Why the Transfer Portal Makes College Football Harder to Predict

Knowledge at Wharton

Ask about Knowledge at Wharton

Episode Description

ABOUT THE EPISODE College football rosters can now change dramatically from one season to the next. That is making forecasts, player evaluation, and even award races more complicated.

Ron Yurko, Director of the Carnegie Mellon Sports Analytics Center, explains how analytics can better account for roster turnover in the transfer portal era. He also discusses the Heisman Hopefuls platform, the challenge of comparing players across positions, and why sports betting and prediction markets require more statistical literacy.

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 Money Ball, welcome to a full hour of sports analytics here on
  2. the Wharton Podcast Network. This is Cade Massey hosting this week with two of my three colleagues in for the
  3. next hour, Eric Bradlow, professor of marketing statistics, and Shane Jensen, professor of statistics, two of
  4. the three colleagues I'm usually on here with. Audie Weiner in the classroom at the moment.
  5. Sadly, we will not have Audie today. We're going to run the normal format, meaning
  6. we're going to go an hour. We're going to do guests in the first half hour. Given the guest and the number of topics
  7. we could talk about, I suspect it's going to go more than half an hour, and we'll round out the show once our guest
  8. leaves with some open lines, open topics. Our guest this week, Ron Yurko, longtime friend
  9. of the show. I feel like we've known him since he was a pup. He's still a little bit of a pup,
  10. but not as much of a pup as he was. Ron Yurko is a teaching professor at Carnegie
  11. Mellon. He helps run the Carnegie Mellon, and they're in the SDS department, statistics and decision sciences.
  12. He helps run the famous and fantastic Carnegie Mellon Sports Analytics Center, which includes lots of
  13. efforts, but notably their conference, which is in just a couple of weeks here. They run an October conference, which is about
  14. as good as it gets in the world of sports analytics. I'm always delighted to have Ron.
  15. He's done teaching for us. We've been over there. He's helped out in lots of ways.
  16. He got his Ph.D. there at Carnegie Mellon. Heck, he went to undergrad at Carnegie Mellon.
  17. I think he is as Pittsburgh as it gets, as can be seen with the terrible towel in the background.
  18. Ron, welcome to the show. Thanks for the introduction, Ted. And, you know, you have a little bit
  19. of a CMU run going on right now. You had Sam Ventura on, another Pittsburgher, you
  20. know, another CMU, you know, integral part of our sports analytics center. And so, I don't know, maybe we can
  21. get a third person here at some point. Now, give us this. Well, we got we have Karim Kassam coming
  22. through to talk to Adi's lab group in just a couple of weeks. And so it is quite the fall, the
  23. fall of Carnegie Mellon here. Well, Ron, we just want to know how much it's going to be a little differently
  24. than that. Ron, how much, how much? Let's let's talk. Let's talk brass tacks here.
  25. How much of the three billion dollars is the Sports Analytics Center getting? Come on, Ron.
  26. That's got to be at least. I mean, I know as much, I know as much as you just put it, but
  27. yeah, it's been exciting times. And yeah, they had a big event on campus, CMU Miami.
  28. The yeah, I have no idea what to expect. I'm just I'm just along for the ride. I feel like Carnegie Mellon's like MIT, the
  29. world has kind of come to you over time. I mean, the STEM tech world that we live in has migrated towards our great STEM
  30. tech universities, one of which is clearly Carnegie Mellon. So it's a good time to be in
  31. Pittsburgh where you are. Yeah, no, it's been it's been a blast, honestly. Like it's amazing.
  32. I just think of fifth year now as faculty, which is just crazy. I remember when I was PhD student out
  33. in the studio with you guys. And it's just kind of crazy to reflect on at times that it's like how much
  34. time has passed. And I just get I just get to be part of this whole big thing all
  35. at once. Well, one of the things that y'all have going on this year that is super
  36. cool is this Heisman collaboration that you have. So some folks who are listening will have
  37. seen it. Some folks won't. So tell us about the partnership that you guys have and what it involves with the
  38. Heisman. Yeah, so really, you know, big thanks to the leadership over at Heisman, Jeff Price, Tim
  39. Henning, Brendan Baumgartner, amongst others, where Heisman is sort of reimagining themselves as an organization.
  40. You know, it's probably the most prestigious award in American sports, I would argue.
  41. You know, this is the award given to the most outstanding college football player in whatever
  42. way you want to go about defining that. And, you know, they they're taking this approach
  43. of like, how do they modernize this? How do they put the Heisman in the conversation all year?
