC United for that. They go into a brand new stadium, Audi Stadium in DC, and the question mark remains against Wayne Rooney. Have you bought into him after the end, after all the good stuff has gone? What I'm interested in is what metrics you used, if you can give any of your IP away, or give us an understanding of what you would focus on that made you believe this guy still had enough left in
11:03 Gary O'Reilly? the tank to get people off their seats and into the stadium?
11:08 Speaker 02 So there are a lot of different statistics that get bandied around these days in soccer. But I preferred, and I started out this way probably because of my background as an economist, I preferred to create a holistic model of the game. So I have one model of winning, actually two models that I use side by side. And every action gets a value within that model. And so we try to see if we combine all the actions that a player's going to contribute, how are they going to stack up? What's their overall contribution to winning going to be?
11:42 Gary O'Reilly? And does that, does that express itself in a numeric value? Is that what happens?
11:46 Speaker 02 Yes.
11:46 Gary O'Reilly? Okay, go ahead.
11:47 Speaker 02 Yes. And then we adjust that numeric value based on the league where we think they're going to play. So, you know, Rooney, in the Premier League, one of the toughest leagues in the world, if not the toughest, was going to come to MLS, which, as Sean Harvey, who's the chief executive of the English Football League, said at a conference I attended two days ago, is probably somewhere between the second and third tiers in England. So he's not just going from the first tier in England to the second, but he's maybe going from the bottom of, to the bottom of the second, or maybe the top of the third tier. It's quite a big jump.
12:22 Gary O'Reilly? Did you feel that assessment was slightly harsh?
12:25 Speaker 02 Actually, my statistics agree with it.
12:29 Gary O'Reilly? Okay, alright.
12:29 Speaker 02 I didn't think so. And that actually brings up an interesting point. You know, we could see that Rooney was going to be making this big jump. And so even as a guy who looked past his prime, perhaps in the Premier League, he was going to kill it in MLS. But, you know, that adjustment used to be smaller. It used to be that MLS was sort of closer to mid-table Championship, which is the second tier in England. And you've got to ask yourself, well, MLS has been getting better. Why is the gap between MLS and the English leagues growing? And it's because the English leagues are just getting better faster.
13:03 Gary O'Reilly? Yeah.
13:03 Speaker 02 I mean, there's so much money in the Championship right now through the parachute payments from the Premier League and all of these owners pouring money into their clubs to try and get to the Premier League, that the quality is just going up really.
13:15 Gary O'Reilly? Dan, I'm just going to explain something for Chuck. So as this gets quantified with a dollar sign in front of it.
13:20 Chuck Nice? Yeah.
13:21 Gary O'Reilly? If you are relegated, so finishing the bottom three, and Dan knows this, in the Premier League.
13:27 Chuck Nice? Right.
13:28 Gary O'Reilly? And you get relegated into what is now the Championship, the second tier of English soccer, football. You will, you will earn more money than the team that wins the UEFA Champions League.
13:40 Chuck Nice? Wow.
13:40 Gary O'Reilly? Fact.
13:41 Chuck Nice? So you're in the bottom of...
13:42 Gary O'Reilly? Yeah, you're relegated, but you will have earned more television money than the winner of the Champions League. That's why this acceleration with English soccer is so, so quick.
13:58 Chuck Nice? Wow.
13:58 Gary O'Reilly? Am I, I mean, I'm in the right place without giving too many numbers? Uh, that is the principle behind this. Is that right, Dan?
14:05 Speaker 02 Yeah, you know, the thing is that the Premier League keeps growing mostly because of the foreign business that they're doing. Uh, the, the price of the TV rights for the UK has sort of topped out.
14:15 Gary O'Reilly? Yeah.
14:15 Speaker 02 But the price of the foreign broadcast rights keeps growing.
14:18 Chuck Nice? Yeah, I mean, they're, they're here, they're, they're on Saturday morning here, that's for sure.
14:21 Gary O'Reilly? I mean, it is the most valuable sporting property on the planet.
14:25 Chuck Nice? Wow.
14:26 Gary O'Reilly? Without a doubt. What I'm, what I'm interested in, I've got to focus on Wayne Rooney, not because it's Wayne Rooney, but how... You quantify, because Wayne has something you can't, to my mind, and I'm up for being taken to school for this one and learning, how do you quantify his mental attitude towards the game of football?
