Hockey Analytics, Simulation, and Predictive Limits
Ask about Knowledge at Wharton
Episode Description
ABOUT THE EPISODE Micah McCurdy, mathematician and creator of HockeyViz, explores how simple models, simulation, and player level data shape modern hockey analytics and why getting beyond the basics of prediction is harder than it seems.
ABOUT THE PODCAST Sports is a game of numbers. Wharton experts Eric Bradlow, Shane Jensen, Cade Massey, and Adi Wyner team up to tackle the world of sports, from current events to longstanding issues such as: What sports streaks are the most impressive? How do you rank the best players? Can athletes be compared across sports? Moneyball Highlights explains how decision-makers in the game can avoid the common mistakes and embrace the data. Episodes are recorded at the Wharton School.
#moneyball #hockeyanalytics #sportsanalytics #datascience #wharton ----- The Wharton School of the University of Pennsylvania is committed to sharing its intellectual capital through Knowledge at Wharton, the school’s online business analysis journal.
Launched in 1999, Knowledge at Wharton offers free access to:
- Articles, podcasts, and videos highlighting Wharton faculty research and analysis of current business trends - Interviews with book authors from Wharton School Press - In-depth series of curated content like Wharton Executive Education’s Nano Tools for Leaders - A searchable database of more than 10,000 articles covering all aspects of business - A weekly newsletter that delivers Knowledge at Wharton insights directly to your inbox
Visit the Knowledge at Wharton homepage: https://whr.tn/3qUpSBE Subscribe to the Knowledge at Wharton YouTube channel: https://whr.tn/3DEAgAU
Read the full transcriptHide the full transcript
- Welcome, welcome everyone to the Wharton Podcast Network here on Wharton Money Ball. I'm Eric
- Bradlow, professor of marketing statistics and data science here at the Wharton School.
- I'm joined today by my two longtime friends, collaborators, and co-hosts, Adi Weiner and
- Shane Gentian, professors of statistics and data science. Some combination of the three of us,
- and Cade Massey are here every week on Wharton Money Ball. Guys, I've always said from the
- that the best part of our show is we get to interview people that are doing what we do for
- a living, but actually in the real world, if you'd like. I mean, we have day jobs and our next guest,
- Michael McCurdy, longtime recurring guest here on Wharton Money Ball, everybody knows him. He's a
- mathematician. We were just discussing off air what he teaches, but you would expect a mathematician
- at a university to teach. He was educated in Canada, England, and Australia. He returned to
- his home of Canada to raise a family, teach math, and of course the most important part of his life
- to help us better understand hockey. He runs the, and I don't know why he put this word in his own
- description, the moderately popular, we think it's very popular, hockey visualization site,
- HockeyViz. He lives in Halifax, Nova Scotia with his wife and two children. Micah, welcome back to Wharton Money Ball. Thanks for having me.
- Well, it's great to have you back, and you know, this is an exciting time for hockey.
- It really is. It's, you know, it's a big part of the season. Let me just start with the basics.
- Let's start with HockeyViz, which you could imagine, this might stand for visualization,
- and I know a lot of the work you do is visual. Let's, for our listeners that haven't heard you
- on the show before, what made you think about not only doing, let's call it the analytics of hockey,
- but representing it in a multi-dimensional visual way? So the really short origin story is that I was real homesick for Canada doing my PhD in Australia,
- and I missed having hockey around. And I was a pretty casual fan in Canada, but then in Australia,
- I found because I missed it, I reached out to it and got a lot more interested in it,
- and wanted to do some little simulations to figure out how likely the Senators were to take,
- you know, 2-4-6, somewhere in between points from a California road trip. And so I thought,
- well, I'll have to simulate that, and the teams aren't the same strength, and then it kind of all
- got out of hand. That was 15 years ago now, and it was just a hobby for myself, just for friends,
- just a little, you know, a little curiosity. But the visualization part came about not because of
- any real deliberate strategy, except the fact that I, despite stereotypes, am not particularly
- fond of numbers. In particular, I don't understand information by reading numbers.
- And so some of the academic training that I had from some of my previous stuff was all strongly
- emphasizing just graphing data all the time. Just if you think you understand what you're doing,
- just make a graph. And that you can use your visual intuition about two-dimensional space
- together with your symbolic intuition with your other knowledges. And so it was, at first,
- it was just a debugging tool. I just didn't understand anything until I saw it in a picture.
- And then I started sharing it first on Twitter, now on Blue Sky. In fact, first on Facebook forever,
- then on Twitter, and then on Blue Sky. And gradually people started to say, oh, you know,
- can you look into this for me? Can you look into that for me? And so what started as purely
- descriptive statistics so that I could understand my own simulations, you know, just sort of slowly
- snowballed. But it's not sort of deliberate, like I'm going to sit down and make a business.
- That part came afterwards to my great surprise. Although my wife doesn't like it when I say
- that out loud. She says, I always knew it was going to work, and I shouldn't be so self-deprecating.
- So there you go. Let me ask you a question. A lot of our listeners here on Morton Moneyball might say, wait a second,
- I thought you said that Mike is a mathematician. What's he doing all this simulation stuff for?
- So could you tell us like how much of what you do, and look, I'm only looking at Adi,
- I'm looking at for our listeners here, Adi is trained as not only a statistician,
- but a probabilist that spent maybe the first half of his career, I'll call it proving things,
- deriving things, et cetera. And now we live in a world of simulation, which can allow us to
- model richer things, but maybe not in a more as general a way in some ways. How do you think of
- the role of simulation in your work versus mathematics in your work? Well, so I made a serious career change away from science into mathematics,
- into pure mathematics. In fact, my PhD is in extremely pure mathematics. And while I love it
- still, it does not engender itself to study employment, even more so than all the other
- kinds of mathematics. And moreover, I found in addition to that homesickness, I found that I was
- also kind of jonesing for that computational aspect. In some of my high school work for myself,
- and in some university work too, I did some physics papers I wrote with a man named Andrew
- Rutenberg, who's still a mentor to me. We're singing a choir now together. But he's a simulationist.
- And that's his primary work. And he taught me a lot about simulation physics. And so it's funny,
- I left physics behind, so I thought at the age of 22, only to recover it 10, now 20 years later.
- And part of it was the sort of fun of just saying, well, here's a little gadget. What happens when
- it has a kind of a crafty aspect to it? You don't have to convince me about the joy of
- figuring something out from first principles and then convincing somebody else of it,
- just on a sheet of paper or just on a whiteboard, like that kind of thing that's really satisfying.
- But it doesn't have that kind of like, oh, here's a little thing, what does it do?
