Episode Transcript
[00:00:13] Speaker A: For p. 97 welcome to the second half of our discussion of Hockey Analytics. Our guest today will be Dom Lucian of the Athletic. I have posted a collection of Dom's writing from the Athletic on the course page, as well as a little bit of a sampling of his other work. If you don't have a subscription to the Athletic, if there's a particular article that you're interested in the Athletic that you don't have a subscription to, send me an email and I can pass along a PDF version of it. If you can afford a subscription to the Athletic, I heartily endorse it. I think it's a wonderful site that provides a lot of quality journalism, but I'm not here to give you endorsements for purchasing the Athletic, just that it's a good site and that Dom writes for.
Previously, Dom has written for the Hockey News, he's written for the Nation Network, he's written for Hockey Graphs, and he's one of the most, I think, talented writers at explaining the implications for advanced statistics in hockey as it relates to gambling, as it relates to your fantasy lineups, and as well as relates to roster construction and the overall relationship between advanced statistics and success on the ice for your favorite team and every team in the NHL.
We're extremely lucky to have Dom come in and talk about some things, and before we get started with his guest lecture, I want you to sort of pay attention to some of the topics that Dom speaks to.
First of all, again, the way in which the word analytics is defined was one of the questions I'm going to ask him. But again, and it will be something that I'm going to be asking you to do, both in writing and in your audio responses.
So throughout Dom's discussion here, I want you to keep in mind the relationship that exists between professional sports in general, in this case the professional hockey industry, and the notion of statistics, advanced statistics, or as they are somehow defined, analytics. Now, again, we're going to be digging in deeper into the definition of what analytics is in a little bit, and that's obviously something that's important. But I want you to keep this in mind, even if this is something that, again, does not necessarily pique your interest as a fan. Recognizing its importance and its prevalence in the professional sports industry is, I think, something that is very important. Something that I think we are going to develop a little bit more further in the teaching of this course in years to come and other courses is the ways in which these types of performance evaluations for on ice talent has affected the entire industry as a whole in terms of the way of evaluating the success of a front office or a marketing team or a sponsorships team. I think that you've seen this certainly in baseball and I think that when we broaden our scope a little bit to talk about this in maybe other courses where I'll discuss the way in which baseball teams have started running entire operations like they are the on ice analytics teams themselves, they'd be, they've become much more data focused in every aspect of the sports industry. I think that that's something that if you're not again interested in this sort of performance evaluation side of things, that's okay. But just understand this notion of, you know, how to quantify success, how to justify headcount. All these sorts of things have changed drastically over the last few years and continue to change. And as we move into this post sort of data revolution and into data being a mainstream part of hockey and of most North American professional sports, if not all. So that's enough preamble. This is Dom Liu Shizian again. If you have any questions, please pass them along to me and I can pass them along to Dom on your behalf. Dom was happy to answer any questions about the industry itself or any specific analytics questions. And again, I do encourage you to do reach out if you have a question for Jack Han as well. He does. He is looking for questions and I am eager to pass them along to him. So again, don't, don't be, don't be shy. They are actively soliciting questions. Both, both figures. Byron as well from earlier, Ryan pike as well. They're all extremely generous. I've had a few people pass along some questions for Tori and that's been fantastic. But my goal is to again be able to connect you with some of these people and introduce you to their stories and their, their experiences. So if you've got any questions for Dom or any questions for Jack, please.
[00:04:14] Speaker B: Send them along via email.
[00:04:15] Speaker A: And here is our guest lecturer and our visit from Dom Luchazen.
[00:04:26] Speaker B: We are so fortunate to be joined by Dom Luz Chidian of the Athletic.
[00:04:30] Speaker A: Dom, thanks so much for joining us.
[00:04:32] Speaker C: Thanks for having me.
[00:04:33] Speaker B: Dom, if you could take us through your initial interest in hockey statistics and analytics. How did you first develop an interest in this field?
[00:04:43] Speaker C: There were two reasons.