14:46 Speaker 02 Great question. So the two models that I told you about already are what I call mechanistic models. As I said, they assign a value to every action, and we can observe each action and we attach a value to it. I have another model, which I would call an agnostic model, which is sort of like a plus minus in hockey, or we have adjusted plus minus in the NBA, where you're looking at the overall contribution, but you're not saying how the athlete does it. You know, the patron saint of this in the NBA was a guy named Shane Battier. And there was a big article by Michael Lewis in the New York Times Magazine saying, why do Battier's teams win, even though he doesn't score 30 points?
15:25 Gary O'Reilly? Yes. So why is that? I read that article, but go ahead. I want you to say.
15:30 Speaker 02 So he makes these intangible contributions. He does things that we don't necessarily track in this mechanistic data.
15:37 Gary O'Reilly? Right.
15:38 Speaker 02 And so I have a very complicated formula based on something called a Shapley value, which comes from a Nobel laureate Lloyd Shapley in the 60s, and try to figure out this overall contribution, and we don't know how we do it. If you take that overall contribution and subtract the mechanistic contribution, what's left are the intangibles, and that's how we measure them.
15:58 Gary O'Reilly? Interesting. So you're able to assign a value to something that is intangible itself?
16:06 Speaker 02 Yeah. You look at the whole, and you take away the stuff you can measure, and what's left is...
16:11 Gary O'Reilly? That's frickin' fascinating, dude. I can't, that is amazing. So, well, back in the day, and it's way back, right? And we're gonna take a break, and just before we do, I'll throw this in. We did all of that by standing or sitting in a seat, watching. We didn't have numbers, we didn't have any laptops, we would know that as a fact. Right, I'm gonna leave you with that.
16:35 Chuck Nice? Yeah.
16:36 Gary O'Reilly? We are gonna take a break, and are you gonna have to hold that thought? Hold that thought, right?
16:40 Chuck Nice? Oh, it's killing me!
16:41 Gary O'Reilly? Hopefully not. No, right, we're gonna take a break. More from Dan Altman, who, I'll be honest with you, this is as close as I get to nerding out. It's fabulous, I love it, and we'll be back very, very shortly.
16:56 Gary O'Reilly? Welcome back to Playing with Science. While we were away Chuck's head exploded.
17:01 Chuck Nice? Yeah.
17:01 Gary O'Reilly? But we've managed to bring it back together again. There'll be some pieces missing. I'll look under the table. Right, we are in the company of Dan Altman, who is just outrageously thoughtful on the game of football, aka soccer, and how you analyze it, not as a coach, but as a data analyst.
17:22 Chuck Nice? As a data analyst.
17:23 Gary O'Reilly? How you see things numerically, how you can quantify what is sometimes to people unquantifiable.
17:29 Chuck Nice? Yes.
17:29 Gary O'Reilly? And bring it to the table Chuck. Let's go again.
17:33 Chuck Nice? So I was losing my mind because I was like, you know, what Dan has done is just brilliant. You know, and what I call it is filling the negative space, right Dan? You're filling the negative space. And, you know, so what you said Gary was, we used to stand on the sideline and basically you were making judgments.
17:50 Gary O'Reilly? Yeah.
17:50 Chuck Nice? So now here's what I want to say. So what you've done essentially is, and here's what I'm going to ask you Dan, because this is your model, so you're going to have to be really honest here. With judgments, there are always biases that must be overcome. When you, Dan, build a model, and not just you, I'm talking the collective you, all of you analysts, when you build a model, how do you not build in the biases to that model so that the model itself is kind of mimicking the judgments that may be erroneous, like what Gary was talking about?
18:29 Gary O'Reilly? Yeah.
18:30 Speaker 02 Yeah, so there are always biases for a programmer or whoever's designing a model, mathematician. With one of these agnostic models, we try to cut out those biases because we don't actually say what's important and what's not. We let the model tell us.
18:43 Gary O'Reilly? Okay.
18:44 Speaker 02 But, and machine learning is really involved in that whole theme as well. But what I do is really complementary to the guy standing on the sideline.
18:52 Gary O'Reilly? Yeah.
18:52 Speaker 02 It's not a substitute.
18:53 Gary O'Reilly? Right.
18:54 Speaker 02 And you're much more powerful when you get both of those working together. Now, both of those are going to be wrong sometimes. I mean, coaches and scouts don't like to admit it, but they are wrong sometimes too.
19:04 Gary O'Reilly? Oh yeah.
19:04 Speaker 02 It's not just the statistical models, which are always going to be wrong sometimes. The key is to be wrong in different ways. Because if you're wrong in different ways, when you put those two things together, you can come up with a really good decision algorithm. I was working with a club in the Premier League and that's exactly what we did. We had a filter based on all of these metrics that we had coming out of the models. And then when we got a list of players from that filter, we would have a video analyst look at them. If the video analyst thought the player was promising, then we'd have a human scout go to a game and watch the players. And our hit rate going down from that original filter was about 70%, which is great and it gives us plenty of options as recruiters. But when I was working with a team in MLS, we took it even a step further.
