How the NHL's New Cap Era Will Change Team Building

Knowledge at Wharton

Ask about Knowledge at Wharton

Episode Description

ABOUT THE EPISODE More cap space does not automatically create a better hockey team. It changes the value of every contract, draft pick, and roster decision.

Samuel Ventura, Vice President of Hockey Strategy and Research for the Buffalo Sabres, explains how NHL teams are preparing for a major salary cap increase. He discusses the analytics behind contracts and player evaluation, the promise of new tracking technology, and why core analytical thinking matters more than ever in the age of AI.

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.

----- 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
  1. Welcome, welcome to Wharton Moneyball. Welcome to a full hour of sports analytics
  2. here on the Wharton Podcast Network. This is Cade Massey hosting this week with two thirds of my usual co-hosts.
  3. Shane Jensen is in here from his home in Center City, Philadelphia. Audie Weiner is in here from his home,
  4. home office, both of them in their home office on the main line of Philadelphia.
  5. And I'm coming to you from Austin. We're recording on Tuesday afternoon as we usually
  6. do. We're gonna run our usual format this week. That means guest in the first half hour
  7. and open lines in the second half hour. We are delighted to welcome onto the show.
  8. I don't think it's Sam's first time, but it's the first time in a while we've got Sam Ventura here to talk hockey with
  9. us. Sam is many things to us. We've known him in many capacities over the years.
  10. Presently and importantly, he is Vice President of Hockey Strategy and Research for the Buffalo Sabres.
  11. That is, those are my Buffalo Sabres, by the way. I cut my hockey teeth living in Buffalo
  12. in the early 90s. I'm a big fan and they had a terrific run in the playoffs last year.
  13. Sam's also affiliated with Carnegie Mellon, kind of our sister operation across the state in Pittsburgh
  14. running a fantastic sports analytics group. He's affiliated with their stats and data science
  15. department where he also got his PhD. He, before being with the Sabres, he worked
  16. in hockey operations and research for the Penguins there in Pittsburgh. And he helps run the Carnegie Sports Analytics
  17. Conference, one of the best sports analytics conference in the business. He's a co-founder, probably, co-runner now
  18. of that conference. Our friend Ron Yurko is in that circle as well. Sam, maybe more important than any of this,
  19. Sam is single-handedly responsible for the volume of hockey content on this show.
  20. He shamed us. He has shamed us historically. If it- Justifiably so, justifiably so.
  21. So we were underplaying hockey for too many years. For too many years. That's true, that's true.
  22. And we've been part of your ball chain. I mean, we rely on you. The Canadians. I was using the Royal, I was kind
  23. of using the Royal We there. Of course it was on me. Yeah, of course, that's right.
  24. Well, we've rallied. I have to say we've rallied and Sam gets some credit for that.
  25. Sam, that's a long introduction. Thanks for making time for us. Yeah, well, thanks for having me on and
  26. the Sabres, thank you for your support, Cade. But yeah, no, the numbers are up here
  27. with the hockey guests. I think you had at least three in the spring, either leading into the playoffs or
  28. during the playoffs. So that was really nice. We had the soon-to-be champion general
  29. manager of the year on the show. That was a good get. That was prescient. It was fun.
  30. It is fun. We're doing hockey, guys, in the middle of football season because hockey is beginning tonight or
  31. even this afternoon, depending on what time zone you're in. Hockey's kicking off a little bit earlier, seeing
  32. September regular season hockey for the first time in a long time. Anyway, they added a couple of games this
  33. year, which is one of the reasons for the earlier start. But they got five games kicking off tonight.
  34. Buffalo's not one of them. When do you guys start, Sam, and where do you start? We start Thursday in Columbus.
  35. Okay. All right, Sam, I know you split time between Buffalo and Pittsburgh. Where are you at the moment?
  36. And how are you splitting time? This has been a pre-season. Hockey has a pretty intense pre-season in
  37. the month running up to season. What has been your schedule for the last month? Oh, a lot of watching hockey, honestly.
  38. Our pre-season kicks off with a prospect tournament in the middle of September, and then
  39. four pre-season games in six days, plus some more for our American League affiliate.
  40. So it's been a lot of watching hockey between games and practices, and it's been exciting.
  41. It's a shorter off-season than we've been used to in recent seasons. Right, that's a good thing.
  42. We're excited about that, and yeah, just geared up for the season here. So Sam, you came into hockey through the
  43. analytics route. The layperson might not have known that the person who runs analytics and research for Buffalo
  44. would say they've been watching a lot of hockey in the last few weeks. Talk to us about what role watching hockey
  45. has to play in your job and your understanding of the sport. Yeah, I mean, I've been watching hockey my
  46. whole life, but I think I learned early on when I was starting out in my career with the Penguins that scouts and coaches
  47. and players have a lot of knowledge that I didn't about the game. And so learning how to look at the
  48. game through the same lens that they look at the game is really important. You start to think of things in different
  49. ways. You start to actually, I mean, it's honestly one of the best ways of generating ideas
  50. about hockey analytics is sitting at a game, watching something, and thinking, huh, I wonder if
  51. I could quantify that. And usually the answer is, yeah, I can. And so watching the game is one of
  52. the best ways to generate ideas about analyzing the game in my opinion. Can you tell us a little bit about
  53. what it means to be in the front office in hockey these days? It feels like it's blown up a lot.
  54. There was this year, you could probably name the year. There was a year, 10 years or so
  55. ago, when hockey kind of famously poached all the guys who were doing public analytics, four
  56. or five guys around the North America. And all of a sudden, public analytics went
  57. quiet because they went inside in-house. That in some ways was the beginning, it
  58. seems to me, of in-house analytics, but that's very different now. What is the typical statistics and analyst research
  59. shop inside an NHL building look like these days? Yeah, well, first of all, I think there's
  60. been a ton of people doing great work publicly in the intervening years, and they continue
  61. to do so now, but. It did. Yeah, but yeah, I mean, working in hockey these days is a lot different than it
  62. was even when I started, yeah, I guess that's 11 years ago now. Staff sizes are up across the league, regardless
  63. of if you're talking about scouting or analytics or whatever. You know, you end up, like, I think
  64. maybe at the start, there was some, maybe some people were unsure about hockey analytics, and
  65. now it's pretty much everybody in the league is 100% collaborative with their R&D
  66. team, R&D staff, and what we're trying to do most of the time is provide objective information that we and others in the
  67. organization can use to make optimal decisions. So, I mean, I'm not breaking any news
  68. there with respect to how sports analytics works in any sport, but the change has just
  69. been that everybody is extremely collaborative and welcoming of information these days.
  70. Is it so much so, Sam, that the fact that a famously quantitative GM worked for the team that won the Cup last year,
  71. so Eric Tolsky came out through the quantitative route as well, headed that team, personnel power
  72. and everything else, and they win the Cup after knocking on the door for a long time, that would normally change the dynamic in
  73. most, the other four major sports would benefit from such a thing. Do you think it makes a difference in
  74. the continued integration of analytics in hockey? Maybe. I mean, I think we'll be able to
  75. answer that question 10 years from now when we see how it's all played out. I mean, there's only so many changes a
  76. single organization can make in the span of one off season, whether you're talking about players,
  77. staff, DM, anything. So, Eric did a great job. He's been working there for over a decade.
  78. And I think if they did have any advantages, they built those up over making good