- And so simulation was part of why I chose that project as like a fun project. And then from
- there, it just gets more and more powerful. Once you have measurements for players that are half
- decent, now all of a sudden your simulations are a lot more interesting than just here's a fun little
- toy. And because it grows naturally like that from a little toy into something serious, which you
- don't get so much with a kind of like, I'm going to sit down and give you a treatment of some
- subject from the bare metal. If you can do that, then so much the better. But you can kind of
- bootstrap your way into something interesting with simulation work. And that's no pun intended
- on bootstrap. This is the right show. That was a good pun. Hold on a second. Wait, wait. I got
- a lead in for you, Adi. You're next, but I got to make a lead in. I'm sitting here knowing Adi for
- 25 years. I have no idea whether he's going to ask you about math, simulation, or you're singing
- right now. It could be any of the three of them, given who he is, but Adi Weiner, please.
- Yeah, I'm going to pass on the singing, although I'd love to talk about that offline.
- But I'm actually, you know, I'm preparing for my course next week, which is on sports analytics.
- It's to our NBAs. And so I want to teach them, if they had more math, I would show them how to do
- empirical Bayes, conjugate priors, and forecast, say, next year's or the second half of the
- season's batting averages or whatever it is that you're doing with sports analytics in the classic
- mathematical framework. But I can't do that because I don't have that time and they don't have that
- knowledge. So I've discovered that I could teach this purely with simulation and essentially first
- principles. And it's remarkable. You can use split samples and use, oh, you have to do a little bit
- of math just because you can't get away from it 100%. But it's incredible what you can do. And
- there's a whole school of statistics teaching, which is really apropos to sports, where you
- essentially dispense with the asymptotics and all the glorious work that has made statistics possible
- for so long because we didn't have computers and replace it with things like bootstrap,
- permutation tests, split samples to get prediction intervals and confidence ranges
- and bootstrapping. And it's remarkable. And I wonder whether or not all that theory, which was
- wonderful for us to get us to where we are, will eventually just be just one of the tools in the
- bag and not the main event. So it's funny you should put it that way because I have been trying
- to persuade my university to let me offer a course a little bit like that, targeted not at MBA folks,
- but instead at people who want like a funny quirk on a math degree. Would you like some
- sort of data science-y kind of angle? Would you like some sort of... But I, despite being a
- mathematician, pure by training and applied by work, I'm not a statistician. And I've made a lot
- of friends with a lot of statisticians, but every now and again, when they talk real turkey, I have
- to say, sorry, you have to prove this to me. I don't know what to talk. And so I'm in this sort
- of peculiar situation of knowing a lot of stats, but not actually knowing any stats in a way.
- I'm going to give Shane the next question, but this will be the last generic question. Because
- actually I want to talk to Mike about this year's hockey, but he's so interesting just in the way
- you're thinking about solving stuff. But please, Shane, go ahead. And of course, another pride of
- Canada here. It is another generic observation, Tom, maybe feeding off of what Adi said about
- how we're doing more kind of simulation-based teaching and less of teaching maybe the
- asymptotics. I kind of feel like even when I was teaching in intro courses, the asymptotics,
- I was teaching it kind of via simulation as a thought experiment. I feel like when you think
- about the sampling distribution of the mean or something like that, you're automatically kind
- of doing a thought experiment simulation in your head anyway. I kind of feel like it's maybe like,
- maybe they're kind of more deeply linked even than that. But regardless, that was just kind
- of the observation. And I feel like maybe my background or whatever, I always taught
- asymptotics via simulation anyway. Sports really encourages that. Yep. There's no question about it. So, Michael, here's the first question. It's not on your list,
- but given we're in the home soon to be of champions, the Flyers. So let me ask you to
- think about the Flyers. No, but I have a very specific question. Okay. So you come up with a
- visualization that says right now, maybe the Flyers have, what is it, about 60 something percent
- to beat the Penguins? Something like that. Yeah. Okay. So when you think about that, there are two
- factors I can imagine. One is they're up two to nothing. And so what you do is you compute the
- probability that they would win given they're up to nothing. Okay. Another thing is that you're
- going to do that because they are up to nothing, but have to adjust your assessment of their
- underlying theta because they're up two to nothing. When you think about this, how stable do you think about the strengths when you're coming up with these probabilities?
- So I actually take a pretty doctrinaire approach to this kind of thing. And it may not be what
- people expect. In particular, the attitude that I take is that everything that you're measuring
- is either a person or it's geometry. And you can't do physics maybe even more broadly than
- just geometry. And so the people in particular, I understand to change really slowly. And so if you
- have a good bead on the ability of the players, that's a big if, but if you do, then the fact
- that the Flyers have won one game or two games or even 10 games weighs surprisingly little on the
- ability of the players. And so the simulations that I'm doing, I'm starting from, this is how
- good I think the people that the Flyers are going to put on the ice are, as well as this is how good
- I think their coaches, this is how good I think anyone else who makes a relevant decision for on
- ice stuff. And so I try not to include anything about laundry, if you like, anything about the
- teams themselves, like I, as such. So I- Is the whole equal to the sum of its parts? Are there interaction effects like between,
- like how do you, let's imagine I give you all these strengths of the players, the coach, etc.
- Do you a little, is it a summation or is it something more complicated than that?
- Well, so, so in my case, the, the, I have a simulation model at another level where I'm
- simulating the game itself, how, who is going to play with whom, who is going to actually be on the
- ice against whom. And so I'll just run any number of simulations of the game. So, so it's not just
- that I'm simulating the remainder of the series, but, but then of course, and of course the shifts
- are also in a way being simulated once you've, so it's sort of simulation all the way down. And,
- and that, that if you like is just physics training, I suppose. But I find it really
- satisfactory to explain, at least to myself and sometimes to other people too, even though,
- of course it has to be simplified in places, you know, it's not actually the hockey game.
- I saw Shane's hand up. I just want to ask a clarifying question on this simulation,
- just to be clear. So let's imagine I've got, I'm even showing you my lack of knowledge.
- There's a goalie and five, six players on the ice at once. Five as a rule? Five. Yeah. Goalie and five. If I told you the strengths of this five players and the strengths
- of this five players playing against each other and the strengths of the goalie, would your
- simulation then output a probability of scoring? Like those are the primitives and that's when you
- simulate and players are going in and out, but at any point in time, there's my six strengths
- against your six strengths and that's, what's determining what goes on. Yep. In fact, it's, it's sort of two or three models, but then taken together, they do exactly
- that. Ah, great. Sorry, Shane. I saw your hand up. Yeah, no, and obviously you're, you're, you're modeling and simulating this at kind of a,
- at a, you know, shift by shift level, it sounds like, as opposed to kind of like a, you know,
- a series by series level. But I, I wouldn't kind of, when I, I thought about sort of simulating,
- say the hockey playoffs. And I, I specifically have been thinking a bunch about this kind of
- the president's trophy curse or whatever, the fact that, you know, we often have the top team
- in hockey knocked out in the first round, or that seems to happen more probably than it should.