19:47 Speaker 02 And we said, you know, we want to make sure that we're going to bring in players from this filter that the coach is going to like right away. So we gave the coach a questionnaire and we said, here are all the attributes that we're measuring. Tell us how important they are, either very important, somewhat important, or not at all important. And if he said very important, we gave 100% weight to that. If he said somewhat important, we gave 50%. If he said not important, we gave it zero. And then we had a new formula to look for these players in leagues around the world. Then the hit rate started hitting the roof because we would show him guys that were spotted by this bespoke metric and he loved them.
20:22 Chuck Nice? I love that bespoke metrics.
20:23 Gary O'Reilly? Yes.
20:25 Gary O'Reilly? Can you tell us which the Premier League and MLS club is or would that get you a rap across the knuckles?
20:30 Speaker 02 I have some non-disclosure agreements.
20:32 Gary O'Reilly? I thought I might, I might have guessed they were there, but you don't mind me asking. So the thing is, I spent four days in Turin talking with the good people of Juventus. This was back in 98. And I remember sitting in there and you all know who this is, Roberto Bettiga, the Silver Fox, legendary Italian striker. And I remember sitting in his office with the translator and going through, how do you approach your transfers, your signings? And he said, we do not.
21:06 Gary O'Reilly? Take the coach knocking on the door and say, I need a striker. I want a goalkeeper. I need a winger. I need a midfield player, holding, attacking, whichever it is. What we do is we sit down as a group and then we decide the thoughts of the group. And the underlying message is this signing has to be of benefit to Juventus Football Club. What we bring in has to have a value while it's here, but then be of value to us with a plus plus once it leaves. And there will always be something that comes in that breaks that up. But invariably, that was their model.
21:46 Chuck Nice? Interesting.
21:46 Gary O'Reilly? It was a very interesting thought process. Is that thought process in play with more clubs now? Or, as you said, does the weight land in different areas because of the different thinking that's in place?
22:01 Speaker 02 If you had a process like that Juventus process and you didn't have any data analysts, you would still be ahead of a lot of clubs that do have data analysts and no process.
22:11 Chuck Nice? Wow, wow.
22:13 Speaker 02 Because there are some clubs that don't have regular recruiting meetings, not weekly, not monthly. They don't have a recruiting committee. They don't have this filtration via, you know, video analysts and scouts. A coach can come in, you know, a couple of weeks before the transfer window opens, put down a list maybe of two names or three names and say, I want these instead of alternatives of 10 at each position. And this happens even at clubs at the highest level. And if you do that, you're just taking on a huge amount of risk.
22:44 Gary O'Reilly? Dan, you just broke my heart.
22:45 Chuck Nice? Yeah.
22:46 Gary O'Reilly? We were talking about progress and a wonderment in bringing your thinking and your like in that part of the game to my passion for football. And then when you tell me there are still clubs anchored into way back into the 20th century, it's heartbreaking to hear that, but part of me is not surprised.
23:09 Chuck Nice? They'll change when they lose. That's how it works.
23:11 Gary O'Reilly? All right. So let's, I mean, apart from the recruitment side of it, because it does have such a presence as we've, as we've highlighted and most, most obviously with someone like Wayne Rooney, which our American listeners will equate to because of his standout performances this season. What other areas can you take?
23:33 Gary O'Reilly? The area of analytics too, for for instance, our team, you're on our team and we have this group of players, but we we think we can get more of them out of them. But how? What areas do you feel data analytics can move into that would allow us to achieve even more?
23:52 Speaker 02 Well, tactics is a big area and every club I've worked with, we've done pre-match opposition analysis and we do post-match analysis as well to see what's gone right and what's gone wrong. The pre-match packages can be very comprehensive, they can be tailored to each opponent as well. You know, we could look at how their back four distribute the ball, we could look at who the most important nodes in their passing networks are, we can look at how they set up for corners, where the completions are, we can map all of their set pieces. You know, we could, we could do a bunch of stuff like that. But we can do it for individual players as well. For example, the club I worked with recently, Swansea City, sent Jordan Iu, a Ghanaian striker on loan to Crystal Palace, one of your former clubs, to gear it this season.
24:33 Gary O'Reilly? Thank you, yes.