  44. And they reached out to fans and, you know, they were hearing perception that, you know,
  45. fans viewed this as effectively a QB or a running back award. And, you know, they wanted more data.
  46. They wanted more analytics. And fortunately, the folks at the Heisman came out when the NFL draft was in town
  47. in the spring and they connected with us and, you know, figuring out about ways in
  48. which they could create this new Heisman hopefuls platform where the idea is literally every position
  49. in every program at the start of the year, every single player, you know, regardless of
  50. what position the vision is, they are a Heisman hopeful. And so for someone like me, who's been
  51. working on problems in football analytics for a decent amount of time now, you know, this
  52. is this is this is the holy grail of football, right, is how do we actually evaluate objectively every single position right on the
  53. field? How do we go about doing this? And, you know, they want to tap into this for college.
  54. Your your goal then is to you'll know you succeed when an offensive guard wins the
  55. Heisman trophy. Is that kind of the mark? Yeah, I don't know if they're going to
  56. win, but like if we see, you know, a lineman there, you know, in the in the in the final, you know, whatnot on
  57. the final day, that could be pretty neat. Just to be clear, Ron, are you looking
  58. at cross position value or value of players within position or both? Yeah, great question.
  59. So I will say this, like we're just getting started on it this year of like what gets provided, because obviously when we think
  60. about like how do we actually get data on evaluating offensive linemen, right, or data evaluating
  61. on cornerbacks, even right, things are very limited, right? You know, we talk a lot about like
  62. having access to tracking data to get new types of metrics. And I would still argue, you know, what
  63. research that we've done with tracking data, I can't give you a good number to say
  64. how much more valuable this linebacker was compared to this corner compared to the safety like
  65. on any given play. All right. So I would what I'm going to say right now is we are taking advantage of
  66. their relationship with EA Sports and such that EA Sports ratings for every single player in
  67. college football and FBS are being provided so that fans when they go to the Heisman
  68. Copal's website, they could see, hey, what are the ratings for players within a position?
  69. And so every single lineman, you know, actually has some number giving some way similar to
  70. how folks use Madden ratings where, you know, there's objective information that's going into that, what
  71. data they have access to at EA Sports and also their team of scouts that they rely on for subjective evaluation.
  72. And it's dynamic, right? They get to update that over the course of the year. So that's surprising.
  73. That's surprising to me. I think of the Madden ratings as things that came out in the beginning of the
  74. year. And that's a big deal when you find out what the number is. And then that's kind of fixed.
  75. Maybe I was wrong about that. But PFF is an organization that updates their numbers continually.
  76. They'll rate every play and they give a season level running, running score. So you're saying that EA Sports numbers that
  77. you're working with are dynamic. They're updating on a weekly basis. Yeah. EA Sports this year for the first time
  78. in college football, they're updating it weekly. They're not just frozen at the beginning of
  79. the year. And they've been doing that with Madden ratings, actually, as well. And I think, you know, college, you know,
  80. is a very, you know, it's a very different space compared to professional football, right, of
  81. how do you actually go about this evaluation of players subject to the wide range of
  82. talent that you see and how talent gets matched up between each other. You know, it's much more difficult, right?