- Um, and it got me kind of thinking about Eric's, uh, frequent, frequent kind of comment on these
- shows, which is momentum versus, you know, kind of badly, you know, a simulation of hockey
- playoffs where the two factors are basically either the quality of the team or the momentum
- of the team. And I do kind of feel like hockey playoffs are uniquely designed to kind of pair
- because, you know, the top teams by Colorado, say, for example, are always kind of going up
- against wildcard teams. And could we see, you know, so basically is there kind of a curse to
- being in the top place in hockey because you get paired up against the team that is most likely
- going to be on like a momentum-based run going into the playoffs just because of that kind of
- one versus eight sort of seeding. It doesn't seem implausible, but, but I have a sort of asceticism
- of sorts where, where if, if there is such a thing, then I, I don't quite consider it legitimate to
- include until I can model it somehow, you know, where, where does it, well, let me just tell you,
- it's real. All right. But sorry. I don't, um, the, yeah, I, I guess the way, the way one,
- if it existed, I agree. If it existed the way I think you would build in the model is you said
- before that your primitives like player ability to player ability essentially are very slowly
- evolving, maybe not actually even evolving once you get to the playoff parameters. And this would
- be kind of like, it'd be kind of like doing, doing some kind of more like, kind of like short-term,
- you know, maybe for a subset team, some of those parameters are, are changing more rapidly.
- Um, that, that would be kind of how I think I would build a concept of momentum or something
- like that. What you could do is we're in, in Mike, maybe Micah does this already, which would be,
- let's imagine we, we take, we're sitting here at game 82, just before the playoff start, right?
- We're going to take a window going backwards. Maybe that's all 82 games. Maybe it's a shorter
- number of games. You could fit the window size that gives you the best predictive ability
- historically over the dataset. So another simple way to do what Shane's talking about would be
- to just take model abilities in some more localized window around the playoffs and call that,
- if you want to call it non-stationarity, I'm fine with that too, or some form of momentum.
- So I actually, I don't do this right now, but I have done it in the past. Um, and, and, uh, in, in fact, I found this anecdotal, of course, but I found that 25 games,
- um, was a useful window to look back. Uh, not, not instead of, but in addition to a longer
- time, um, uh, precisely how those got mixed is a little bit ad hoc. Um, this is part of why I left it behind. Um, but, but, but now I use a sort of more
- instantaneous, like kind of trying to get a beat on player ability. That's a little bit more like,
- like the speedometer in your car, like, um, where it's understood that it's, it's a continuous measurement. And, and then I suppose if you were going to layer on more
- characteristics on top of that, you know, the, the question is going to be, well, how, you know,
- and, and so you could, you could just take a window like that, but then I always sort of want to get a little bit deeper into it, you know, into some, maybe it's something
- about coming from science rather than coming from statistics is that, is that if I can't
- get ahold of something where I can say, look, this is, this is what people are like, specifically
- these interesting people, you know, with their extremely fancy jobs and, and, you know, where
- they play hockey for us to, for our amusement, you know, that, that aspect of personality,
- even if I have to get at it very clumsily, I would still rather try and fail than just say,
- well, let's just throw some big number on it. I like it. Yeah. Adi, please. All right. So I've got two questions. Um, I'll start with this one because it's been
- burning on my mind based on a competition that we just did with high school students
- with a simulated hockey season. Shane was, was part of this as well. Um, and it has to do with the goalie. So we built a model where the goalie effect was actually quite
- large. It was simulated hockey, not real hockey. And part of the reason we did that is that if I,
- based on real data, um, if I told you, for example, that the, the two factors about a team,
- one was the goalie's historical G over XG. In other words, how many goals did he give up versus
- how many shots would you expected versus your opponents, you know, the XG differential for
- the two teams. So like the quantity of quality shots they're been making, they both were kind
- of equal in terms of their predictive quality. Um, I thought that the goalie would get,
- get trounced by, you know, the quality and number of shots that each side was taking,
- but it turned out the goalie is really, really important in describing what happens in the past.
- And I'm really not sure, sure whether that really describes what happens so much in the future.
- Um, or is this another goalie performance? We've talked about that on our show is very important
- in figuring out who wins a match, but, and I'm, but I'm not so sure how important it is in
- predicting who will win matches. Um, and so what, what we hear is goalie actually matters a lot
- and goalies can get hot or is that hot because it's non-stationary or is it hot because it's
- just a psychology? It's not really hot. It just looks like good, good, good work. Or is it, um,
- uh, you know, uh, or is it genuine psychological heat of the classic momentum definition?
- So I'd love to hear your opinions. And once you do that, when I have a second question,
- so I want to cue that up. I, so I, I, on the one hand, I strongly suspect that, that there is,
- that there is something just latent, just that we're not catching that I don't have data for,
- or that I'm not perceiving correctly, you know, a really classic hidden variable, kind of, you
- know, it looks blurry to you because you're just not resolving the things that you should be
- resolving. There are features here. You know, how much time does that guy have to take that shot?
- How much time does that goalie have to actually make a read? You know, that, that kind of stuff,
- like how occluded, how good is that screen? You know, all kinds of stuff like that, that we're
- just, you know, that we were getting really, I'm getting really clumsy proxies on if, if anything.
- And, and so, but on the other hand, it's, it seems tempting to me to imagine that you could have
- some kind of, some kind of literally psychological treatment there, where especially, part of why
- it's on my mind actually is because Lena Solmark, the goaltender for the Ottawa Senators, took a
- leave of absence to deal with anxiety problems in the middle of this regular season. The, and,
- and then was extraordinary last night. And, and it's not just, you know, it's not just unusual for,
- for a male athlete to do such a thing in the middle of a pro season, but it's also striking
- because you could see him playing badly before it. And then he took a good-ish chunk of time off.
- I don't know exactly what kind of care he got, but it seems to have been reasonably good. And
- now he came back and he's been much, much stronger. And so that, you know, anecdotal, I suppose,
- but it points to a reasonably plain psychological explanation for on-ice results. The, you know,
- the mechanism of which I think is going to be really tricky to model. Adam, you had a second question?
- Yeah. Here's my second question is, so when, I love to be what I would call smart, stupid
- modelers. So what does that mean? Take a sport that I don't know much about that makes me stupid
- and then model it in what I would call decently smart techniques. I've done this with like
- football, for example, and I've demonstrated pretty convincingly that while I can't beat the
- smart, smart modelers, I can get pretty close in terms of forecasting probabilities of outcomes
- with very, very simple stuff. So in hockey, my question is, if I were to just model teams based
- on score differentials, you know, I mean, a schedule adjusted score differential type thing,
- not, so that's what I mean by smart, stupid. How would I do in terms of predicting outcomes of
- these playoff games, as opposed to someone who did fancy simulations and was smart, smart?