24:34 Speaker 02 And it was obviously a benefit to both clubs to have him succeed, right? Because Swansea wanted to see his value increase and Palace obviously wanted him to do well. So I gave their ownership an owner's manual for Jordan Iu to show which situations he'd been most successful in, where he'd created the highest quality chances, where he had the best quality touches. And there was a really coherent narrative about how he moved across the pitch and what his options were at each point along the pitch and what other players needed to be in different positions for him to have the maximum effectiveness.
25:08 Gary O'Reilly? That's amazing.
25:08 Speaker 02 And I think that can be a very powerful tool.
25:10 Gary O'Reilly? It's interesting because Crystal Palace have American co-owners who own sports franchises here in the USA. So they're more than aware of this type of thinking.
25:20 Chuck Nice? Makes sense.
25:20 Gary O'Reilly? And it's interesting, you just gave them the roadmap to get more out of one of their players.
25:25 Chuck Nice? And that is so granular that I'm wondering, is it, will we see at some point, right now you have teams as your client, will there come a point where your clients are actually agents who are able to take your information on that granular level and increase the value of the players when they go into negotiations? Are we going to see that happening?
25:52 Gary O'Reilly? Interesting.
25:53 Speaker 02 Yeah, I think we certainly will. It's now the case that there are enough data providers that agents can get access to the same quality of data as the clubs do. And we saw something like this happen in hockey. I remember a couple of years ago, there was a big contract negotiation for a player, I believe it was with the Toronto Maple Leafs. And, you know, the club knew that this was their guy, this was the guy they wanted, they want, they had done all the analytics, and then the agent comes in with a huge binder with all the same numbers. He knew just how good his player was, and he wasn't gonna take a dollar less.
26:26 Gary O'Reilly? That's the guy I want as my agent.
26:27 Chuck Nice? Exactly.
26:28 Gary O'Reilly? I mean, okay, we're talking about them and they're not in the room, so let's invite them in, the players. Some do, some don't, am I right? Engage in the analytical process. Some will tell you they've watched it, because the analysts will provide a video package for, say, a fullback who has a, all right, we're gonna play Paris Saint-Germain, so he's a one-on-one with Neymar Jr. So they'll produce a package on Neymar, tracking and mapping. Not that if you're gonna play against Neymar, you are fully focused, because if you're not, you're going to get smashed to pieces.
27:09 Gary O'Reilly? Oh, by the way, but you take that sort of one-on-one confrontation in a game. How many players engage and how many players just don't care?
27:19 Speaker 02 I mean, there's a huge variety. We saw it in the NFL recently, there was a player to whom the team started sending blank tapes to see if he was really watching the video. He would say, "Oh yeah, I watched those blitzes, yeah."
27:31 Gary O'Reilly? That's awesome.
27:33 Speaker 02 They found that out pretty quick. But, you know, there's a variety. I think that it depends on the culture and it depends on the players that you're gonna bring in. It starts with the head coach and filters on down from there. At Bolton, for example, when Sam Allardyce was the coach, they had an incredibly intense data culture. They made the players buy into it. They made sure the players they brought in would. They got down to, in terms of granularity, Chuck, they got down to where the players were sitting in the video room during the sessions and then testing their comprehension.
28:02 Chuck Nice? Unbelievable. Unbelievable, man. That's...
28:05 Gary O'Reilly? It's there.
28:05 Chuck Nice? It's so crazy.
28:06 Gary O'Reilly? When someone, I mean, Big Sam has a reputation, and that's the reputation. Some people love him, some people don't, but what he did was embrace... This thinking and in and and the more outlier type of thoughts back in the mid nineties that no one else wanted to know about, you know, what did what did he say about, oh, no, man, that's voodoo.
28:28 Chuck Nice? Yeah.
28:29 Gary O'Reilly? Wasn't that what Sugar Ray Leonard said to us?
28:31 Chuck Nice? Sugar Ray Leonard said it was back then, that was voodoo.
28:33 Gary O'Reilly? All the compute, all this sort of thing that we're discussing. Whoa, voodoo stuff, man. He wasn't having it. And it's interesting, that kind of echo still vibrates today and I'd be surprised. All right, here's a question. Do you think there's room for a data poor player in elite football?
28:50 Speaker 02 A data poor player, do you say?
28:52 Gary O'Reilly? Correct, yes.
28:53 Speaker 02 One who doesn't pay attention?
28:55 Gary O'Reilly? Mm hmm.
28:57 Speaker 02 I mean, if if you're a player who's talented and you work very hard, you shouldn't need to pay attention because the coaches are going to take the insights on board and then convey them to you.
29:06 Chuck Nice? Right.