  79. decisions over a number of years. And so, like I said, 10 years from now, we might learn if that helps other
  80. teams, but I think it's a hard question to answer at the moment when we're just sitting here today.
  81. You know, we should recap that Eric, we had him on this year before they won the Stanley Cup, and we welcomed him to
  82. the show, and he said to us, welcome me? I've been here before. He was one of our earliest guests.
  83. We got him just as this hockey blogger, and we just didn't connect to that hockey
  84. blogger that was on our show when we were first getting off the ground. I ended up seeing Eric Zelsky.
  85. Literally our first year, I think. It had been 12 years. And he said something snarky like, let's do
  86. it again in 12 years. But you know, Sam, I wanted to ask you a quick question. Obviously, you talk about the way analytics is
  87. really used to make the best decisions and optimal decision-making, but in most sports, the
  88. effect of analytics is visible in the way the game is played, managed, run. Obviously, baseball, some would say to its detriment.
  89. Football, people like to argue it's added a lot of, a lot of, you know, risk -taking, which we enjoy.
  90. Basketball, instead of, how about hockey? If I'm not a regular hockey watcher, but
  91. if I'd been, you know, I pop in and watch every now and then, what would I know?
  92. Is there anything I would notice in hockey that would be, you could say, oh, analytics
  93. did that, or had a role in that? Yeah, it's a tough question. I mean, I think, you know, one interesting
  94. trend is that the last three cup winners, so Carolina, Florida, and Florida, have played a
  95. relatively similar style. And so, I guess it'll be interesting to watch what happens this season and in subsequent
  96. seasons if other teams start to take up that style. And so, you know, I don't know if
  97. Eric talked about this in your interview with him, but, you know, the general idea is
  98. that, you know, you try to play a very aggressive style and you try to cover up those mistakes whenever you can.
  99. But, you know, the core tenet of aggression is sort of what underlies the styles of
  100. those two teams over the last three seasons. I would be very hesitant to say that
  101. hockey... Can you elaborate for the lay hockey watcher? Super aggressive means, let me guess, lots of
  102. forechecking, but then also, that's kind of obvious, but also more challenges in the middle zone
  103. between the lines? Is that right? Like, what does it mean to be a more aggressive style of play?
  104. Sure, yeah. I mean, aggressive in puck pursuit all over the ice is kind of the way to
  105. think about it. So forechecking, backchecking, you know, pressuring opponents in
  106. the defensive zone, et cetera. I would say, I would be hesitant to attribute that to hockey analytics, but I do
  107. think, you know, in as much as you can quantify what's happening, it might, you know,
  108. shed some light on what these teams are doing and it might cause other teams to try to play a similar style.
  109. But I don't think we know the answer yet, I guess, is the real answer to the question.
  110. But, you know, I think in the next few years, we'll have a better idea. I mean, analysts doesn't have to change the
  111. way the game is played. I mean, that's okay too. And in some level, but the answer, the
  112. reason for that could be because it isn't getting through the front office coach barrier, if
  113. you will, or because it's not really about telling them how to play the game. Or, I mean, maybe, Sam, you got an
  114. opinion on this. Like, you know, I feel like maybe relative to something like football, there's a few less
  115. like kind of in-game strategic decisions that are going to be kind of obvious to
  116. the viewers. Like one thing I think about over the last, say, 20 years is that definitely to
  117. bring in like pulling the goalie, I think much like earlier, you know, and I would
  118. assume that that was probably an analytics motivated kind of, you know, essentially strategic change that
  119. is noticeable, but that's like, you know, one small kind of, I guess, you know, there's
  120. more, there's more of those kinds of things in football where you have fourths and downs
  121. and various other things. Yeah, for sure. I mean, the pulling the goalie thing, you
  122. know, that's something that kind of came into play, you know, a decade ago or so.
  123. I mean, there've been other changes that, you know, for example, like I think 10 years
  124. ago, teams might've been more focused on controlling play, generating more volume.
  125. Now teams tend to be more focused on quality. But again, like these are changes that have
  126. happened hand in hand between data and, you know, the on-ice strategist, the coaching staffs,
  127. the players, et cetera. And I just, yeah, I mean, just to, I'm trying to give a fair answer to
  128. Adi's question. I wouldn't be comfortable attributing those to analytics on its own.
  129. I think, you know, these are data informed decisions that teams have made over the years.
  130. And I think the league has shifted, you know, as a whole, as a result, but, you know, it's always in collaboration with knowledge
  131. of the game and, you know, the other people who are, you know, stakeholders in these
  132. decisions. Are there any broad trends in roster building or is there always differences of opinion?
  133. And so there's always gonna be some people going left and others go right. Are there trends in roster building?
  134. Yeah, well, I mean, one, you know, hot topic right now that we've been monitoring closely
  135. is the, at least the expected unprecedented rise in the salary cap in recent seasons.
  136. So back in, I think it was January of 2025, the league announced that the cap would be rising about eight or 9%
  137. each year for the next three seasons. And so these are increases from, you know,
  138. I think it was 88 million to 95 and then to 104 and then to 113. And now there's a report that the salary
  139. cap for the two seasons from now will go as high as 127 and a half million, which would be a 12% increase
  140. over the preceding year and over a 50 % increase in the span of five seasons. So, I mean, that's a pretty big jump
  141. in the cap numbers. And, you know, I'm sure, do you guys remember what happened in the NBA 10 years
  142. ago when the cap went up? What was it? 33% overnight, something like that. I mean, I don't know if we can
  143. walk through that, but, you know, basically, you know, there was an explosion of high value
  144. contracts given to players that, you know, the perception at the time was that they didn't
  145. really, you know, they were unusual. I don't wanna say players deserve contracts, they
  146. deserve everything, but it was a very interesting overnight shift. And I think we're going through sort of
  147. a phased in version of the same thing right now, where, you know, these increases in
  148. the salary cap have big implications for, you know, how we make decisions. So, I mean, when we got that information
  149. in 2025, I actually had my staff do a deep dive and go through the scenarios. You know, let's pretend we're this team, let's
  150. pretend we're that team. You know, how does this change how we value our own unrestricted free agents?
  151. How does it change how we value our restricted free agents? Does this change anything for arbitration?
  152. Does this change how much a draft pick is worth? You know, what will the market for salary
  153. cap dumps look like? And, you know, we sort of went through all these scenarios and we made a bunch
  154. of predictions about how everything would play out. And, you know, we tried to sort of
  155. set a course for the upcoming, you know, handful of years based on that sort of logical thought process.
  156. I think we did pretty well in our predictions. And, you know, I think we've navigated the
  157. current cap increases pretty well so far. And, you know, it'll be interesting to see
  158. how it plays out moving forward. Sam, in football, I think of analytics and cap management as mostly separate.
  159. They obviously come together at about the general manager level for most clubs, I believe that's
  160. fair to say. Is it similarly divided in hockey or do people like you, or does that fall under
  161. your rubric? Are you in close collaboration with other clubs? Or is it more of a collaboration with
  162. the guys doing cap stuff? Close collaboration, I would put it. Yeah, I mean, we have actually two people
  163. who are experts in that area, two assistant GMs, Josh Flynn and Mark Jakabowski, who run