- What fraction of the signal do I get by being smart, stupid? A lot. I don't know. It's sort of not precisely well posed as a question, but something like 80,
- 85%. It's funny, the phrase that you use, the smart, stupid, I call physicist thinking,
- just because there's this really kind of uncomfortable habit of physicists to sort
- of wander into any discipline and just say, you know, have you considered approximating this with
- a tangent plane and like, you know, sort of swanning out after doing nothing useful whatsoever.
- But, but, you know, really basic simulation principles and a head on your shoulders can
- take you a very long way. Like I noticed, for instance, one of like I run a prediction contest
- for the regular season every year where I give whoever wins a lot of sour candy. That's for
- historical interest. But one of the things that you can really tell is that some people,
- some people are just winging it. They're just putting in whatever numbers based on, you know,
- whatever they feel like. And other people are actually doing simple things like making sure
- that the number of points adds up to a number of points that is possible for the, for the whole
- season, you know, or possibly going further than that and actually like simulating something,
- who knows. And so, and you see a really, really big difference. And then after that,
- you, so as long as you do something sensible, you can get sort of to the same place that all
- of the other people doing something sensible get to. And then, then the real struggle comes from,
- you know, now what do you do? What are you going to try to actually get some juice out of something?
- And there you find almost immediately, you know, well, let's try this. It doesn't help. Let's try
- that. It doesn't help. You know, you can, like I would say, if you've got, if you went one step
- further than what you said and actually used both goaltenders and finishers, so just measured with
- goals per expected. And then also something about chances, pure chances without finishing
- ability. Like if you, if you separated out those two features into two separate things,
- you would already be, you would already have something like 90, 95%. I think the other nice thing about, I want to move on to one other question out here,
- but I think the other nice thing about simulation of course, is you can answer a lot of other
- questions along the way that aren't necessarily, that's, I mean, that's the thing is that once you
- have, as Shane mentioned earlier, the, like the shift level, once you have a full simulation,
- all kinds of things you could answer. And in some sense, it's consistent and coherent
- because it's coming from the same underlying primitives. In fact, if you like, one of my
- favorite examples of this is that during, during one of the worst parts of COVID, when everybody
- was stuck in their bedrooms, I dug out the simulator and just loaded up the 1819 Detroit
- Red Wings. And I forget precisely which year, the darkest depths of the tank Buffalo Sabres
- and the two worst teams I think we've seen in the last two decades and imagined a seven game
- series between them. And I just got the simulation to sort of just tell me it's debugging output.
- And I just turned it into a story and, and people loved it. It was just, it was just a,
- just a hoot. There wasn't any real point to it, but once you have a simulation model,
- you can play games with it. You don't even have to do real work with it. So let me ask you a couple of questions from your list in the last few minutes that we have. So
- one of the things you wrote down is isolating individual player and referee impacts on
- minor penalties drawn and taken. So without knowing what you're going to do, I'm going to
- go back to my ETS days and say how I would answer the question. Okay. And then you'll tell me whether
- what I'm saying is totally stupid. Let's imagine a table where the rows of the referees and the
- columns of the players. Okay. And so we have entries when certain referees are calling penalties on
- certain players, right? In the educational testing world, one could be examinees, the other being
- test items. And we estimate row and column effects. So some players are more likely to get penalties
- than others. Some referees are more likely to give penalties than others. And this, depending
- on what model we fit, we can decompose the two of those things. So is what you're doing anything
- like that, or am I totally off base what you're thinking about or doing? Surprisingly similar in
- the end. It's slightly more sophisticated in terms of structure where I can like nest things inside
- one another. But broadly, yeah, that's the approach. You know, does your presence on the ice
- cause your team to take penalties or the opposite or neither? And I didn't get as
- far as actually putting individual referee names on stuff. I've been yelled at before.
- Okay. Okay. So, Michael, let me ask you another question. One of the other topics on here was age curves. And I love the way it's phrased, solving age curves alongside
- teammate and competition effects. Like, am I old or the people next to me suck or I just happen
- to be playing against really good people? How do we figure all that out? Well, so this is something
- that I have gotten some progress on. But so far, until very recently, the aging work and the sort
- of specific ability work has been separate. So I have like a model for who's a good finisher,
- a model for who's a good shot generator, a model for who's a good shot suppressor, all these other
- specific hockey skills. And then I can go and take those measurements afterwards and say,
- okay, now let's figure out an aging curve for those things. But of course, it doesn't take
- too long of doing that, in my case, a year or two before you realize you really want to do those
- things at the same time. Because among other things, you get into the what I call the moving
- elevators problem, where if you take an older, strong player who's declining, because they're
- and then pair them with a younger player who's improving. Sometimes you do this, sometimes,
- you know, you have to worry about bias all the time. Sometimes these things happen accidentally,
- you know, or sort of along the way. But other times they happen on purpose. You know, we like
- that as a coaching choice. We're going to put this guy, you know, his legs are no good anymore.
- So we're going to play him with a young guy whose legs are amazing. Also, this young player who's
- with amazing legs is actually a superstar in the making. So we hope and we would like him to
- get better. We'd like, you know, the specific example that I had in my mind from last year
- was Claude Giroux, Flyers legend, now in his older age for a hockey player in Ottawa, with
- being routinely paired with Tim Stutzler. And worrying that I wasn't getting the credit
- portion right, because you have two players who are staple to one another. And that's already a
- problem, except you know, in sort of generalities that the aging curves are taking them in the
- opposite direction. So if you take something naive, where you're not including the aging,
- the measurements that you're going to get themselves are going to be slightly wrong
- for that reason. And so I think, I think at last, I have figured out a way that I can start to
- integrate the aging curve measurements into the model measurements. And so we'll see if that works.
- Maybe you can have me back on in a month or two. But I considered this to be a huge pain in the
- neck until sort of all of a sudden, I have a model idea for a regression technique that might
- be able to solve with the computers I currently have. So maybe just in the last minute we have,
- so how do you see this Stanley Cup playoffs playing out? Like, you know, do you have a,
- well, let's wait a couple things. One is, according to your sims, there has to be a
- predicted winner by the definition of a sim, like it will give you a predicted winner of the playoffs.
- Both, who is that? And how certain are we? Like, I'll ask you in two ways. What's the probability
- of whatever team that is actually winning? Like how much higher is it than, I've got this right,
- there are 16 teams in the playoffs right now. And so how much higher than one over 16 is it?