29:07 Speaker 02 They, you know, there are some players who are going to be able to consume that stuff directly and there are some players who may need an interpreter, just like there are some coaches who may need an interpreter.
29:15 Chuck Nice? Gotcha.
29:15 Speaker 02 And and and that's fine because if they still have ability, you need to be able to incorporate that.
29:21 Chuck Nice? Now, how about betting?
29:22 Gary O'Reilly? I'm I'm it just. You you can't help but monetarize this, can you?
29:26 Chuck Nice? Yeah.
29:26 Gary O'Reilly? All right.
29:27 Chuck Nice? It's got it's got it's like, what do you do? And can predictive analytics actually help the better? And then are you going to have guys that are just getting into this so that they can lay money on the line and perhaps make some dough?
29:40 Gary O'Reilly? Already have.
29:41 Chuck Nice? Yeah.
29:41 Gary O'Reilly? Really?
29:42 Speaker 02 There are, there are plenty of guys who have, yeah. And and what's interesting is some of those guys have gone on to buy clubs as well.
29:47 Chuck Nice? Get out.
29:49 Speaker 02 Yeah, Matthew Benham at Brentford and Matillan and Tony Bloom at Brighton, another one of your former clubs.
29:54 Gary O'Reilly? Yeah, I know them, I know them.
29:56 Speaker 02 These, these are guys who made their money in betting on football and sometimes other sports and now they're running their clubs with an analytics bent.
30:05 Chuck Nice? I've wasted my life.
30:07 Gary O'Reilly? No, you haven't.
30:07 Chuck Nice? I have wasted my life.
30:10 Gary O'Reilly? No, you haven't.
30:12 Speaker 02 That's what I'm actually looking to do too. I have a group of investors I put together and we want to buy a club because we think one of the best ways to monetize my analytics platform is to take a club from a lower division in Europe and get it promoted to a higher division where it's going to have a lot more value.
30:27 Chuck Nice? Well, something tells me you're going to be damn successful, sir.
30:31 Gary O'Reilly? I'll tell you what, Dan, we must have another conversation.
30:35 Chuck Nice? Yeah, man.
30:36 Gary O'Reilly? That won't be recorded.
30:37 Chuck Nice? I'd love to have that conversation.
30:41 Gary O'Reilly? It has been a pleasure, sir.
30:42 Chuck Nice? Hey, Dan, man, thanks so much. What a fascinating conversation.
30:45 Gary O'Reilly? Yeah, and thank you for bringing your mind to my game.
30:48 Speaker 02 It's been my pleasure. Thank you for your game.
30:51 Gary O'Reilly? On the shoulders of giants as we all stand. Thank you, Dan Altman.
30:57 Chuck Nice? Yeah, Dan Altman.
30:58 Gary O'Reilly? Who knew you'd like soccer, only because you would bet on it and get lots of money?
31:02 Chuck Nice? Yeah, damn right, buddy. Let me tell you something, I'm changing my whole career. I'm changing it all. I'm going right into sports betting and I'm using analytics and I'm buying a soccer team.
31:12 Gary O'Reilly? Or should I say, a football club? I'm going to play you a Pink Floyd track called "Money". That might just dampen your thoughts for a while. Right, we're going to take that break. Fabulous Dan Altman has been our guest and when we come back, we'll have the fabulous Dr. Howard Hamilton, another data analyst for football. Can't get enough, can we? Back in a minute. Welcome back to Playing with Science. This is my running around in my own backyard, nerding out like a true geek. We have spoke to Dan Altman from North Yard Analytics and I can say now we have the privilege and pleasure to speak to another sports analyst, Dr. Howard Hamilton, founder and CEO of Soccer Metrics Research.
31:57 Chuck Nice? Research? What a great name, too.
31:59 Gary O'Reilly? All right, let's get straight into it. Tell us about Soccer Metrics, how it came to be and where it is headed right now.
32:08 Speaker 02 So Soccer Metrics is a data analytics firm that performs quantitative analysis.
32:16 Speaker 03 On data that the football industry uses every day, that the football use, football industry generates and uses every day from team performance, player performance, team and player valuation, uh, op- operational statistics related to the per- the performance of leagues, uh, the performance of teams at the front office. If it involves data of any sort, I'm interested in analyzing it and generating actionable information.
32:50 Gary O'Reilly? So it's not, it's not just focusing on the game, the players, the outcome and why. It might be things into the front office. So for things like ticket sales, uh, are there areas like that? Have I got that wrong?