  164. the day-to-day of cap management and contract negotiations and everything. But we work in close collaboration with them,
  165. you know, every day and sort of modeling out what contracts might look like, what player
  166. values might look like, how things are shifting around the league, building tools for them to
  167. use to, you know, try to comp players and everything. So, yeah, I mean, there's a lot of
  168. work that goes into it. And, you know, my staff, you know, in particular, Matt Barlow has been huge in helping
  169. sort of guide us through some of that from a data perspective. And, but yeah, it's a fun aspect of
  170. what we work on. It's a, you know, one of many, I guess. It's super complicated.
  171. I mean, you'd want the cap guys to fully integrate with the analytics guys because it
  172. takes a lot of modeling to figure out the right, the optimal, even the shape of the contract when you've got those kinds of
  173. increases down the road, right? You can push a lot of money strategically down the road if you're gonna see those
  174. kinds of increases. Shane. Well, yes, yeah, kind of following up on that, when you were kind of sitting out,
  175. like how different types of teams would react to this kind of increasing cap? Was there, like, is there like a, do
  176. you kind of expect over the next few years, like kind of almost like an increase
  177. in the variety of, like increase in the variance basically as certain teams, like, you know,
  178. like, do you have enough sort of like prediction of that heterogeneity kind of coming out
  179. in kind of team strategy? Or is it? Yeah, I mean, it's a really good question. I think two things come to mind.
  180. I think first there's been a lot of public research on, you know, modeling contracts and
  181. whatnot. I think a lot of those tend to ignore what's probably the most important factor, which
  182. is how much money is available in the system, right? So during the COVID era, where the cap
  183. was flat for a few years, players were, you know, if you just kind of washed away what was happening and just assumed contract,
  184. you know, contract would be assigned as they always were, you would systematically underestimate or overestimate
  185. the contract values. And now, you know, with the cap rising more than it has historically, you're gonna systematically
  186. underestimate it because, you know, again, the amount of money available in the system is going
  187. to dictate, you know, how teams behave. So I think, I mean, that's sort of like a economics theory that, you know, I
  188. think very obviously came into play over the last, you know, six years, I guess.
  189. But then the other part is just that there's a big premium on players and on talent, you know, more than ever, right?
  190. So you can have 30 million in cap space, but if there's no one to spend it on, then what good does that do
  191. you, you know? So identifying talent, acquiring talent, player evaluation just becomes so important.
  192. And, you know, that's, those are some of the core problems that we work on every
  193. day from a data perspective. Do you feel like the dynamics around that have changed in hockey since you've been involved?
  194. You said going back 10 years, working with the Pens before you've been with the Sabres.
  195. You talk to NFL teams these days, and they're saying now what baseball said probably eight
  196. years ago, that there's just, the boards seem to be similar across clubs now, that teams
  197. tend to draft the same, they seem, i .e., they seem to be valuing more similarly
  198. now than they used to. Have you seen the same kind of dynamic in hockey or is that, there's not enough
  199. commonality yet? Yeah, no, I think that's definitely true, especially at the draft.
  200. You know, it does seem like teams are more aligned with each other in terms of their draft list than they had been when
  201. I started. I think that, you know, you could also make the argument that's because there's been more
  202. use of video and data, right, so, you know, historically, you would have an area scout,
  203. scout a player, and, you know, that person's opinion was the primary source of information about
  204. that player. Now, you know, we have, we can have any scout in our organization do a follow
  205. -up video report on the same player so much more easily than we were ever able to historically.
  206. So I think that plus, you know, the influence of data probably has, you know, driven
  207. teams towards more similar draft lists. What's your sense of what role public data
  208. plays in that dynamic? As you mentioned, I mean, even though the public world got hoovered up 10 years ago,
  209. there's lots of good public sources, not lots, but a number of good public sources out
  210. there now, great folks we have on the show periodically. I think of this as happening in other
  211. leagues. Heck, I know baseball teams who in court now, even some football teams who bring public
  212. data into their player evaluations for the boards. There's just so much better and more public
  213. data. I'm assuming that plays a role in hockey as well. Yeah, I mean, I think it depends on
  214. what you're talking about. You know, like we're going to use our own internal models for player evaluation, but there
  215. is a lot of information out there publicly. You know, if you're talking about like public
  216. draft rankings and you can build those into your decision-making process at the draft to
  217. try to get an understanding of where players will be selected or how the market views
  218. these players. So, but you know, I think when it comes to player evaluation we're really only using
  219. our internal scouting systems and our internal R &D systems to make those decisions.
  220. Okay, Sam, this brings up the question of where edges come from these days. Like where, and you can only talk about
  221. this, you're not gonna talk about this in detail, but maybe you can talk about this
  222. in abstract terms. Why does your shop have an edge relative to other teams when it comes to, let's
  223. just stay with player evaluation? Like where does that, even abstract terms, where
  224. would that come from? Yeah, well, we don't know, I guess. We don't know if we have an edge.
  225. We hope that we have an edge, but we can never know that for sure. I mean, I think, you know, thinking about
  226. this from like a broader perspective, like one thing we've really focused on at all stages
  227. was trying to create knowledge about the game through the research that we're doing that will
  228. not only make us smarter as an organization, but it'll help with the player evaluation process.
  229. It will help with the strategy evaluation process. It'll help us prepare for opponents.
  230. So like when I think about the big changes in hockey analytics over the last few
  231. years, like it's the introduction of puck and player tracking data, thinking ahead, it's going to
  232. be, you know, we're gonna start getting Hawkeye data. So, you know, that skeletal pose data and,
  233. you know, what are the implications of that? So like, you know, we're gonna have a
  234. point in three-dimensional space for each key part of the body, you know, so head,
  235. shoulders, knees, and toes, but then elbows and hips and ears and wrists and, you know,
  236. points along the shaft of the stick and the blade of the stick. And so, you know, it really is gonna
  237. open up a lot of possibilities for what we can do with that data. And I'd like to think, you know, we've
  238. spent a lot of time, excuse me, thinking about that and generating ideas about that.
  239. And I'd like to think that, you know, we're forward looking enough that we'll try to,
  240. you know, maintain any competitive advantage we already have, or maybe we're way behind.
  241. Like I said, we can't really know for sure. Give us a couple of examples of what
  242. you're gonna be, just something you can, something not proprietary, but you will be able to
  243. tell with this new data. So I've heard, for example, that with the old data, you couldn't even know whether a
  244. guy was skating backwards or forwards because you don't have that kind of positioning, which is
  245. pretty fundamental in hockey, but you're gonna learn better stuff than that. So what's an example of something that you
  246. can do with this new data? I know there'll be a zillion, but like what's something people are talking about they know
  247. that you're gonna come up with? Yeah, well, I don't wanna minimize it. I am very much looking forward to knowing
  248. which direction a player is skating, but so that is very important. I mean, the low hanging fruit would probably