- And then number two, how many teams would I have to take? We always like to ask this question
- so that it would be even money. Like I give you three teams and I get 13, does that, is that an even money bet? Or you have five and I have 11. So I'm just wondering,
- what's the distribution of winning percentages? So how do you view both? What do you see for both
- of those things? So first of all, the short answer is Colorado. They're the best team in the league
- this year. Their probability as of right now, and they're up one, nothing in their series.
- Their probability is about 37% to win the cup. That is unusually high for a single team.
- Most years, so to answer your second question, most years, if you wanted to make an even money
- field, it would be five versus 11 would be typical. Sometimes six, sometimes four. This
- year it's two. And the second team is Carolina, 20% as of the moment. And so then 20 plus 37,
- there's you're already over 50. And they weren't both over 50 or very, very close even before the
- playoffs started. Carolina, of course, is up to nothing on all the way to the first round.
- But that is unusual. This year specifically, there are very few bad teams and a lot of
- middling teams, some of which made the playoffs and only a couple good teams. And so they look
- really good by comparison to the very average field. Well, Micah, that's fascinating. Shane,
- before we let Micah go, any reaction to 37 and 20? Does that seem like, oh, my God.
- It sounds high. But I mean, you know, when you were first starting to talk about Colorado,
- I was thinking to myself, oh, well, what about like I was thinking, oh, maybe Carolina should
- actually have the higher percentage because it's maybe got the easier path. But the fact that it's
- basically the second highest anyway, maybe maybe if you could talk a little just very briefly,
- like the 37 percent, like you've got kind of Colorado almost twice as much as the next
- high, like Carolina kind of those are two relatively independent trajectories, at least
- the two of those that have. Do you like is it kind of in the privatives? Is Colorado really
- that much better? Yes, that's and that's part of why I don't I don't feel bad of saying
- so, even though, you know, for six, seven, eight years in a row, I would have said, oh,
- that's a mistake. Thirty seven percent, you know, that's far too high for a single team.
- But this year, I feel a lot happier about it because I've been watching the primitives,
- if you like, about, you know, Nathan McKinnon, Cale McCarr, the Wedgwood having a tremendous year,
- you know, a lot of just really high quality skaters. Maybe a corollary to Shane's question.
- Then we're going to let you go. If Carolina were to play Colorado, I guess that would be in the
- Stanley Cup finals, right? Yeah. What would Colorado's win probability? How much of that
- 3720 is coming from a head to head matchup if they just get there or how much of it is
- Shane's intuition that a lot of it could be the path to get there? So the path is easy for
- Carolina. That's part of. So it would be even greater. So you're saying then if the path is
- easier, then it's got to be even more extreme than 3720 in the finals. It is. Yeah, it is.
- It's more like 65, 62, maybe. I looked it up the other day and I don't have it right in front of
- me, but it's in that territory. It's hefty. Wow. Wow. Yeah. And Carolina are no slouches. They're
- extremely good. Huh? Well, Mike, as always, it's great to have you here on Wharton Moneyball. We've
- been joined by Mike McCurdy, as he told me before the show started, 100% time mathematician,
- 100% time father and husband, 100% of the time the creator of hockey biz. Micah, thank you again for joining us here on Wharton Moneyball. Thank you all very much.
- So please stay with us and join us right after the break. Welcome back. Welcome back to Wharton
- Moneyball here on the Wharton Podcast Network. Eric Bradlow, professor of marketing statistics
- and data science here at the Wharton School. I'm joined by my two longtime friends and
- collaborators, Shane Jensen and Adi Weiner, both professors of statistics and data science.
- And again, some combination of us and Kate Massey here every week on Wharton Moneyball.
- Guys, I just want to say, I'll tell you why I felt so good about that interview we just did
- with Micah McCurdy. First, he's great. And I think his insights are tremendous. But secondly,
- more so for Adi and myself. I mean, Shane, I expect you to ask intelligent questions about
- hockey, but I don't think Adi and my questions were totally moronic. I think we may have actually
- learned something over 12 years. What the hell? No, yeah. The only thing the only thing moronic
- was that expectation that I would only ask intelligent questions. But yeah, yeah, no, no.
- Either way, we're all doing fine. We're doing fine. So, guys, as you know, in the second half
- of the show, we always we typically do what caught our eye in sports. Adi, I'll start with you.
- There's lots to talk about. Baseball. Yeah, plenty to talk about. Any NFL draft coming up? Whatever
- you want to talk about. What caught your eye? There's a lot. I mean, so as you guys all know,
- everyone, our listeners know, you guys know, I try to catch most Yankee games. I actually went to a Yankee game. I went to a Phillies game. So two games since we last got
- together. It's always nice to go to a baseball game. So I'll just point out one thing general,
- one thing specific about baseball. First of all, the American League is getting its ass handed to
- them by the National League. Only four teams in the American League have a winning record
- as of actually as of before the American League teams played the National League team.
- Interleague play, you mean specifically? No, no, no. Just in general, their records.
- So it has to I mean, you'd think that there should be some more or less balance if they're
- only playing each other, but they're not. They do play a lot of interleague. So there are nine
- teams in the National League with winning records and only four in the American League. That was
- prior to last night's games. I didn't I didn't update what happened last night. And I'm thinking I'm because it's like the Yankees have the best record in the AL and
- it's crap. I mean, and there's only a couple other teams who are like hanging in there
- and and in the National League, like whole divisions are above 500. So just an observation,
- it definitely caught my eye. And I'm kind of wondering whether this is just the vagaries
- of small samples or is it really. Like last year, I feel like around this time, maybe preseason last year, we went in saying like, oh, well, all the basically all the you know,
- the top like there might be one or two AL teams that are in the top eight. Right. Right. Going
- in. So I kind of feel like this is maybe I mean, maybe it's not the same teams or whatever.
- But no, I kind of feel like this is a lot of the kind of elite player movement, at least that we've seen. It's gone to the NL. Well, the Dodgers have been sucking them up.
- And the Mets and the Mets, Shane Mets, you look at just to build on that, if you look at fan graphs,
- let's just go through it quickly. Top eight. This is projected wins by the end of the season. Last
- time I checked, Dodgers National League, right. Braves National League. Yankees American League.
- Last time I checked, Pirates National League. Cubs National League. Tigers American League.
- Mariners, I think, are still in the American League. Padres National League. Phillies National
- League. Brewers National League. So I've just named the top ten. Did you say Pirates?
- At four? Pirates are really good this year. They're shocking. Yeah. The Pirates are predicted
- 85 total wins. I mean, it's a four-way tie for fourth or whatever. The Pirates are projected
- 85 total wins right now on fan graphs, which is fourth highest projection right now.