33:05 Speaker 03 Yeah, ticket sales is part of it. Um, there was one project I did, uh, very early on where I was looking at,
33:14 Speaker 03 you know, the sensitivity of ticket sales to the current record of a sports team. And, and there was some interest in that. I presented a project, I presented that work at, uh, Penn Wharton about five, six years ago.
33:28 Gary O'Reilly? Okay.
33:29 Speaker 03 There was some interest from StubHub on that. It was actually, it was using StubHub's data. The front office work more relates toward a, a benchmarking analysis. Uh, there is, there was a professor out of the UK named, uh, Bill Gerard, he was a, uh, um, he's worked with Billy Bean, um, on a number of projects from Oakland A's to AZ, uh, uh, Alcor. And he, he created a systematic, he created a benchmarking analysis of the performance that teams got relative to the amount of.
34:08 Speaker 03 Found the resources they were putting in.
34:10 Gary O'Reilly? Okay, let's, let's take a step back.
34:12 Speaker 03 The amount of points that took to win, the amount of payroll that took to win a match or a league point, and I applied that to Major League Soccer.
34:20 Gary O'Reilly? All right, let, okay, let's take a step back, not, not a big one. You mentioned, you mentioned the B word, the Billy, Billy Bean. Um, AZ Alkmaar, Chuck, is a professional team in the Dutch league.
34:32 Chuck Nice? Right.
34:33 Gary O'Reilly? At a division, and I'm not mistaken, Billy has joined a group that's bought a football club in England, Barnsley in Yorkshire. Uh, is football money 2.0 as we've been thinking it is? Is it ripe for the kind of moneyball scenario that we saw the Oakland A's and what is now prevalent through Major League Baseball?
34:57 Speaker 03 I think you've seen it in a number of, of places. Um, you know, Matthew Benham, you know, and, and his clubs from North Shaolin to, to Brentford, you know, have, have employed, I guess, a moneyball type of approach. They had a head of, of football strategy, Ted Knudsen, um, who, who started Stats Bomb. Um, has done a lot of work with Matthew Benham. Um, Barnsley, I'm not really aware, aware of.
35:31 Gary O'Reilly? Oh, but look, we don't have to focus on that. Let, let's come, let's come out of that scenario.
35:35 Speaker 03 Um. Yeah, but I would say to, to answer your question, I, I would say that it's, it's ripe for it, but it's a very, it's, it's a, it's a very gradual process right now. I think there's still a lot of cultural resistance to it.
35:51 Gary O'Reilly? Yeah.
35:52 Speaker 03 Not, not totally unfounded in my view, but I think that there will be some cultural resistance to that.
35:59 Chuck Nice? So let me, let's switch gears for a second, because I want to ask you about some of the stuff that you've done. I was looking at your website and some of the things where you, you did an, uh, an analysis of effective playing time in the World Cup for different referees and players, I mean, I'm sorry, and teams with those particular officials.
36:20 Chuck Nice? What do you do with that as a coach? Like, do you actually prepare in your game plan for the officials that'll be officiating? And does that, do you plan on your teams being more rested or more taxed? Or what, like, you know, why do you do something like that?
36:36 Gary O'Reilly? It's interesting, and I'm interested to hear your answer to that. Because as a former player, whoever is the referee on the game
36:44 Chuck Nice? Uh huh.
36:44 Gary O'Reilly? can not always make a big difference.
36:47 Speaker 03 Yeah, it's interesting you talk about that analysis. And I've applied that analysis to various competitions from Major League Soccer to the English Premier League, to Japan's J-League, and to the World Cup. And what I found out was that the identity of the referee really doesn't make that much of a difference as to the amount of effective playing time.
37:14 Gary O'Reilly? Okay.
37:15 Speaker 03 The, it's the two teams that make more of a difference. For example, for example, you know, going back a few years in the Premier League, any match involving Stoke, involving Stoke City would almost always have a playing, effective playing time of less than, of less than 55 minutes, which is well below the Premier League average. And, you know, I remember sharing that with some people in the Premier League, and without my communicating which team was associated with the least amount of effective playing time, they all said, I'm sure Stoke City is at the bottom, and they were.
37:58 Speaker 03 I think where, I think where the referee is important is in the amount of, say, time between stoppages. Because that's where the, time between stoppages, time between fouls, because fouls are those events that the referee is actively involved in stopping play. And there's a lot of matches where, again, that's, that's a function of the two teams that are involved. But the referee has some input in that as well.
38:34 Speaker 03 Yeah, so I think the referee matters more as to the amount of flow in the match, as opposed to the amount of effective time that is actually being played.