Neither of them had to do with school because I am a journalism major, so I don't have any formal mathematical background. I was good at it in high school, but for some reason I thought why not try writing? Because I want to do something close to the sport. But at the time there wasn't this whole analytics movement. So I didn't know that was really a path I could go down. But when I was in school, this was around the early 2010s when the Leafs were almost good, but still quite bad. And I am from Toronto, so I'm a Leafs fan, and I saw them play, I saw them win. It didn't make sense. And I was on Twitter a lot and I was reading a lot of what people were saying and a lot of things made sense. The Leafs couldn't get the puck out, they got hemmed in a lot. They got out shot, they got out chanced, and they still won some games somehow because they had Kessel who could score a lot, and they had some offensive players and goaltending sometimes. But yeah, it didn't make a lot of sense. That was my, I guess, introduction into all of it. And the second part was me being an arrogant 20 something, thinking I can make money off of this by betting and I should do something with that. So I started just putting numbers together sort of randomly because again, I didn't have any formal training, I didn't do the Googling, I didn't learn anything yet on how to actually make a model. And so I was sort of just going with that and seeing what fit and what stuck and what made sense. And I played hockey for my whole life, so I sort of went into it with that kind of background, especially in terms of seeing the Leafs and seeing that it didn't work and finding stats that made sense of all of that.
[00:06:45] Speaker B: Perfect. That's a. That's a great description of how, how.
[00:06:48] Speaker A: You can sort of take your own.
[00:06:49] Speaker B: Experiences as well and apply them to your writing. So you develop your own model. You said there, you developed it sort of from the ground up. What did you learn from the early process of you giving it a try to try and create your own mathematical interpretations of what you were seeing.
[00:07:04] Speaker C: Yeah, so obviously it was a trial by fire thing because I didn't have any formal mathematical training and I would just follow other smart people on Twitter and see what they were saying. And if there was a word or concept I didn't recognize, I would just learn about and Google it and try to apply that into my own things that I was doing. And I was putting my stuff out there and people were reading it and critiquing it. And the other thing with gambling is that if you're wrong, you will lose money. And people don't like losing money. So I was one of those People, I didn't like it and I just strive to learn more and do better and figure out how to actually model things a bit better. And I was learning from a lot of people on like the betting sphere on how to do that and how to do that properly. And I think there's a big thing to be said about having skin in the game and in that sense, because if you are beating the odds makers of beating Vegas, then you're probably doing a good job because there are a lot of other smart people putting their money where their mouth is. And if you can beat them, that's a good sign that you, you're doing something right with the numbers you're using. And if you're not, then what's really the point of putting them out there? Because if you're just going to get crushed, then there's probably something wrong with it.
[00:08:28] Speaker B: Absolutely. And in your experience through your the time that you've been in this space, have you noticed things have changed a little bit in terms of the way the teams have behaved have the way the teams have acquired players or deployed them. Have you noticed a change since you started writing?
[00:08:42] Speaker C: A little bit. There will still be.
It sort of goes both ways where the analytics movement, I think at the start felt they knew it all and they knew everything. And these teams were stupid, these coaches are stupid. And that was obviously wrong and false. I think there has been a lot of movement towards the middle of both sides trying to like gain understanding of what the other side is seeing.
So with a player like Ben Sherrod who has terrible underlying numbers, he gets traded for a first to fourth and a prospect who won the third round last year.
The initial take is this guy is terrible, his numbers are terrible, the team overpaid for toughness or whatever and I was definitely among those people. But you look at the team that acquired him and it was Florida. And Florida has done this amazing job of finding players that publicly don't look that great. And as soon as they get to Florida they look amazing. So there's obviously some sort of skills they're targeting and they have a better grasp on what fits into their system, what fits with their players. And these players start thriving with Florida. And it wasn't that they were bad, it's just probably the way that the public measures these things might not be as accurate as they think it is. And Florida has some very smart analytics people themselves. They have access to better data. So when they do something like that should be a bit of an eye opening experience from this side but there are still times where you see that the things people are doing, the public do have some value and some merits. A player like Brett Kulak, for example, I don't think would have ever fetched a second round pick in previous years, but because he has strong underlying numbers and has shown that in the past, even if it wasn't this season specifically, there is value in those players. I think teams are starting to pay a lot more attention to how players drive play and the specific skill sets that lead to those things. And one of those things I think it was near the start of the 2010s. Eric Tulski developed this paper on zone entries and zone exits and the importance of them towards out chancing teams. And you see a lot of teams not only start targeting those players, but certain players start playing in that way where carrying the puck has become a lot bigger of a deal.