  249. be something like shooting mechanics, right? So, I mean, that's been studied in depth
  250. in baseball, golf, basketball, other sports. I'm actually a little skeptical that we'll be
  251. able to make any huge discoveries with respect to shooting mechanics. I think we'll be able to quantify it,
  252. but I'm not sure how much value it'll provide. I mean, hockey is just so different from
  253. other sports. In baseball, you have a stationary bat or swinging a bat. Each swing looks pretty similar.
  254. There's maybe some mild differences depending on the features of the incoming pitch.
  255. Same thing in golf, stationary golfer, swinging a club. Sometimes the length of a club is different.
  256. There's a different lie, the grade of the ground is different, but hockey is so dynamic,
  257. right? And that's what differentiates it from other sports. And so, there's all these other factors that
  258. will play a role in determining what the optimal shooting mechanics are in a certain situation.
  259. We're gonna try, I guess, and we'll see where we get. I'll tell you what my dream would be
  260. if I was working for a hockey team is looking into shot blocking. I feel like that's something that I think,
  261. probably teams are studying that in whatever data they have out there right now, but I
  262. feel like this is the data you'd kinda need to do a good job, I think, of studying whether certain players are good or
  263. bad at shot blocking. Maybe even whether or not there could be more optimal shot blocking kind of strategy for
  264. a particular team. I think that would be cool. I don't know if you have the detail
  265. to study the fletch. The same thing for like, is it gonna be so high res that you can actually
  266. tell which players are good at deflecting pucks and stuff? That would be cool too.
  267. Theoretically, yes. Again, we don't have the data yet, so I can't say for sure, but those are
  268. definitely two very good ideas. I'm gonna write them down, and I'll let you know in a couple years when we
  269. have the answer. Don't tell Edmonton. Keep them at Buffalo. Okay, Sam, I'll give you a bad idea.
  270. That's from a Canadian. This is from the West Texan. Do y'all already know something about, very
  271. common thing that happens in hockey is that a bunch of guys get in a scrum around the puck.
  272. The puck's loose. A bunch of guys get in a scrum, and then some team comes out with the
  273. puck. Do y'all have a sense of guys that add probability to your team getting the
  274. puck in those circumstances? It happens dozens of times a game, and so it's a possession question.
  275. Are there guys that are reliably better at that or worse? And if so, might this higher resolution data
  276. give you some, start giving you some indication of how and why that happens? Yeah, I mean, we definitely have a good
  277. idea of which players are adding value to their teams. In that way, we don't have any idea
  278. what the reason is, right? So like, I mean, you could create stats about your rate of battle engagement or your
  279. winning percentage in battles. And those tend to be very stable and repeatable across seasons, across teams, all that good
  280. stuff. We don't, if you think about what this looks like in data form, right? It's just a bunch of dots, right?
  281. Next to the puck, and then one of the dots comes away with the puck. We have no idea what's going on in
  282. there. So again, theoretically, with the Hawkeye data, we'll have the locations of sticks.
  283. We'll be able to see, maybe certain players are getting lower in establishing body position or
  284. getting more leverage or something like that. But we don't know yet. And so, I'll add it to Shane's list
  285. of good ideas to explore in the next year or so. We're gonna run out of time with you
  286. in a minute, Sam, but we wanna hear something about your upcoming talk at the Carnegie
  287. Sports Analytics Conference. Again, this is one of our favorite conferences of the year.
  288. It's always a late October thing in Pittsburgh, and you've got something in there about hiring
  289. in the age of AI or LLMs and hiring. So tell us what you're, give us a preview of what you're thinking about.
  290. Yeah. Yeah, so I'll be talking about this at the upcoming Carnegie Mellon Sports Analytics Conference.
  291. Please register if you haven't already. We look forward to seeing you there. So, I mean, I spent a lot of
  292. time this summer sort of immersed in the hiring process and thinking about what it means
  293. to find the best candidate in the age of AI and LLMs. And I mean, just to give you an example, who among this
  294. group is the worst programmer? Who wants to self-identify as the worst programmer here?
  295. Probably me. I'm the least officially statistical. Okay, perfect. I'm a decent programmer.
  296. I'm the least, if it's gonna transition into more vibe coding or using AI, I suddenly
  297. become the worst, because I just don't use it. Okay. Oh, Shane's out. Shane's out.
  298. You're killing my example, but five years ago - By that measure, I'm probably getting to
  299. be the best, because I do that. I don't vibe code, but I do a lot of AI coding. Yeah, yeah.
  300. So five years ago, I would say there's something like a 0% chance I would hire a Cade self-reported as the worst
  301. programmer. So a 0% chance I would hire you to work on my team, because at that time, high-end programming skills were way
  302. more important than whatever knowledge you have about sports analytics, whatever intuition you have with regard
  303. to stats, modeling, practical problem solving, all that good stuff. But now, I would be much more likely
  304. to hire you, Cade, because you have all the hard-to-find traits, that innate understanding
  305. of stats, modeling, how to apply it to problem-solving and sports analytics, and LLMs will
  306. give you a modern toolkit with respect to programming and creating production-level code.
  307. And so I'd be much more likely to hire you so it's sort of a shift, I think, in terms of the traits that
  308. are most important when you're thinking about bringing people on. So we actually did something really interesting this
  309. summer, which was we collected data on all of our candidates throughout the various stages of
  310. our hiring process. So just as an example, we do a technical assessment to kick off the process, and
  311. it's essentially a take-home exam. And we told people, it's totally fine to use LLMs, just we asked them to self
  312. -report the degree to which they used them, how they used them, et cetera. And we also made a subjective judgment about
  313. how much they used LLMs. And then subsequently, we rated them along a bunch of different
  314. categories, so relating to what they did in their technical assessment, how they performed in their
  315. interviews, et cetera. And we found some really interesting things. So I mean, first of all, I would
  316. say one really fun result is that the candidates who scored highly on the degree to
  317. which they used LLMs ended up underperforming expectations during the interview round, and that the better
  318. interviews came from people who had more of what I would call the soft skills. And I'd probably include some more core technical
  319. skills in that soft skill group, so just like an understanding of stats and modeling and
  320. whatnot. And so I think that's really interesting. I think it's useful for students who, I
  321. think it's probably fair to say many out there are over-relying on LLMs to learn
  322. and understand. You guys are closer to that than I am, but I've heard from my academic friends
  323. that it's a bit of an issue. It's a disaster is what it is, but anyway, well, we can, yeah.
  324. Yeah, and then I have a lot of really good stories as well about some of these interviews and whatnot.
  325. I mean, we had at least one person feeding my audio into an LLM and then reading the output live on the video call,
  326. and the misadventures that we had going through all this. But we collected data and we tried to
  327. be very objective about it so that we could learn from it and we could put something out there, which is gonna take the
  328. form of this talk and maybe a write -up afterwards that will help not just teams,
  329. but students and job candidates as well think about how to approach the sports analytics hiring
  330. process. Well, say one more line about that. What is your prescription at this point for
  331. those who want to get on, whether it's with a team or with a media group or whatever?
  332. Like how does one differentiate themselves these days as an entry-level analyst?