  83. It's much more difficult to infer how good is the linebacker at New Mexico versus, you
  84. know, within the NFL comparing linebackers between the different teams, right? Ron, have you been asked to provide just,
  85. I'll call it, measures of performance as opposed to some sort of weighted score that gives
  86. some overall, like you'll let the consumer of this information decide what they want to do
  87. with it. But in some sense, they're not asking you for, like, you can imagine using, treating this
  88. like a supervised learning problem. You take last, you take the previous highs and winters, you see what things are predictive
  89. of that. And so you're not trying to predict the winner. You're trying, you're an information provider of what
  90. you think are useful statistics. Or do I not have that right? Yeah, you're absolutely right.
  91. You know, what we're, what we're working on with them is figuring out, like, what are
  92. the right types of metrics to bring over that hopefully give fans and also give their
  93. voters, right? And those voters are past Heisman winners, members of this electorate, sports journalist reporters, what information
  94. that can help them gauge what's going on. A big part of this then, especially in
  95. this first year of what we're working on with them is this fan vote. And so folks didn't realize, I think that,
  96. you know, Heisman's had this fan vote that it actually counts as an official ballot in
  97. the final Heisman vote that's cast. Now it's one vote of many that they have, but one of the coolest things they're
  98. doing this year is it's going to get announced on a conference championship Saturday to sort
  99. of be the first ballot that's officially cast. And, you know, they want it to be
  100. like a signal of what the fans are saying. And part of the motivation is last year,
  101. Jacob Rodriguez, this linebacker at Texas Tech won the fan vote. And that didn't get reported until after the
  102. Heisman, when Mendoza officially won it. And that's when it came out. But there was a little bit of a
  103. campaign. I think his girlfriend or wife was tweeting about it all the time, trying to build
  104. up like fans to vote for him. But he had a fantastic college football season and maybe could make a case for being
  105. in the Heisman race. And so one of the things that we're doing this year that it created was, you
  106. know, they're very much interested in having fans vote the entire year. But if you think about that, you know,
  107. a vote after just one week of games, right, is based on a single week of information.
  108. And then, you know, fans are going to keep voting every day throughout the entire year.
  109. So how much weight should you give to a vote in week one versus a weight to a vote at the end of the
  110. year? And, you know, you can think of like a naive way of doing that, of going, oh, one week of games, it's worth one,
  111. two weeks of games worth two. But, you know, I'm a statistician. So I took some time and was reading,
  112. of course, classic work by Mark Glickman and also Shane. There's this cool paper thinking about, you know,
  113. how much they were looking, I think it was like for design of matchups and selecting,
  114. like how much information in a game. Yeah. And the pair comparison test, right, and pair
  115. comparison modeling. And so what I decided to do was, you know, in college football, there is very
  116. limited, I would say, like public available college football staffs to work with, right?
  117. I don't have statistics yet on every single position to say, like, how much information have
  118. I learned about every player in college football? But what I do have is I have game outcomes.
  119. And so I can model head to head outcomes of games. And this is going back to classic Mark
  120. Glickman, how Stern model for football outcomes. And I have strength parameters that I learned
  121. for each team. And I can allow that to evolve from week to week, as well as from season
  122. to season. And what I what I ended up doing was I was curious about, like, OK, how
  123. much information am I gaining? How much information am I observing from week to week?
  124. And so what I decided to set up is if I viewed the the ratings I effectively learned from this state space model at
  125. the end of the year as sort of the anchor, I could see how far away are we from that end of the year
  126. at week 12, at week 11, at week 10, and so on, all the way back to week one. And for statisticians, just do this with KL
  127. divergence. And I could get, OK, a notion of this distance. And I scale it then in such a
  128. way that, all right, the the ratings, the weeks that are closest to the end of the year have higher weight.
  129. And one of the things that I thought was actually kind of cool about this was I did this over the past five years
  130. of what I call the NIL era. Things behave differently in the NIL era compared to before.
  131. And what I actually observed, because I allowed this model to take information from the year
  132. before as the prior entering the following year, you would actually see weeks two and three
  133. that could be further away from the end of the year than week one. And so it was this interesting period that
  134. kind of matches, I think, with intuition. We think about in the college football season
  135. where the first few weeks of the college football season, they can be kind of noisy.