[00:11:15] Speaker B: That's a really interesting concept. The idea that the way that the game is being understood and actually played at the playing level has been altered by this sort of statistical analysis approach to games themselves. Zone entries and zone exits being one of them. Do you see that there are inefficiencies still when you watch the game? Do you find that there are still tendencies that are legacies of an older time or do you find the game pretty uniformly transformed?
[00:11:38] Speaker C: There are definitely still concepts out there, but it is hard to say based on what we have in the public, for example, because our data is so limited.
I think one of the ways we're seeing a divergence right now is in expected goals models where the public only has so much information in terms of shot location and angle and they can infer rebounds or whatever. And the public has pre shot movement. Better understanding of rush chances, rebounds and their data will be a lot stronger in that regard. In some teams, rather than target location as the goal, they target pre shot movement as the goal instead. And their numbers publicly don't look very good, but they're still great teams getting great results. And there is a bit of a disconnect with that, I think the inefficiencies right now, it is hard to say, but I feel like market value is still one of them. Where teams will still overpay for aging players and not exactly go after younger players with say an offer sheet, for example, because there's this code or whatever or they don't try to pry a player out through that realm. I think we saw the best example of that recently was the Avalanche with Devon Toews. They didn't sign offer sheet, but they basically offer that compensation back to the Islanders in exchange for him because they didn't have the room to sign him. And now he's one of the best defense in the league. And there are opportunities like that around the league where teams are really capped out and these young players get pinched when they can provide a lot more value than the contracts they're actually being given.
[00:13:29] Speaker B: We did talk in our class. We spoke about the contractual mechanisms that are in place to keep costs low, being the entry level deal, the restricted free agency process. It is a fairly convoluted process when you try and break it all down.
[00:13:43] Speaker A: That notion of market value, I think.
[00:13:45] Speaker B: Is a really fascinating aspect of this as well, because I often encourage students to imagine what Conor McDavid's actual market value would be if he ever was able to. If he had a series of one year deals and a LeBron James model, what would that be in terms of what is he actually worth as opposed to what is he worth in a cap world?
Shifting gears a little bit though. I want you to sort of. I want to give you the freedom here.
How do you define the word analytics?
[00:14:12] Speaker C: I think I'm going to cop out of this one and go with what every new GM says in their introductory press conference when asked the same question. It is just information and how you present it and how you communicate it. I think anything can be analytics if you want it to be. Even traditional scouting, if you just mark down what you're seeing and sort of tabulate it and calculate it, it can turn into some sort of analysis. If you just track those specific things that a scout is looking for, even if you're just grading a player by his game, those can be analytics still. It's not just stats. It's not just numbers. It's whatever information you can quantify.
[00:14:55] Speaker B: Really good definition there. Fantastic. And you've mentioned this before now you've alluded to it in a few of your answers. This notion of what is publicly available data and what is privately available data. And I do recall a period in time in which people were relatively skeptical of any organization that came around that claimed they had some sort of black box model for data. Do you believe that the notion of public and private data has kind of settled this debate?
[00:15:21] Speaker A: Is there valuable private models that are out there or do you still believe.
[00:15:25] Speaker B: In the value of having a publicly available model?
[00:15:28] Speaker C: I think there's value in a publicly available model. It's just they. People who make those have to understand the limitations behind them and how many things we are Inferring based on the models.