38:46 Gary O'Reilly? Some of the work you did, and thank you for explaining that, some of the work you did regarding the World Cup we've just seen in the summer in Russia, and this caught my eye. So please explain, as regards the passing networks that are produced and provided by a number of sites, yours included, please explain to Chuck, myself, and probably our listeners, what is an eigenvector centrality? Because I'll be honest with you, Howard, it sounds as if you stole that from an episode of Star Trek. Ooh, I like it now. You know what I mean? It's straight out of Gene Roddenberry's textbook, that one, for sure. So please explain. She'll never hold cotton. She'll never hold... That is the worst Scottish accent ever.
39:28 Gary O'Reilly? That's not a Scottish accent. That's actually Scotty from Star Trek. See? Still not good. Sorry, Howard, please. The eigenvector centrality. See, now that's a Sean Connery Scottish accent. Yes. Please, we've interrupted his answer. Go ahead, go ahead.
39:47 Speaker 03 In network theory, where you have these nodes, and I'll just make it really specific to passing networks.
39:56 Gary O'Reilly? Okay.
39:56 Speaker 03 So networks consist of nodes and edges. Your edges are the passes, your nodes are the players, and your edges are the passes between each player. You can represent this in various ways. You can represent those nodes and edges in terms of a matrix, just some numerical representation of this network. Now, the eigenvector is, this is real scientific, I'll generalize it.
40:34 Gary O'Reilly? No, no, be as deep as you wish. Seriously, be as deep as you wish.
40:39 Speaker 03 It's a it's a representation of of that metric of that matrix. And centrality is the the relative importance of each node to to the network. And eigenvector centrality is one specific type of of measure of importance of a node, in our case, a player, to the passing network. So the reason why I used an eigenvector centrality was that it represented the player who was most important to
41:21 Speaker 03 to the success or the structure of that network. If that player wasn't present in the passing network, in that in the passing network, that passing network would look significantly different. And
41:37 Speaker 03 yes, there there's some natural variation from match to match, but even the World Cup, where you have a minimum of three matches, some teams play four up to seven. You can still identify those players who have a who have a consistent impact on the success of their passing network. South Korea is one example.
42:01 Chuck Nice? Through Son.
42:02 Speaker 03 Yeah, Son. Yeah, Son was Son was by far the most important player of the passing network, which is extremely unusual.
42:08 Chuck Nice? Because he's a forward player.
42:10 Speaker 03 Yeah, because Son's a forward player.
42:12 Chuck Nice? Interesting.
42:12 Gary O'Reilly? So what I mean, it's funny, I would watch the game and then what what someone like Howard would do was quantify it numerically. Quantify it in, if you've seen ever seen a passing network, it's a diagram.
42:24 Chuck Nice? Yeah.
42:24 Gary O'Reilly? And and and you can then visualize, but what
42:27 Chuck Nice? It looks like thirty NFL plays layered on top of one another.
42:32 Gary O'Reilly? Yeah. So and then what happens is, from the analytics point of view, if you realize South Korea are using Son as one of these.
42:42 Gary O'Reilly? Points, these nodes in an advanced area, you then structure your team to disable that to happen while they are trying to construct every way to enable Sun to do his thing. So it's quite interesting, the way I would look at that. But what Howard does and he's, he's, he's a, the like that's out there, they show you in a visual form that's easily condensable.
43:07 Chuck Nice? Interesting. Let me ask you this.
43:08 Gary O'Reilly? Yeah.
43:09 Chuck Nice? Okay. So France won the World Cup, right?
43:11 Speaker 03 Hmm, right.
43:13 Chuck Nice? Right. And so everybody was just like, wow, France won the World Cup. Wow. Okay. Cause it, you know, that's.
43:19 Gary O'Reilly? Oh, congratulations.
43:21 Chuck Nice? Yeah, congratulations. And by the way, Johnny, Johnny Bonhomme is in our, our control room. He's our producer of this show and he is, he is French. So, you know, I'm sure he's very happy. He's like a thumbs up right now. But here's what I want to know. Was France, from an analytics standpoint, the best team to win the World Cup? Cause everybody was like, whoa, France won the World Cup. Wow. But from an analytics standpoint, should they have won the World Cup?
43:48 Speaker 03 I think France are the most consistent team at the World Cup. I think this World Cup was, was very inconsistent on a number of levels, which may be entertaining for us as fans. I think this was one of the most unpredictable World Cups in a while. Like Germany not making it out of the group stage for the first time in 80 years. Um, yep. Spain, you know, Spain falling stage, they did, you know, England going all the way to the semifinal. Um, yeah, a team, you know, they're, they're very, you know, Argentina, Argentina imploding, Brazil not make, not making.