- Yeah, I did. I said the Pirates. Interesting. All right. All right. I mean, the Bucs fans in Pittsburgh are quaking. This can't be right. There's got to be something
- wrong with the map. Whether you want to take the top ten where it's seven to three or the top five,
- it's four to one. The Yankees are the lone team sitting there amongst the Dodgers, Braves, Pirates,
- and Cubs. And so we're not even counting some other teams that are off to some really good
- start. Well, I mean, that's the thing is, you know, I mean, in part that's fan graphs. Fan
- graphs sounds like it's waiting tremendously towards in-season so far versus preseason
- expectations. If you had any kind of... No, fan graphs is pretty slow learner. What was the Pirates preseason? Were they in the top ten?
- It's just a little over 500 because right now the Pirates are playing roughly 600 ball,
- but fan graphs only has them going 520 the rest of the season, like 72 and 68. But you add that
- to the 13 wins they have now. To be fair, the Pirates are really building. They have a tremendous
- young talent, Paul Skeens at the top, but still only 520. We're not expecting them to be that
- good. Let's go back to you. I forget if that was your specific or general... No, so that was my
- first observation. The second observation, I'm going to bring this up every week until I get
- an answer. We can actually discuss this. We talked about it a little bit last week.
- What do you do with Trent Grisham, the Trent Grisham problem? I talk about this constantly.
- He did manage to hit a home run the other night. Here's a guy who I looked at his record carefully,
- six years of complete lower than mediocrity, barely getting a win over a placement.
- Last year, he had a couple. He was pretty good. The Yankees gave him a $20 million contract.
- I thought that's just insane. And of course, coming out of the gate this year...
- Is the problem that the Yankees actually have mediocre players at some positions?
- To me, it's that they would fall for... What's the problem here? To me, serious analytics staff shouldn't fall for recency bias,
- the way they appear to me to have fallen pretty hard for recency bias. And your model should take care of that. I haven't done the real gritty work yet.
- Maybe they're not looking at just performance data. A lot of things that the actual
- teams are looking at is this peripheral information that we don't get. I mean, we could get or we don't generally dig into the hard hit rate, the pull rates,
- the expectations of how you'll do in Yankee Stadium compared to... He was playing in San
- Diego before. That's a crap park. And he pulls the ball. And they clearly hit more home runs last
- year. But maybe they're seeing something that I'm not. But I'm just, as again, let's be what
- I called before, smart stupid. So I'm not going to take all the fancy information that the high
- tech people will use. That's the stupid part. But I'm going to use the information that you
- have smartly. And it just doesn't seem to me that that was a good move. And so that's mine.
- I'll just keep my Trent Grisham watch going for the rest of the season. Obviously, as a Yankee
- fan, I hope he cleans up and does fantastic. But as a statistician, I'm thinking not a good move.
- All right. That's my take. Shane, what caught your eye in sports? Maybe it's around hockey or it could be around any sport.
- Yeah, I'll talk baseball as well. But first of all, I just want to say, Audie, I'm glad you've got somebody else to kvetch about. Is that the right way to say it?
- Kvetch. You've got to say your units properly. Kvetch. Kvetch about besides Anthony Volpe. It sounds like you found a second target there.
- Yeah, definitely. But yeah, I guess I'll do another baseball one because we are kind of, I think, still
- in sort of the silly kind of like overly interpreting small sample kind of part of the
- season. The one that kind of stood out to me is Mason Miller. The closer for the Padres
- has faced 38 batters this season and struck out 27 of them. He's actually slumping, Shane. I looked at his stats like four days ago and he had struck out
- 23 of 29. He's only struck out four of his last nine. Shane, it's not looking good.
- K-rate coming down from like 15 or whatever the heck it is right now. It's just like.
- By the way, last time I checked, Shane, I don't know if this is true. When I looked, he had struck out 23 and walked zero.
- Yeah, I mean, his whip is like point three or something. No, but I'm saying he had struck out 23 of 29 and walked zero.
- So so I mean, obviously, we're all our reactions all say there's no way. I mean,
- this is not sustainable. And I guess it kind of leads me like I'd love to hear from you guys.
- What do you think the best sustainable relief like what the sustainable version of like this
- is like? What like what? You know, if you had to get a percent strikeout rate, that's like a whip, a whip, more like point like is whips going to be double that by the end?
- Oh, yeah. Yeah. Oh, yeah. But no, no, very many walks. Shane's bringing up three interesting points. One interesting point with three prongs to it.
- One is there's three things that are amazing by what he's doing. His strikeout rate,
- his era, which, of course, zero and his whip. Which of these three do we think will be the most
- outlying or, you know, in a multidimensional space, which of these will be most outlined by
- the end of the season? And is the combination of these three going to be even more outlying?
- Like he'll be at the ninety ninth percentile of strikeout rate, maybe era and rip, which in a
- three dimensional space will make it even more impressive what he's doing. That's what I'm
- talking about. Are you talking Manolobis distance? Yeah, probably. Yeah. Yeah. Collectively. I mean,
- listen, there's there's been absurdly good seasons by relievers over historically
- era's around or even less than one ridiculous strikeout rates. I think he's in line to be
- one of those. Will he compete? Will he continue what he's done so far? No, that's that's that's
- immortal. It'll it'll regress down. But is a long way to regress down and still be in the
- greatness area? Yeah. And I guess like I'm using him as a way of thinking about like what is kind
- of the what is the reasonable like lower bound on era for a full season and whip? And it's like
- got to be at least double what he's doing right now on the whip front, at least. So I'm going to
- stay with baseball with what my caught my eye. I've been texting you guys about this.
- So as if we didn't have to talk about another great thing of Shohei Ohtani. So he actually is currently on a 52 game on base streak. Now, that, of course,
- includes walks hit by pitches, other stuff. Now, just to remind everybody, everybody knows Joe
- DiMaggio streak is a hit streak. That's 56. The all time on base streak is Ted Williams.
- Eighty four. OK, the next highest after that, I think, is like in the low 70s. So Ted Williams
- is very outlying. But Shohei's 52 is the 28th longest all time, but it is the longest since
- A-Rod in 2004. So and the longest Dodger one ever is 58 games, and I think it was Gil Hodges or
- somebody like that. But like it's getting to the point where, you know, he's setting records at
- a very fast pace. And the other thing I did, which was fun for Chat GPT, I actually like Chat GPT's answer here. I said, which one's more impressive,
- Ted Williams, 84 game streak or DiMaggio's 56 game streak? And it did a bunch of calculations.