  136. All right. You can have roster shakeups going on. You have weird, you know, Penn State looked
  137. amazing playing Marshall and Buffalo and Temple. Right. And then, you know, conference games kick in.
  138. Right. We don't have too many of these great matchups at the very beginning of the year.
  139. And how much information are we really gaining in those first few weeks? And so what this leads to is this
  140. curve that at the end of the day, as a statistician, I was thrilled. It was like, all right, I feel like
  141. I did real statistics. And then I got a very intuitive looking line that basically says, you know, weeks one
  142. through four, it's like a slow increase in weight. And then once you get into week five,
  143. it's just this really steady increase until the end of the year. And that's and we end up getting this
  144. this chart that basically says, like, hey, at the end of the season, your vote after
  145. all the weeks that have been observed, every single game that's been observed, it's worth 18
  146. times the amount of a vote from week one. Right. Because you've observed all of these weeks and
  147. it's not just 13 times more. Right. It's relative to what we're actually observing over
  148. the progression of the year. So that was that's the biggest contribution that I have for this platform this year.
  149. And it's for for fans, it's, hey, this Heisman fan vote race isn't going to be over in this week or week six or
  150. whatnot. Like Trinidad Chambliss, I think, is currently leading. Jeremiah Smith's probably going to take over in
  151. the next week. But every week moving forward, their vote is going to give more and more weight.
  152. Yeah. And so it's all based on this idea of how much information are we actually observing
  153. that's translating across the season. That's awesome. And it's awesome that the Heisman guys were
  154. willing to give you say in that matter. And we're willing to go with such a sophisticated and, you know, opaque, probably to most
  155. people methodologically approach. But it's really, really cool. Let me ask you a different question.
  156. This is just kind of a stepping away from the Heisman thing in particular, the Heisman
  157. Hopefuls platform, though we'd encourage people, Heisman Hopefuls, easy to Google, easy to vote.
  158. They have a running leaderboard of the EA scores by position. So it's fun that way.
  159. Let me just ask Ron, the analyst, Ron, the football fan, Ron, the cutting edge football
  160. analytics guy to weigh in on Jeremiah Smith versus Chambliss. And there's no right answer to this question.
  161. But we often bemoan that the Heisman always goes to the QB. We've had a couple of examples recently where
  162. there are exceptions. But over the over the history, hold on, didn't did Hugh Green win it as a
  163. defensive guy for Pittsburgh or I'm making that up? Way back when, that's before my time, come
  164. on, Pittsburgh, famous Pittsburgh Panther, there have there. I mean, Charles Woodson, I mean, we've had
  165. defensive, we've had defensive backs when it. OK, but it's rare in recent. Travis Hunter won it, but he played Travis,
  166. but he was a two way player, really kind of exceptional in that respect. I'm asking for kind of an analytics.
  167. Perspective on how exceptional would a wide receiver need to be if we take we have
  168. to define what we're looking for, like like let's say most valuable player by advanced analytics,
  169. how good would a wide receiver have to be to top the best quarterback? You know, now I'm not saying this is
  170. what the actual criterion is, because there's an eye test, there's a specialness that Jeremiah Smith
  171. seems to be hitting right now. But if you were to approach it analytically, just give us a sense.
  172. What do you think? How exceptional would a wide receiver have to be to top the most valuable position on
  173. the field by any measure? Yeah, it's a great question. I mean, it gets at the heart of
  174. this problem of how do I actually get a sense for how good any single individual
  175. player is on the football field, right, while controlling for the fact of who is their
  176. quarterback, who are they facing, who could be going up against them, right? And I still don't think we have good
  177. numbers on that, right, to say, like, you know, is Jeremiah Smith the same value as
  178. a QB? And I don't think I've seen a good way of doing that in college football yet,
  179. partly because publicly, right, we don't have the information on who who's on the field for
  180. every play, right? I can't just straight up adjust for that. But that being said, right, when you have
  181. a wide receiver that is putting up, you know, record like numbers, and if I think
  182. we go back to the examples of wide receivers that were putting up these ridiculous numbers
  183. in the past, you know, they're getting themselves elevated into the Heisman race, right?