  333. Yeah, well, I mean, I should probably also add like I'm super biased, right? The things that I care about in my
  334. team members might be different from other people or people in other sports, but in my
  335. opinion, I think it's totally fine to use LLMs to help you get tasks done, but it doesn't replace the need to be able
  336. to communicate about your work, to explain what you did, to see the flaws in any
  337. automatically generated analysis, to think about things in terms of counterfactuals.
  338. Like that's just like a general idea that people apparently don't understand. Even the LLMs don't really understand it.
  339. So I would say like focusing on like really building a core understanding of how to
  340. use modern tools, right? So whether that's statistics, machine learning, LLMs, whatever,
  341. but you have to really understand what you're doing. You can use tools to help you, right?
  342. Like I'm not opposed to people using a hammer to get a nail into the wood, but yeah, go ahead.
  343. Let's take it one more step then because the question is you want them to understand
  344. how to use the tools. Is there any alternative in building that understanding to repetitions?
  345. Is that not ultimately what it comes down to? It's like you need reps using these tools.
  346. You've got to solve problems and then do it again and solve problems and do it again and as much of that as possible.
  347. That's always what I feel about data science. I don't think it's any different now that
  348. we have AI. Yeah, absolutely. And I think, I mean, just, I'll have some more complete results in my talk here
  349. but I'll just give you like just as one example of that, I don't think people these days, candidates these days really understand how
  350. to perform like a proper data analysis. So, you know, some absurdly low percentage of
  351. people who submitted these technical assessments to us did any form of model evaluation whatsoever, right?
  352. Almost no one did like the minimum level of exploratory data analysis, right? So there's like, there's these core ideas that
  353. we as, you know, statisticians, modelers, data scientists, whatever you want to call us, you know,
  354. we've been doing it for a long time. We've been living through that repetition. We know how to structure an analysis in
  355. a way that, you know, will lead to some semblance of truth at the end. And I get the feeling that, you know,
  356. a very large proportion of job candidates don't. And so, you know, again, come to my
  357. talk if you want to hear all the specifics on that, but it was eye-opening for sure.
  358. Okay. Good, good. Sam, we will let you go. Thank you for making time for us, especially
  359. on opening week. Enjoyed this first day, five games to watch around the league and good luck to you
  360. guys with your opening out on Thursday in Columbus. Also, we should have mentioned this when we
  361. talk about Tulski last year, let us continue through the tradition of interviewing a front office
  362. executive of the that season Stanley Cup winning team. Let us continue that tradition and we'll revisit
  363. - See you next year. See you next year, Sam. I hope you're right. Sam Ventura, Vice President with the Buffalo Sabres,
  364. affiliated with Carnegie Mellon, longtime friend of the show. Thanks, Sam. Welcome to the second half of this week's
  365. show. Cade Massey in here with Shane Jensen and Audie Weiner, Eric Bradlow, out doing Eric Bradlow
  366. things. He'll be back. Some combination of us are here almost every week of the year.
  367. We're talking 48, 49, 50 weeks of the year, working hard. Have been for 12 and a half years
  368. now. Just off the line with Sam Ventura, friend of the show, Sam Ventura. Mostly behind the scenes, some fun getting on
  369. the stage. Sam, Vice President R&D with the Buffalo Sabres after pulling the same kind of work
  370. with the Penguins. The multiple round winning Buffalo Sabres last year, the outperforming Buffalo Sabres last year.
  371. Fun, fun, fun. Hockey tonight, Shane. Have those Sabres had more playoff wins in
  372. the last decade than the Bills? Well, I don't think so. I mean, I guess game by game, yes,
  373. but not in terms of, if you did series to game equivalents or something. They've had a lot less heartbreak than the
  374. Bills, let's put it that way. Same owner, by the way, in case you're not paying attention.
  375. That's the same owner, which is a pretty strong affiliation. They just opened their new stadium.
  376. Buffalo's doing well. That city's doing well. Shane, five games while we're on hockey.
  377. Five games tonight. They did us a solid. They gave us a lot of, if you're gonna open, might as well give us some
  378. rivalry games. Look, the Habs and the Leafs are playing. That's fun. The Rangers and the Bruins, that's fun.
  379. Vancouver and Edmonton, I assume. That's kind of interesting, Shane. It leaves you out.
  380. It is, it is. Yeah. It's tough tonight. It's tough for the next couple of games,
  381. going up against the Wild Card round in baseball. We basically have hockey and baseball all the
  382. time for the next couple of days. Right, right, right. So it's that time of year, we start
  383. overlapping three sports. It's good fun. And then we'll have basketball rolling here, and
  384. for a week or two, we'll have all four underway. October's peak sports. Well, before we get to baseball, let's talk
  385. a little bit, because we can talk baseball for a long time. The Wild Card is underway as we speak.
  386. We'll come back to it. A little bit on football. Shane, Adi, did y'all take in any
  387. college football this weekend? Are you paying any attention? Shane, Adi, we've pulled you into pro football.
  388. I don't know how well we've pulled you all the way down to- Yeah, college is hard because of my Saturday non-watching
  389. properties. Yes, right. So I do try to catch up a little bit. I'll try to catch up a little bit
  390. on Saturday night to see if there's any big games, but I'm really mostly just watching
  391. a couple of things. I don't really see it. I'm just reading a little bit about it.
  392. I'm interested in Texas on your behalf. Okay, hold on, Shane. Give me some, I'm always trying to learn
  393. things from you. If you can use a bathroom with the light on as long as you don't turn
  394. the light on, couldn't you watch a TV if it happens to be on in the room as long as you didn't turn the
  395. TV on? So now you're asking a deep, deep religious question. The answer to that is complicated.
  396. So there's two, yeah, you could leave it on the entirety and just leave it on, and people have done that.
  397. And I could sort of find myself wandering in somebody's house next to our neighbor who
  398. has it on, all those things are possibilities as well. But then there's the third reason why we
  399. don't do it is that it really kind of, it's not the right spirit. That's not the right call.
  400. It's not in the spirit of it all. Yeah, I understand. Okay, but there's plenty of sports all day
  401. long. I mean, listen, yeah, college, but the other reason college is not in my bones and
  402. the way you guys, is Northeast is just not a college football, you know? Yeah. It's just not that.
  403. Yeah, and in fact, the best Northeast team lost unexpectedly this past weekend to Wisconsin.
  404. So that's an interesting little wrinkle. Which is the best Northeast team? Penn State, well.
  405. Penn State. And you're in, you're out. And that's like practically Midwest, right?
  406. Come on, let's be clear about Penn State. It's only four hours west of here. Yeah, yeah, yeah, Pennsylvania is a big, big,
  407. big, big state. If you had said Rutgers, I'd be like, oh yeah, that's right five minutes from my
  408. house. But no, that's not a good team. I'm glad we've operationalized the definition of Midwest
  409. in the eyes of a Northeastern guy, of an I-95 guy, of an Acela quarter, Acela quarter guy.
  410. I don't know where it is, but somewhere between the Acela line and State College, Pennsylvania
  411. becomes Midwest. So there's like a four hour window in there somewhere. Yeah, I mean, come on.
  412. Deep dark Pennsylvania is essentially Ohio. Okay, so there's a cultural aspect to it
  413. as well. I got it. All right. Okay, here's the thing about college football. Well, here's one of many things about college
  414. football. We've talked about this before. There's so many more games on a Saturday
  415. that the probability that something crazy is gonna happen is just higher. It's already a higher variance game than the
  416. NFL, but then you just get more rolls of the dice. So crazy stuff's just kind of happening in
  417. every viewing window of the day. It's just fun. There's always something that happens.