You hear the term like play driver a lot in terms of players, and that generally just means that when they're on the ice, their team does well. When they're off, the team doesn't do so well. And there'll be certain mathematical formulas to try to individualize that as much as possible. But there can still be error in that because they're not specifically driven by individual skill sets. They're not accounting for one player being the guy to carry the puck in, one guy who is for checking and retrieving pucks, one guy who's exiting the zone, one guy who's making all the passes. So it's those individual skill sets that can tell us more about who is actually the one driving the bus on each of these lines, where the private models will have access to that data and the public models won't. But at the same time, it's still a matter of who is modeling and what they're doing and whether they're doing it effectively. You can have the best data in the world and do a terrible job because you're not doing the right things with it, or you can have this crappy public data and still do a decent job because you're making the right choices and assumptions based on the data you have offered to you.
[00:17:00] Speaker B: Wonderful. And that's. That's, you know, reminds me so much of those earlier days on the Internet where there was not very much information out there, and yet some people were doing some really cool stuff with it on their own, private blogs, yourself included.
This is something that I think is. Is fascinating to me is this idea of what is publicly available and what is privately available, and this new frontier of this sort of tracking information and things like that. If I gave you $100 million and I said I want you to create an analytics company, what information would you. What technology, what kind of information would you invest in first?
[00:17:30] Speaker C: Like an analytics company for hockey or.
[00:17:34] Speaker B: Yeah, for hockey specifically.
[00:17:40] Speaker C: I guess the number one thing is tracking every team, obviously, and getting all the necessary data points and then hiring a bunch of data sciences scientists, data engineers, people who can work with such a large data group and figuring out what makes sense, what doesn't, what is worthwhile, what isn't, and having someone who can communicate what the data people find to the hockey people and sort of be a liaison between those two worlds. Because if you can't speak hockey and convert what the data people are saying, then the hockey people won't do anything with it. They won't understand it, they can't because it's not in their language. You have to be speaking their language in order to get results. And I think that is where a lot of data, people who were hired and who weren't exactly listened to, that's, I think, where they.
That's where the miscommunication was and why they weren't relied on. Even if they were the smartest people, they weren't speaking the same language.
[00:18:50] Speaker B: On that topic of knowledge translation, which is, I think, what you're describing there, how do you. Do you struggle with this in your own writing? What's the part of this process that you find the hardest to translate to a more traditional hockey audience?
[00:19:03] Speaker C: It is always hard because people intrinsically don't like math, and math is hard. And a lot of people's eyes glaze over when any mathematical subject gets, I guess, put out there. I think early on in my career, when I was just interning with the Hockey News and I was pitching these ideas, they basically said, you have to remember that your audience is not very smart and try to speak not down to them, but like to their level so they understand and try to make everything as easy to understand as possible. And I've sort of stuck with that throughout my entire career. And I know there's times where it's a bit difficult to sort of explain the information, but I try my best to keep in mind the casual fan and try to not only help them understand, but if they don't understand, make them feel welcome enough to ask.
And I try to make a point to explain concepts to anyone that is confused unless they come at me from a place of bad faith and immediately try to be rude. And in that case, I'll just be rude back because they aren't.
I'm a busy person. They don't deserve my kindness if they're going to be rude first. But if people come at it from a good place and are nice and genuinely curious, then even though they don't agree with what I'm saying, I'm still always happy to explain and have that debate and sort of understand what they're seeing as well. Because again, this public data that we have isn't the best. And there's always blind spots that will still be had. Other are fans of teams that'll say, I can't believe your model likes this guy. I think he sucks. And I'm like, all right, tell me more about why this guy sucks because his numbers are good. So what do you see? And then you have a discussion about what the two fields are missing and sort of go from there because there's always that gap. The idea is to bridge the gap and have both people coming out smarter from the discussion.
[00:21:28] Speaker B: It's a fantastic message, especially for people that are, you know, seeking careers in the hockey industry. I think that's really, really, really great advice. When you're, when you're looking back at your career so far, is there something that you look back on? You think to yourself, this was a really important thing. This was a really important skill that I developed early on. You mentioned that you weren't traditionally trained in mathematics, but you clearly have many other skill sets, including a background in journalism. Is there a skill that you thought to yourself maybe less in terms of a practical sense and more in terms of yourself personally? What skills do you feel like you developed throughout your career so far?