  184. DeMonte Smith, right, he won the Heisman, being someone that was putting up these incredible numbers
  185. at Alabama. And right now, if I had to like make my vote of who is the leader of this Heisman race, I would I would
  186. say Jeremiah Smith, just given the consistency to, right, if you know, that's my fan hat
  187. coming on of like, how do I think about someone that is like Heisman deserving, you
  188. know, it's not just single moments, it's like a level of consistency that goes through the
  189. entire thing. And so, you know, I think like this question of how much does, you know, someone
  190. need to elevate at a certain position to, you know, exceed what we just think of
  191. as, hey, the most important position in football is the quarterback, right? And there's no doubt about that.
  192. But where is someone within position so many like standard deviations away? And is that the signal for us that,
  193. hey, this is this is a generational wide receiver or what they're doing is good?
  194. I was going to say, with regard to the decision making, like, is it is it just about whether that like wide receiver stands
  195. out historically or relative to his peers? Or do you also need that? It's just not an exciting quarterback year or
  196. whatever, like like like like if there's a quarterback that stands out, I think they they
  197. can just kind of have the you know, that's that type of thing. I can so like, you know, do you
  198. kind of need this combination like for like a defensive player or a wide receiver or
  199. some skill position player that's not a quarterback to be elevated into that conversation?
  200. Do you also kind of need to not be kind of on the top end of the of the of the quarterback position?
  201. And the QBs, I feel like will also get hit a little bit more by their record than other positions, too.
  202. Yeah, right. The Trinidad Chambliss will take a little bit of a drop because it's, well, oh, Ole
  203. Miss lost, right? They lost the Florida. It's the flip side of being able to basically having your hands on the ball like
  204. every other player. Yeah, it's interesting. So, Ron, if they're going to have you
  205. work on the Heisman, why not have you work then on the Unitas Award and the Bolitnikoff Award?
  206. And, you know, there are awards, Butkus Award. There are awards for high profile awards for
  207. people at other positions. Like, if you're going to provide metrics, why don't we provide metrics to help support those
  208. other awards? And it's a great question. You know, Heisman is its own single organization
  209. for just this single trophy at the end of the day, right? And, you know, the winners of the Heisman
  210. get brought into the Heisman fraternity. And but yeah, you're absolutely right. You know, we could see it across other
  211. sports. You know, Major League Baseball kind of took an initiative of bringing advanced defensive metrics into
  212. your Gold Glove winners. And they were trying to take a step in that direction, too.
  213. So hopefully, you know, we see more efforts in that way and people make more of a contribution from that sense.
  214. It would be great. Let's change gears, but on the way out, I want to say just a touch of
  215. research reveals that Hugh Green was the runner up in 1980 to George Rogers. I remember George Rogers being a very not
  216. exciting Heisman trophy win, but he was running up to it with the highest ever. What's that?
  217. I was going to give you another pit player that got snubbed, Larry Fitzgerald. Yeah, that's a good example of a wide
  218. receiver. But what about Tony Dorsett? Did Tony Dorsett win the Heisman? He did.
  219. And that's actually I think it's the 50th anniversary of Tony Dorsett winning the Heisman.
  220. I was going to say 76. If you had me guess, I was going to guess 76 for Dorsett, another pit player.
  221. Hugh Green won the Lombardi, the Maxwell and the Walter camp. And he was the highest finishing at the
  222. time defensive player in 1980. Okay. You mentioned NIL era. Shane saw you give a talk about NIL.