  418. That's worthy of a movie. Because I do kind of feel like college football, I usually go into college football with
  419. this like mixture model of games that are competitive versus games that are not.
  420. So if you kind of bend based on whether or not there's like a two touchdown point spread, is craziness gonna occur more often
  421. in the predicted to be competitive games or the not? No, you get both flavors. You get both flavors.
  422. Periodically, you get the true college football craziness is in that second bend, the not supposed
  423. to be competitive bend. But there's enough of those that eventually one of those is gonna turn.
  424. And that's, I mean, there's fun whenever they're tight games as well. But anyway, I'm pretty sure- For example,
  425. so Iowa State beating Michigan in the last minute. That wasn't predicted to be close.
  426. Or that wouldn't have been a blowout prediction or that was a blowout prediction?
  427. No, that was a, Iowa was supposed to be good and Michigan, there was always a question of whether there are any good.
  428. It's Iowa, not Iowa State, by the way. They weren't supposed to win, but it wasn't
  429. that surprising. Now they got beat pretty good all game long. And so the fact that they won at
  430. the end added quite a bit of flavor to it. But no, that wasn't a complete shock.
  431. This thing, but we're talking about this competitive. I'm gonna try one thing on y'all.
  432. So there's this thing that's starting to be said about the SEC that I've heard from
  433. a few different sources in the last two weeks, that the SEC is beginning to, the
  434. experience is more like the NFL experience now. And what they mean by that is teams
  435. are more competitive with each other. Like the NFL, almost literally any team could
  436. lose on any Sunday. That's just the way it is. And what they're saying is the SEC feels
  437. more like that with all of the transfers and NIL. It's just kind of, now there's still a
  438. bottom team. There's not as much of a top, but there's just more competitiveness in the middle.
  439. And I think there'd be an easy way to ask this statistically. I think we're fundamentally talking about the spread
  440. in quality of teams, the spread in team strength as a function of the standard deviation
  441. and game score. And the NFL, the college historically has had bigger standard deviation than the NFL, but much
  442. bigger spread of quality. And so you just don't see as much craziness happen, but the NFL, there's spread, the
  443. top team to the bottom team, but it's not that much. You know, it's like- Yeah, all right.
  444. What you're talking about is like within team variation versus between team variation, right?
  445. Well, that's not what I'm trying to, that might be another point. That's not what I'm saying.
  446. I'm saying- Oh, okay. I thought you'd kind of like, because you're talking about two different, like the college football
  447. has more kind of, I guess, within team variation that the same team, same two teams
  448. matched up in each other will like produce a lot different, higher variation of outcomes over,
  449. if you were to repeatedly kind of see that. That's probably true. Versus like two teams are just kind of
  450. different in terms of quality, that's the between team variation, right? Differences in team quality.
  451. Really, really all I'm playing with here is the between team variation. And the NFL is, if you put it
  452. in units, standard deviations, if you normed it for the spread in games, so you norm
  453. college and pro, and you just ask what the NFL's fraction of that is, it's gonna
  454. be a little bit less than a standard deviation, I think, from top to bottom. And the SEC used to have a good
  455. normal spread. The Big 10 has this massive spread. ACC has this big, massive spread.
  456. I think the SEC has gotten to be much tighter. So you just see anybody beats anybody.
  457. That's, you know, it's a long season. We're only four games in, but I think we're gonna see even more of that going
  458. forward. I wanna say one other thing about college football, because I'm gonna talk about it a
  459. lot over the rest of the fall. I've built this little shiny app. Adi, you'd be proud of me.
  460. Talk about vibe coding. I saw it, I saw it. Shane, avert your eyes, plug your ears.
  461. Did some vibe coding. Operationalized a concept I fantasized about for a few years.
  462. It's all right. If I was to judge you, I'd have to start judging everybody. I can't do that.
  463. You can look at it and say it's pretty. It's this frustration, because people always argue when
  464. it comes to the college football playoff about best team or most deserving. Essentially, the debates come down to best team
  465. versus most deserving, or resume versus eye test. It's always the same two dimensions.
  466. And we can just, we have objective measures of these dimensions. We can plot them, and we can start
  467. talking about, okay, what's the trade off? There's a, we can decide, okay, what's the
  468. weight here? What's the right, 50-50, 80-20, whatever you want. Let's just measure them objectively, choose the trade
  469. off, and see what it gives us. Now, I'm not saying that to end the debate. I think that should be the beginning of
  470. the debate. Let's set the table, essentially, with what the models say. Let's make it real explicit, transparent, use objective
  471. measures. But you can do that by a little two by two where your most deserving's on
  472. one axis and your best is on the other axis. And so I've started updating that through the
  473. year. We're gonna post it on the Wharton Money Ball Twitter account over the course of the
  474. year. But we've also built this app where you can go play with your own weights.
  475. You can choose which power ranking system you wanna use for best. You can use Bill's SP Plus, you can
  476. use ESPN's FPI, you can use Andrew Percival's metric consensus. And then you choose your own weights.
  477. You want 80-20 most deserving, or do you want 75-25 best team? You choose, and every time it adjusts and
  478. gives you the rankings in the top 12, the playoff field or whatever, according to that.
  479. And you can just, now we can debate it at the kind of philosophical level instead
  480. of the inane comments that always, the inane discussions that always. It's early in the season to be worrying
  481. about this, but I wanted to kind of get it underway and see what it looks like and get it going.
  482. So more to come, more to come on that front. Okay, this coming weekend, one big game of
  483. note, probably, you know, there's multiple, but Alabama -Mississippi State is probably the most interesting game
  484. coming up. The Mississippi State quarterback, out of the blue, one of the best players in the country.
  485. The Alabama quarterback, brand new guy, first-year starter, looks amazing. So this is a battle of two of
  486. the best young quarterbacks in the league. And one of the old, you know, always good teams and one of the hasn't been
  487. good in a while teams should be fun. Also, Ohio State, Iowa had Michigan last weekend.
  488. They traveled for that. Now they host Ohio State. So right on the heels of doing that
  489. to Michigan, they get one of the better teams in the country come in. All right, fellas, what about on the NFL
  490. side? I know you both pay a little more attention on the NFL side. Yeah, I mean, I, what happened to Drake
  491. May? What happened to my team? That's kind of what I'm thinking about on this Tuesday is I just have to note,
  492. you know, I mean, I've been watching every game, but like Drake May, I think he has seven turnovers in three games now.
  493. Going back to the start of the playoffs, it's something like 16 turnovers in seven, the
  494. last, his last seven games. Extraordinary number. It's an extraordinary number. And this is a person that a lot,
  495. you know, we debated last year whether he was kind of, you know, product of a good schedule and all this type of stuff.
  496. But I mean, he was playing at a near MVP level and now he almost looks unusable.
  497. And I just, how do you go between those two so quickly? It's fascinating. Like tragic, but fascinating.
  498. It is. Is quarterback play really that fragile? It's maybe early in a guy's career.
  499. I don't know. I mean, he does, I mean, I of course have no idea what's going on.
  500. I'm kind of hoping it's something that can be reversed at the same rapid pace or
  501. something like that. Maybe he just needs to, you know, chill out. I mean, you know, his coach says Drake
  502. just needs to relax. I find it, I hope that helps if he's able to do that. I would find it hard to be, I