[00:22:00] Speaker C: I think it was an asset to study journalism and not math because my skills are in communication and writing and getting my thoughts out in a clear, succinct way that hopefully a large number of people can understand. Because there are probably many smarter people at math in this field than me who are doing much more difficult things and much more complex things, but they don't have the same ability to sort of explain those ideas in a way that the majority will understand or make sense of. And I think that is probably my biggest strength. Strength is just speaking other people's language and helping them understand these concepts that at first may seem a bit scary.
[00:22:53] Speaker B: That's wonderful. And one of the topics that have come up with many of our guest speakers so far has been the importance of failure. Where, you know, we've had many people come in and discuss things that they have, they've, they've learned from and things that they've, they've failed at, frankly, many times over and over again. Is there an aspect of your, your work so far where you look back and you think, wow, that was really something that I learned a great deal from. I'm reminded of, you know, in the early days of the Internet discourse around this topic, these, the so called shock quality wars that occurred. You know, it seems so quaint now in retrospect those, those days, but were there, were there parts of your work now you look back and you think, wow, I really wasted some time or I learned from or any of those sorts of things.
[00:23:35] Speaker C: I think failure is never a waste of time because you're learning and you're making yourself better and you're becoming a better person. For me, I think the turning point was the 2019 playoffs where the St. Louis Blues won. And the model I had at the time wasn't really believing in them as a strong team while other models did. And I think the big thing there was a lot of people were still debating between course scene expected goals, and I was still of the mind that shot attempts were the way to go because it was easier and I was using goals as well. So I thought the mixture of the two made sense. And the other thing was that the Blues were a strong defensive team, and in my mind, defense wasn't extremely predictable. So I focus a lot more on offense and sort of had a bias towards offensive teams specifically. And while it's true that offense is easier to predict and you should have a bias towards that, there is a line where you're obviously too much in line with the offensive teams and ignoring defense. And I think that specific final was St. Louis vs. Boston, two of the best defensive teams in the league. And I knew that something had to change there. And the signal for me was not specifically that those two teams did well. It was that when I was betting, it wasn't necessarily that I was losing money. It was that I can see the market not move towards what I was doing. So if I bet on a team, the market would either stay in place or go in the opposite direction, or I would have these substantial edges on certain teams specifically, and I probably shouldn't have had such a big edge. And when you get that kind of signal, you think there's something wrong with this team specifically that other people are catching that I might not be?
[00:25:29] Speaker B: That's really helpful information there. And is there a statistic? Is there. Is there a way of evaluating the game? Is there something that you've seen in the last little while here, maybe the last six, seven months last year or so that really interests you? Do you see something that you're thinking to yourself that's really interesting to me. That's something I want to learn more about?
[00:25:47] Speaker C: Yeah. I think for me, it's the idea of all these teams focusing on being heavier and playing physical and how much the playoffs are different from the regular season in that regard. And I mentioned Ben Sherrod already in Florida is a smart team that spent a lot to get him. He has that reputation. He is good at being a physical defenseman.
I am definitely interested in how worthwhile that is in a playoff setting compared to the regular season, whether a team like, say, the Leafs are too soft to win in such an environment.
[00:26:24] Speaker A: Wonderful.
[00:26:24] Speaker B: Just to finish up with a final question. If someone is just just starting out an interest in hockey analytics or in data science or anything as applied to sports. What would your advice be to someone who's just starting out a career or just trying to develop an interesting in analytics or hockey statistics more broadly?
[00:26:41] Speaker C: I think just read everything that's already out there and if you're interested in the concepts they were creating, try to recreate it with your own skill set, with your own curiosities, with your own intuition, and see if you can either come up with the exact results they did or maybe even improve on those concepts. There's a website called metahockey.com that has basically every single analytical piece since the beginning of this era in terms of shaping how we all think and feel. And I think there's probably older pieces that could use a bit more scrutiny and going through them and trying to pick holes in past arguments could be definitely a useful exercise.