- First of all, I like that it reflected some uncertainty. It said, well, you'd have to make
- lots of assumptions, et cetera, et cetera, et cetera. But it basically said there's no
- comparison. It's what Adi said also on the text chain. It said, conservatively, DiMaggio streak
- is probably a hundred times as rare and non-conservatively, it might be a thousand times
- or maybe the odd one conservative. It's not even close, which it's orders of magnitude more rare,
- 56 hits than 84 on base, which was a consistent Adi with yours. Like you answered in 10 seconds
- or two seconds, not close. Well, I actually do that calculation in Moneyball Academy with my
- students in the full Moneyball Academy, the full three-week program. And basically DiMaggio had
- three things going for him that made it even remotely possible. Obviously a very high batting
- average. So when he did put the ball in play, he did- It's 357 that season. That season. And during the streak, it was around not even that much higher,
- closer to 400 during the streak. He also almost never walked, which is generally a bad thing
- for baseball players, but a great thing for hitting streaks. You really need that low walk rate
- in order to get the ad bats to generate these hit streaks. And frankly, nobody does that. No modern
- player today doesn't walk. In fact, DiMaggio was asked in some level, why don't you walk?
- And it really reflects the changing understanding of the game. He said, the fans don't come to see
- me walk, which is an interesting question. It's almost like softball when you're playing
- pickup softball. Nobody wants to walk. You want to get to hit the ball. That's not major league
- baseball. You could also make an argument, Ted Williams had the most famous vision of any player.
- Ted Williams walked a ton. Yeah. Well, it's funny because he also had the completely opposite
- attitude towards the fans. DiMaggio said, they're coming to see me. I need to be as good as I can
- possibly be every single time. And it was quite of his part of his personality. Why does he play so
- hard? And he would say, because someone might be here to see me for the very first time. And Ted
- Williams, his attitude was, I don't give a crap. He was so notoriously anti the fans. And I mean,
- so much so if you've read a biography of Williams, it's fascinating. But Ted Williams would say
- repeatedly, I'm here to hit and I can't hit a ball that's out of the strike zone and no way I'm
- swinging at it. And even more tellingly, I can't hit a pitch that might even be in the strike zone.
- It's just in a place I can't do anything with. And that's a testimony to his brilliance as a
- hitter, which really was, I think, probably unsurpassed by anyone. I hate to admit this,
- but yes. But so DiMaggio had those three things together, the high bat, I didn't say the third
- one, the third one, which kind of, which is actually also very important. He almost never
- struck out. So he didn't strike out. He didn't walk. And when he did put the ball in play,
- he hit it hard. So, so many opportunities to get hits, whether those were picking up infield hits,
- because he didn't strike out, whether not by not walking, he had, he would typically have five,
- sometimes six at bats in a game. And Williams would have three. It was very common, right?
- So you can't have hitting streaks if you're only batting three times a game. And finally,
- of course, he hit the ball so damn hard and getting and on the line. So he got a lot of
- hits that way. So when you talk about the probability of 56, it's 100 times greater.
- And that's for show, hey, you're just not going to see 56. It's never going to happen.
- They said it was something of 250 years. It's done. Nobody hits that way anymore.
- They said Ted Williams could be broken. You know, maybe it's 25 years, 30 years,
- 40 years. But I mean, that's a, I know people say, wow, 40 years, 250. Oh, that's a big difference between 40 years and 250 years.
- The only one who had a chance to beat DiMaggio was Ichiro. Same, same style. Right. Right. All right. Adi, you're next. Let's go round two. Anything else caught your eye in
- sports this week? Anything caught my eye in sports. So I haven't really been paying attention
- much to the playoffs. I was looking a little bit at the NBA playoff sort of lineup. And I,
- in preparation for the class in previous two seasons, there was an extremeness about the
- Celtics two years ago when they won insofar as that they were not only the favorite during the
- regular season, they significantly underperformed in terms of wins, their expected wins using a
- Pythagorean-type identity. In other words, they did great, but we expect them to do even better.
- And they won. Last year, of course, there was just, they were just saying, I guess, who won
- last year? Jason Tatum. Jason Tatum went down with an ACL, with Achilles injury.
- Right. And it was totally, totally different. But last year, it was almost as expected. It
- was tight that there were three teams that all were extraordinary in terms of their differential.
- This year, again, all three top super teams were, none of them over- or underperformed during their
- season. So I think it's going to be, I think, I mean, it's pretty lopsided. It's surprising.
- The real good question is, would you take the field this year versus the other ones?
- I mean, in our previous hockey discussion with Micah, we were talking about, oh, how many teams
- would you have to take to cover 50% of the chance of winning the Stanley Cup? And it was like,
- you know, two in this year, but like, you know, usually three, four, something like that.
- It sounds like you, well, the last couple of years in the NBA, if you give me three teams,
- I can cover 100% of the probability of winning it all. Getting close to it. So yeah, last year, this year, I think OKC is responsible for over half
- easily. It's not half yet. I mean, betting odds, they're plus 150. But, but, and so it would take
- the second team, which is the Celtics, by the way, which are like at plus 500 or 550, it would
- take the two of them to get over 50%. But, and I think it would be hard for me to argue. No,
- what's interesting to me, actually, glad Shane brought up Micah, because I think everybody here would agree, in totality, the stronger teams are in the West.
- San Antonio's in the West. They had a winning record against OKC this year. Denver's in the West.
- They had a winning record against OKC this year. They're all in the West. But still,
- OKC is plus 150 and Boston's 550. It was the same question Shane asked Micah, how the hell is Colorado, you know, 62, 38 or whatever it is over Carolina, when Carolina,
- you know, Colorado in some sense has the harder path. Well, OKC has the harder path,
- and they're still a three to one favorite. It's crazy how that distribution is so skewed in the
- NBA right now. It seems like, I mean, I know in general that happens in the NBA. It's a very
- dynastic. Well, there's so many teams, so many teams competing for the bottom. Yes. It just
- takes like a third of them right out. It's nuts. Yep. Yeah, that's right. So, Shane, anything else
- caught your eye? Well, I'll just kind of throw one shout out to Jose Ramirez. This is a baseball one
- that I noticed Jose Ramirez recently became the Cleveland Guardians all time leader in games
- played. So he is the player that has played the most games for the Cleveland Guardians slash
- Indians franchise in that franchise's history. Yeah. And actually, the list of kind of, you know,
- the list of most kind of like most all time leaders for across the different franchises
- are fascinating when it gives me the opportunity to also we can celebrate Garrett Anderson is
- actually the Angels all time leader in terms of games played. He just passed away this week,
- shockingly. Do you guys know who it is for the Yankees without looking it up? The most games
- ever in Yankee history? Yeah, the Yankee that's played the most games for the Yankees.