  503. would find it hard to relax in his particular life circumstance. Doesn't, don't they play Buffalo this coming weekend?
  504. Yeah, yeah. Relaxing. It's gonna be a nice chill walk in the park. What else around the league has your attention,
  505. Shane? What else do you think is interesting? Well, I mean, I think, I mean, let's
  506. talk, we'll continue to talk about the Rams being kind of this team that had such
  507. preseason expectations and, you know, has kind of stumbled out of the gates now.
  508. I think they've got to be talked about as one of the sort of interesting sort of disappointments other than, you know, the Patriots
  509. on a personal level of this early season. I mean, I see the Chiefs looking incredible,
  510. which is always, there's not a lot making me happy out there but like, you know,
  511. just, I mean, I have eyes. I can see, you know, Patrick Mahomes seems to be doing his usual thing.
  512. And so kudos to him for coming back seamlessly from that, you know, ACL injury and
  513. kind of just hitting the ground running. Right. Yeah, that's kind of what I've sort of
  514. noticed over the last few weeks. The Rams are coming to our town, Philadelphia, to play this.
  515. That's one of the more interesting matchups, I think, of this upcoming week. Well, not least because the Eagles come off
  516. their Monday night game. That was just shockingly bad. Yup, I watched the game last night.
  517. That was my major commitment to football. It was a Jewish holiday of Sukkot this
  518. last couple of days. So not much football on Saturday or Sunday but I did watch the Eagles last night
  519. with a lot of Eagles fan and they were very unhappy. Yeah, it was a tough game, tough game.
  520. I mean, to get beat like that in the NFL is bad but to be surprised by the Bears, by the third strike quarterback.
  521. So obviously, you know, we've talked about this a lot. We've even had some fightin' words about it
  522. but there's a lot of just debate about Jalen Hurts, what quality of quarterback he is.
  523. Obviously, he didn't look great yesterday but one would argue that he is not a great
  524. quarterback. He is a solidly middle-of-the-road quarterback. I was surprised at some of the Eagles'
  525. fourth down decision-making. They were at midfield at fourth and one and they were all set up to do
  526. a tush push. At the last second, they decided too far for that and they punted and punted when
  527. they're down as far as they are, late in the game, what are you doing, punt it, midfield, right?
  528. And- Yeah, that was a ridiculous decision for any team, I think, nevermind the Eagles
  529. with their tendencies. But you know, it does say, I'm sure that if we, I'm sure if we talked
  530. to their analytics folks, they'd be like, of course we should have gone for it. There's probably a very, there's a process by
  531. which the players on the field, coaches on the decision-makers on the field, they have
  532. to make the decisions on their own. I mean, they get some help but there's no electronic system that pops in their ears
  533. and go, yeah, this is what you do, you gotta make a decision. And maybe because they lined up to do
  534. the tush push and they decided it was too far for that, that kind of left them off balance and they had a delay,
  535. I'm not sure. But it was a really weird play. And then later they went fourth and five
  536. and that's like the right move, midfield, when you're down by as far as you are.
  537. It'd be interesting to see the trends on fourth down, whether they've plateaued or are we
  538. continuing to progress? Are we seeing regression? Are people a little less aggressive after hitting
  539. new highs recently? But I haven't seen those data. All right, fellas, why don't we talk a
  540. little baseball? We have got the four wildcard games going today. This is the kickoff of the playoffs and
  541. the Phillies and Braves have been going at it during the show. I've lost track in the last 20 minutes.
  542. And since when we started this segment, there is 75% chance on Colshy to win. Where are they now?
  543. I don't know on Colshy, it's still a three-two as we're sort of watching, as we're taping this for the Phillies.
  544. Well, you know, the Phillies are notorious for not having great, great bullpens.
  545. Yeah, no, I feel like this is, the Braves have the Phillies right where they want
  546. them. Okay. Well put. Colshy, Colshy calls it 72%. So it was 74 when we started the
  547. second half, it's a 72 now. At least there's still a three-two. Something must be happening because runner's on or
  548. something. Yeah, two men on, two men on for the four. Two outs, bottom of the eight.
  549. Okay. So they're coming on. Okay, we've got all the games and I'm just astounded at the American League bracket because
  550. it's half the bracket is the East and the other half are the West and the Central and all the talent is in the
  551. East. And so it's just this absurd, this was the thing Neil Payne was on about last
  552. week, right? The playoff structure being so whacked. I mean, it's just so, it just looks
  553. so wrong. The Jays sitting there waiting for the winner of the Yankees-Boston. And then we have the Guardians waiting for
  554. the winner of whatever they, White Sox-Astros or something. So one of those three teams is gonna
  555. get escorted into the American League championship series. No, it's, I mean, we talked about this,
  556. I think it was even last week, that I think, I mean, if you're gonna kind of have these wild card entries, you really
  557. need to kind of do the seeding by regular season record or something like that. I think it's the fact that the kind
  558. of rays are locked into, they're kind of reward for having the best record in the
  559. AL ends up being, they're locked into probably playing the second best team in the AL,
  560. whoever comes out of that, you know, Red Sox-Yankees series. That's tough. And I mean, like, you know, of course,
  561. they didn't anticipate, you know, when they designed all this, I'm not sure they anticipated, you
  562. know, kind of such a high between division variance and kind of top teams, but it
  563. certainly has happened, so. I mean, it's, it feels like a small sample thing to me.
  564. These are four teams per division. I mean, it's gonna get some high variance stuff.
  565. The odds, the Kulshy market odds, as we start the playoffs, you know, we talk about
  566. this every now and then, and here we go. Two teams own 50% of the probability, according to betting markets, prediction markets.
  567. The Dodgers at 31% to win it all. Milwaukee at 17 and a half. Those are both, those are both National League
  568. teams. Those, that's the- That's amazing. That sucks out basically their entire side of
  569. the bracket. Yeah. Well, they also like- I mean, yeah, I mean, I guess they're, so, I mean,
  570. the odds are implicitly favoring them in the World Series, but that's crazy. That is absolutely crazy.
  571. No, yeah, I mean, you know, under my coin flip model, I mean, that, you wouldn't
  572. do something like that. It's also that, you know, that's Dodgers with a lot more probability than Milwaukee, even though
  573. they've got, they're going the road. Those teams play, they'll end up on the road going to Milwaukee.
  574. So, I thought- Those are the two buys that, I thought I would share a little baseball.
  575. You know, one of the things that I always, I'm going to the game tomorrow. So, my Red Sox fans, buddies are telling
  576. me, oh, you're going to watch the Yankees to get closed out tomorrow. That's your plan, huh?
  577. And I have to say- What a World War Red Sox fan can smack talk a Yankees fan. Yeah, you know, the thing is, is that
  578. - Well, where are we going? It's like the, I feel like the Yankees, they have a wonderful pitching staff, but not,
  579. I mean, yeah, they have a great pitching staff, but I'm not sure their offense can
  580. perform at all in the playoffs. And in fact, so a couple of things that, I've done some research with a former
  581. student who's now started his PhD program. And we've been working on something which we
  582. call exploitability. It's an interesting statistic. And basically, it judges how diverse you are
  583. as a offensive team against the various different types of pitchers that you might face.
  584. To essentially create handedness matching and also type of pitch. So there's fastball and there's off-speed pitch.
  585. And what we noticed is that the Yankees are an extremely strong team offensively, historically, against
  586. fastballs, but just belittlingly bad against all the other situations. And in the playoffs, you get the best