[00:27:31] Speaker A: Dom, thank you so much for sharing.
[00:27:33] Speaker B: Your time with us.
[00:27:33] Speaker A: Us and and everyone in schema 4000p.97. Thanks you for for your time and attention here and, and thank you so much.
[00:27:40] Speaker C: Yeah, no problem.
[00:27:51] Speaker A: So that's our discussion with Dom again. If you have any questions, pass them along to me. And in terms of the site that he mentioned there towards the end of our discussion called Meta Hockey, metahockey seems to be experiencing some technical difficulties. So I've actually included an archive of all the articles from the past. It is an absolutely amazing resource, MetaHockey, but I've included an Internet archive version of it on the Brightspace site. So if there's anything that you'd like to access from the archives of MetaHockey and sort of tracing the development of hockey analytics. The link is on your BrightSpace page under Week 8 worth of reading materials. Go have a look at that. Hopefully it's just a temporary issue with with MetaHockey. It is a wonderful website, but in the meantime you can have a look at the archived version. But there's so many interesting things that I think Dom brought up here in this discussion. If you've done read his writing, I think that you can see that as he mentions here, his. The evolution of his thinking is. Has. Is clearly evident in his writing. And again, if you have a chance to read Dom's writing anywhere, go and check it out.
He is an exceptionally gifted writer and gifted communicator. It's one of the things made him very successful and can make you successful too in the hockey industry is being able to communicate your ideas effectively, not just to an expert audience, but to a wide audience as well. And that's something that I think is, is good advice no matter what aspect of the, of the sports industry you are looking to work in moving forward.
So this brings us to our audio response question. And generally speaking, when I teach this class in person, I like to have these sorts of debates in person, but this will have to suffice.
So I'm going to give you a very specific task.
You have a $10 million number. Imagine yourselves as part of a hockey operations staff and you're sort of been brought in as an external consultant with this sort of number of $10 million.
So this $10 million can either go towards expanding your analytics and your hockey statistics staff in terms of contributing to on ice performance, or you could justify saving that $10 million and spending it elsewhere in the organization.
You must articulate on one side or another using specific reference to something that you've learned this week through the articles, through the guest speakers, what have you, and make specific reference to something that was said by Dom or said by Jack. But you can make specific reference. Now, you are totally within your rights to say, you know what, I've looked at the, the, the information here and I think you're actually better off spending money elsewhere. But I'm going to need to know where and I'm going to need to know why. I'm going to need to say, well, Jack said this, but I think this because of these reasons. I need you to be specific, but if you say, you know what, $10 million needs to be expanded on the, the analytics staff for these specific reasons and maybe even saying doing these specific things, but it's sort of a thought exercise for you. Now is there many NHL teams that can spend this kind of money on an analytics staff? Absolutely not. But there are some Premier League teams, some, some baseball teams certainly that spend.
[00:30:54] Speaker B: This kind of money.
[00:30:55] Speaker A: I don't know what the total number of the analytics budget is for the Toronto Maple Leafs. It might be this size. But imagine just for the purposes of this thought out experiment, this, this case study, we are looking to spend $10 million either on expanding an analytics and analytics staff or creating one in the case of your, your fictional team that doesn't have one, or spending that money elsewhere. But in either event, I need to know why you've decided to make these decisions. And for the purposes of this course, I need you to reference something that came up this week either in the articles or in the guest lecturers. So again, make specific reference if you're going to talk about a guest lecturer. And I need you to do that to the time in which you are you are talking about what they said and otherwise I can't wait to hear this. It normally ends up being a lively debate. I suspect it will be a lively sort of digital debate of sorts as well. And if you've got any questions, send them along to me. And I look forward to talking to you all next week.
[00:31:45] Speaker B: And I hope you have a great weekend.
[00:31:55] Speaker C: I know about you just take home.