- It's not Derek Jeter. It is Derek Jeter. It is. I guess you do know. There you go. Well,
- I'm just saying you think Lou Gehrig, but he was he played every game, but for 16 years, not
- basically 20 years. Mickey Mantle played 17 or 18 seasons. Imagine a lot, much less. He only played
- 13. Yeah. Imagine only played three minutes, like five for the wars. Right. And so those are the
- long time. And then there's a bunch of pitchers. But pitchers don't play every day. Yeah. Yeah.
- So, I mean, all basically looking at just kind of the top five across base of kind of franchise
- leaders in terms of games played, it's it's all position players. I don't think they're.
- Yeah. By the way, just in case people are interested, the Red Sox, the Stremski played thirty three hundred and eight games to the Red Sox. That's top all time.
- And then Hank Aaron for the Braves is next. Then Stan Musil for the Cardinals. And she's ahead of Aaron. That's that's a surprise. Wow. Well, yes, that's right. Yeah.
- All right, guys. So just the last thing in the last minute or two, I can't believe the following.
- Well, if you know the answer to this. So, Adi, I'll ask you in basketball. OK, I don't know the answer then. OK, so let me just say, you obviously know every year they announce
- a defensive player of the year, right? I do. I couldn't name one, but no, that's not true.
- There's only been one in the history of basketball who's been the unanimous defensive player of the
- year has voted on. I'm not surprised, actually. Very surprised. And it was just yesterday.
- This year, it's women. I can't believe you're telling me I'll make it up. Michael Jordan,
- Will Chamberlain, whoever it is, has never there's never been another unanimous defensive player of the year. Well, but those players who are primarily known for their offense.
- Right. Yeah. But I mean, has there ever been anyone? Well, yeah, I mean, I guess you.
- I guess what we're saying is defensive talent somehow doesn't sort of stand out. You'd have
- to look at the distribution of these, quote, unquote, we're talking kind of a defensive MVP,
- right? So it's what the distribution of that those votes looks like. Yeah, I'm just surprised that there is even if it's just a hurting mentality,
- that there hasn't been a herd of votes that has gone towards one player, like, you know,
- whether it's I don't know, Dekembe Mutombo, who is like the five time defensive player of the year,
- was never the unanimous defensive MVP or I don't know, Scott. Yao Ming was Yao Ming really good at defense? Yeah, yeah. Yao Ming. I don't know if he was.
- Yao Ming. Yeah. Dennis Rodman, you know, never defensive, never unanimous. Either way,
- that just caught my eye. One last thing I'll end with, guys. So, so Alkaraz, you know, I love
- talking tennis, is injured. He has a wrist injury. He had to pull out of two tournaments now. He's
- saying he may not play the French. Now, if he doesn't play the French, here's my question to
- you. Two part question. How much odds do you give to Joker? Not Joker. Yes, that Joker. But how much
- odds do you give to Sinner? Remember, Sinner had like eight match points against Alkaraz last year.
- Sinner should have won the French last year. How much probability do you give to Sinner? You have
- to take Sinner well over the rest of the field, right? And number two, does it give Djokovic a
- chance if he's on the opposite side of the draw? Because then he only has to potentially beat one
- of them. Yeah, I guess. Yeah, I mean, I mean, my intuition, can you give me a little bit more
- like what before this injury, how would you, what would you have given? What would be the
- probability of somebody besides Alkaraz and Sinner winning? Oh, I would have said at most
- 5%, 10%. Oh, 10%. Yeah. I would have gone with that. Yeah. Yep. Yep. So I think,
- and again, it's kind of one of those things where, you know, this is what we call the
- violation of the IIA property in marketing, which means I'm not going to redistribute the
- probability in proportion to the market shares of the other brands. I'm giving it all to the
- other dominant brand. I'm giving it all to Sinner. Yeah, that makes sense. All right, guys. Well,
- that's been one hour of Wharton Money Ball. I'd like to obviously thank our guests, Michael
- McCurdy, but on behalf of myself, Eric Bradlow, on behalf of my colleagues, Shane Jensen and Adi
- Weiner, thanks to our producer, Deep Patel. Thanks to our other producer, Marissa Renna.
- Thanks to our associate producer and sound engineer, Dion Simpkins. On behalf of all of us
- at Wharton Money Ball, enjoy your statistics, enjoy your sports, and we'll see you next week here
- on the Wharton Podcast Network and Wharton Money Ball.
Brief Summary
Michael McCurdy discusses hockey analytics, simulation methods, and Stanley Cup predictions on Wharton Money Ball.
Key points
- The Power of Visualization. Michael McCurdy explains how he transitioned from mathematics to hockey analytics through visualization.
- Teaching Sports Analytics. Adi Weiner shares his approach to teaching sports analytics using simulation instead of traditional math.
- Lena Solmark's Mental Health Break. Lena Solmark, the Ottawa Senators goaltender, took a leave of absence for anxiety, highlighting the importance of mental health in sports.
- Modeling Sports Outcomes. Exploring how simple modeling techniques can yield surprisingly accurate predictions in sports outcomes.
- Colorado's High Cup Probability. Colorado has a 37% chance of winning the Stanley Cup this year, a notably high figure.
- Trent Grisham Watch. A fan's ongoing observation of player performance raises questions about team decisions.
Episode highlights
- The Power of VisualizationMichael McCurdy explains how he transitioned from mathematics to hockey analytics through visualization.“I just didn’t understand anything until I saw it in a picture.”3:18Apr 22, 2026
- Teaching Sports AnalyticsAdi Weiner shares his approach to teaching sports analytics using simulation instead of traditional math.“It’s incredible what you can do with simulation.”7:49Apr 22, 2026
- Lena Solmark's Mental Health BreakLena Solmark, the Ottawa Senators goaltender, took a leave of absence for anxiety, highlighting the importance of mental health in sports.“It's not just unusual for a male athlete to take time off for mental health.”21:23Apr 22, 2026
- Modeling Sports OutcomesExploring how simple modeling techniques can yield surprisingly accurate predictions in sports outcomes.“You can get pretty close in terms of forecasting probabilities with very simple stuff.”22:26Apr 22, 2026
- Colorado's High Cup ProbabilityColorado has a 37% chance of winning the Stanley Cup this year, a notably high figure.“This year, Colorado's probability to win the cup is about 37%.”31:39Apr 22, 2026
- Trent Grisham WatchA fan's ongoing observation of player performance raises questions about team decisions.“I’ll just keep my Trent Grisham watch going for the rest of the season.”41:44Apr 22, 2026
- Unanimous Defensive Player of the YearA surprising revelation about the history of basketball awards.“There’s never been another unanimous defensive player of the year.”56:01Apr 22, 2026
- Alcaraz's Injury ImpactThe tennis star's injury raises questions about the upcoming French Open.“Alcaraz is injured. He may not play the French.”57:20Apr 22, 2026