  587. teams who have the most diverse pitching staff that are able to exploit your weaknesses.
  588. And the Dodgers are exactly the opposite side. They don't have anything like that unbelievable punch
  589. against fastball, but are much more uniform against any situation they might face.
  590. So I saw this statistic, not our statistics, we haven't calculated it yet, for this year.
  591. The Yankees are by far the worst team when it comes to facing top pitching among
  592. the playoff teams. Interesting. And the Dodgers and the Cubs, who are terrific teams generally, are terrific teams against-
  593. Will the Yankees have to face that only in the World Series? Yeah. Luckily enough.
  594. Luckily enough. Although Tampa Bay, Tampa Bay is- They're exploitable against really good pitching, but they're the
  595. only team in the AL with good pitching. You're right. Tampa Bay is actually, oddly enough, very good
  596. against top pitching. Although it's hard to know whether this, you know, these things are subject to much smaller
  597. sample sizes and season-long statistics. And you wonder how reliable and a predictable
  598. such an observation is. But if you look at the exploitability statistic, which actually has something structural, the Yankee lineup
  599. is just beatable because you can switch your types of pitchers at the release stage to
  600. just make them not able to produce when they need to the most. And anyone who's watched the Yankees over the
  601. playoffs over the year in and year out, just watch them just not deliver in the situations where you need to.
  602. When you're looking at this, were you looking, was it kind of like just at the team, overall team level, or were- Overall
  603. team level. Because it's like, I kind of feel like exploitable to me, like you could answer that,
  604. kind of investigate that at a deeper level where you're kind of going through the lineup.
  605. Because I feel like the Blue Jays last year really impressed me in their playoff run
  606. because it seemed like they just had like a lot of different styles of players in
  607. the same lineup. And I feel like that would be inherently, like if you've got like a real mixture
  608. of like, but you know, high strikeout, high home run, but also, you know, the on
  609. -base kind of guys, that just says to me that that would be inherently less exploitable
  610. than like in the lineup that's, I guess, like lower variance or whatever, you know.
  611. Yeah, I mean, that's really what it comes down to is low variance, right? And essentially you've got, it's, and in the
  612. playoffs, I mean, that's the twist that I think needs to be investigated. These high, these great stuff pitchers, they have,
  613. I mean, what makes the great pitchers, in some cases it's two pitches or even one
  614. pitch, but it's usually because they have an arsenal. And a great bullpen has variety, you know,
  615. and that's what it comes down to. What do you think, Shane noted this thing in the rundown that the Yankees go into
  616. the playoffs with only 10 pitchers and 16 position players. I don't, I'm not familiar with the breakdown
  617. for playoffs. Is it always a little bit more position player strong? It's usually like, I think I saw somewhere
  618. where somebody broke it down, like usually, like the mode is like 12 and 14 or 13 and 50, like 10 and 16 is
  619. pretty extreme relative to what other teams have done, you know? But so, I mean, yeah, I just thought
  620. it was kind of notable. They got a lot of rest, typically in the playoffs. And you have a lot of, Yankees have
  621. terrific crew of starters. That's a lot of volume and maybe they don't need these, but that speaks to-
  622. They can set their starters to go deep, which is kind of unique for them, certainly.
  623. Yep, I mean, with Kohlbeck and Rodan and Freed and obviously impending soon to be awarded
  624. Cy Young winner Cam Flitler, gonna be great. But they got no hitting, so. Look, another less high profile American League team,
  625. but a good story is the Chicago White Sox coming back and making the playoffs. Shane notes that they were 1.2%
  626. at the start of the year to make the playoffs. I was just in this conversation with somebody
  627. over the weekend. How do we explain this? Is there a story for how the White Sox turned it around so much?
  628. And how unusual is that in baseball? It's not totally unusual. I mean, the White Sox, I mean, the
  629. Red Sox went first, worst to first, I think. I've done it a couple of times in the last decade.
  630. So like kind of going that worst to first sort of thing. And they didn't even go worst to first.
  631. But that's worst in the division. The Sox were like 100 game losers or something.
  632. Yes, no, no, obviously we don't have as much data for that kind of, you know. But I mean, obviously credit to like, you
  633. know, clearly they've been able to rebuild in a very kind of rapid way. All those kind of trading away that talent
  634. for those couple of years that led to those 100 loss teams. Like, you know, they were able to, I
  635. mean, I guess good personnel decisions to turn it around. I think they're also that division.
  636. Well, I mean, the AL in general, outside of the AL East was exploitable, to use honest terms.
  637. Yeah, it was very exploitable. The last thing I want to hit is more from Shane. And I feel like this is something we
  638. should do at the end of every year because it celebrates what teams do with the revenue they have, revenue being so disparate in
  639. Major League Baseball. So Cleveland Guardians, we've got friends there. We've admired how they run their shop for
  640. a long time. And here they are again, top of the list. They're the only team who paid less than
  641. a million dollars per win in their payroll costs. They're at like 873. The next is the Rays, again, famously good
  642. personnel, small budget at 1 million. Marlins are right there at one. The White Sox, we were just talking about
  643. just over one. The Nats, just over one. But the Guardians, top of the list, under
  644. a million, making it happen. They're in the playoffs again. I mean, we talk about the Brewers overachieving,
  645. the Guardians reliably overachieve well as well. We should note while we're talking there on
  646. the other end of the spectrum, just to kind of give people a norm for that, like under a million payroll costs for a
  647. win, the Mets are the worst in terms of how many, the high price of their wins, they pay over 4 million per win.
  648. So it's like a factor of four. Well, they didn't have very many wins. No, yeah, that's right, that's right.
  649. I mean, the Mets, of course, you know, it's, Mets did what they could to kind of make their, that ratio as large as
  650. possible by losing as much as possible and paying as much as possible. But look, the third, fourth, and fifth highest
  651. paid payroll per win, third, fourth, and fifth, all at 3.3 million, around 3.3 million per win, Dodgers, Phillies, Yankees, all in
  652. the playoffs. Yeah. And Dodgers, obviously, the favorite. 31% chance, Kourtney Kalsche.
  653. All right, guys, that has been another show. Before we go, what is the story in Atlanta?
  654. I think the Phillies won, didn't they? Oh, the Braves did a bomb off our relief pitching and won.
  655. What a shot. Yeah, the Braves won. Oh my God. Yeah, yeah, it's almost like we saw it
  656. coming, but whatever. Wow, all right. There's a way to open the baseball playoff
  657. season, gracious me. Okay. Congrats to the Braves fans out there. There you go. Shane Jensen for Shane Jensen, for Roddy Weiner,
  658. for Eric Bradlow, and absentia, this has been Cade Massey for Jacob Grodnick, our producer, Deep
  659. Patel, the boss lady. Appreciate you guys, the listeners. Come back and join us next time.
  660. Between now and then, enjoy your sports.

Brief Summary

Sam Ventura discusses hockey analytics, team strategies, and the upcoming NHL season.

Key points

  • Welcome to Wharton Moneyball. Join Cade Massey and co-hosts for a full hour of sports analytics.
  • Hockey Season Kicks Off. Hockey is beginning tonight, marking an early start to the season.
  • The Role of Analytics in Hockey. Sam Ventura discusses how analytics influence decision-making in hockey.
  • Rising Salary Cap Implications. Sam Ventura explains the expected rise in the salary cap and its impact on team strategies.
  • The Evolution of Draft Strategies. Teams are aligning more on draft lists due to better data and video use.
  • The Role of Public Data in Player Evaluation. Public data is increasingly influencing player evaluations across sports, including hockey.

Episode highlights

Tension

45/ 100Some tension

Related episodes