44:29 Gary O'Reilly? Brazil, I read Brazil stats that you produced and Brazil stats were as spectacular as their failure.
44:34 Speaker 03 Yeah, yeah, totally. If, I think if you looked at the expected goal figures, not just on the, not just on expected goals scored, but also expected goals allowed, Brazil was way out in front in both statistical categories. Neymar. Yeah, I think if Neymar had advanced to, if Neymar and Brazil had advanced to the finals, I think he would almost certainly have been seen as the best player in the tournament. But their, their, their conversion rate was abysmal, which is part of the reason why they didn't advance there for it. I, I think if you looked at that match against Belgium,
45:15 Speaker 03 you could see that. I think France, it, France are a good team. They're, they're a really solid team. They had the best.
45:22 Chuck Nice? Um, I'm sorry, I'm sorry, Dr. Hamilton, they would kind of take issue with that. Uh, you say they're a good team. They would say, we're the best effing team because we won the World Cup.
45:34 Gary O'Reilly? So the thing is, I mean, it's.
45:36 Speaker 03 No, they're, they're totally, they're totally proud to say that.
45:39 Gary O'Reilly? So it's interesting, it's interesting. Here, Chuck, 16.3% doesn't sound a lot, does it? When, you know, we always talk about 100%. That was France's conversion rate from opportunities created to chances converted into goals. And the team that's won the World Cup has less than 20% conversion rate. Is that because everyone is useless? Or, Doctor, is the game of football that we know and love proving to be a lot harder than people give it credit for? And that's why the score isn't 110 versus 103. It's possibly 1-0 or 2-1?
46:18 Speaker 03 I think it's always been like that. I, I, if you look at, if you look at competitions in soccer, from domestic competitions to continental competitions, national team competitions, at best, the conversion rate from shots to goals is around 10%, maybe 11, 12% if it's a higher scoring league or your defenses are, are garbage. Um, so to have a conversion rate of 15% or 18% or more is, is really fantastic for a team, especially at an international level where you don't get a lot of chances to score. Um, if you're converting at greater than a 12% clip.
47:06 Speaker 03 You're putting yourself in a really good position to go really far in a tournament and ultimately win it.
47:11 Chuck Nice? Okay.
47:12 Gary O'Reilly? Well, there you go.
47:13 Chuck Nice? There you have it.
47:14 Gary O'Reilly? It's Dr. Howard Hamilton.
47:15 Chuck Nice? Basically, here's Johnny, Johnny, our producer is looking through the glass here. Johnny, what everybody in the seems to be saying is, uh, France got lucky, so enjoy it. It's not gonna happen again. Okay.
47:25 Gary O'Reilly? So, Dr. Hamilton, thank you so much.
47:26 Speaker 03 You know what? Just might.
47:28 Gary O'Reilly? Oh, he just might, you never know. Um, can happen.
47:31 Chuck Nice? Of course, I'm joking.
47:32 Gary O'Reilly? What a pleasure. Thank you so much for sharing and the new word of the day is Eigenvektor.
47:38 Chuck Nice? Eigenvektor. Eigenvektor.
47:40 Gary O'Reilly? Oh, right, okay.
47:42 Chuck Nice? Ivan Flavin. Eigenvektor.
47:45 Gary O'Reilly? This is the first time he's done it in the show. He's gotta get it out of his system. Alrighty, thank you so much to Dr. Howard Hamilton, um, from Soccer Metrics. Been a pleasure to talk to him. So Chuck, we have learnt a lot.
47:57 Chuck Nice? Yeah, it's been.
47:58 Gary O'Reilly? And we have still only scratched the surface. And we have learnt there are some silly names attached to football that have nothing to do with the game itself, but make it all the more interesting, like Eigenvektor.
48:11 Chuck Nice? Eigenvektor.
48:12 Gary O'Reilly? All right.
48:13 Chuck Nice? Ivan Flavin.
48:14 Gary O'Reilly? Yes. Right, while I recover from Chuck's outburst, uh, I'd like to say, I've been Gary O'Reilly.
48:20 Chuck Nice? And I've been Chuck Nice. Lying.
48:23 Gary O'Reilly? Yes, he has. Uh, and this has been Playing with Science and my little nerd geek fest on football, which I have thoroughly enjoyed. I hope you have too and we look forward to your company very, very soon.
Previous Extended Classic – Let’s Make America Smart Again: The Future of NASA Next Living in Space, with Scott Kelly Page updated September 29, 2026 · transcript and summary are machine-generated