  223. In Japan of all places. Yeah, in Japan. Y'all saw each other in Japan. Okay. I was going to say, tell us where
  224. it was. I go to Japan to get my NIL content as everybody should. That's where you naturally pick up for college
  225. football is Japan. But yeah, this is something I've been kind of hooked on is thinking about, you know,
  226. modeling college football in this very dynamic era that we're in. So team, team, team, like power rankings, essentially
  227. team ratings. Well, and also, yeah, yeah. Thinking about the team dynamics and then hopefully
  228. getting more into the player level, right. With, especially with the partnership with Heisman, but
  229. yeah. So like, you know, folks don't realize this, the transfer portal itself before 2021, when the
  230. floodgates opened with NIL and then also when you could just transfer without having to sit
  231. out a year, you know, the transfer portal is literally like the software is a database
  232. that was created to make it easier to communicate between, you know, prospective coaches and players
  233. about like, Hey, someone moving between schools. And so that led to a little bit
  234. higher number of transfers that can happen. I think it was like 2018 or whatnot when like that software was created, then COVID
  235. happens, right. And then we get some sweeping changes in terms of, all right, because of this, can
  236. players now get to switch between the schools without having to sit out a year? Right.
  237. And they initially, they only got to do that once. Right. And then at the same time as this,
  238. of course, the NIL floodgates takes place where endorsements payments. And, you know, more recently is like the
  239. formal payment from universities to players and universities managing their salary caps in this way across
  240. sports. But then also in 2024, we started to see they dropped, you know, you could transfer
  241. as many times as you want without having to sit out. It was actually, if you transferred a second
  242. time, you would have to sit out a year, but then in 2024, I don't know what caused this, but they dropped that completely.
  243. And so we're in this era now where I think we're around like 3000 or more transfers just within college football.
  244. Right. And just players within FBS moving around. And so one of the questions I had,
  245. and this honestly came from me teaching my sports analytics methods class, where I was teaching
  246. Mark Glickman's how stern state-based model for modeling team ratings. And when I was doing some, you know,
  247. literature review and whatnot, and he had these extensions about modeling the variances between seasons.
  248. So entering a year, like how much uncertainty do you have about a team rating entering
  249. the year? And he, he was doing things way back in like finance examples of potentially allowing that
  250. variance to blow up in some way, because of whatever could be volatile in the current
  251. environment. And from reading that and preparing for my class, I thought, huh, that's kind of funny
  252. because maybe that could work in college football right now, given dynamics that we have.
  253. And so what I ended up doing, and I'm still working on the writeup of this, hopefully it gets finished here soon, is I
  254. modeled then the variance describing the change from one year to the next for a given
  255. team as a function of their transfers, right? How much, how many transfers do they have
  256. entering into the season? Is it purely just the number or do you look at other features of that transfer?
  257. So right now I'm just doing literally the number of transfers and I explored like trying
  258. to do it like by positional groups as well, but there's some weird selection bias you
  259. observed there. Uh, but then the other thing I tried to account for was like, all right, I'm
  260. doing this over this playoff era. So let's also see in addition to just regular temporal dynamics, what happens in 2021 in
  261. this NIL era? And one of the things that really is interesting. And so intuitively more transfers blows up uncertainty
  262. about what you have about a team. I don't think that's necessarily surprising. Just to be clear, just for our listeners,
  263. um, I guess would the intuition be that if you do not allow the variance to be a function of the number of transfers,
  264. what you'll probably end up with is too low a variance between seasons. Therefore, you'll have so many unexplainable events early
  265. on in the season, which could potentially be explained by allowing the variance to be larger.
  266. Is that for our listeners? Is that the essentially the intuition? Yeah. And Indiana is a great example of this,
  267. right? So Indiana in 2024, when Kurt Zignetti takes over, right, it's a completely different team, completely
  268. different roster. And so if you were going into that year, uh, in these classic models where you
  269. just said, Hey, the between variance is not a function of everything. Everybody gets the same variance, right?
  270. You would have this basically a forecast. And what I did in this version of this work is like, given the information we
  271. have about transfers entering the year, let's forecast, uh, how, you know, the, the distribution for
  272. their number of wins that we would see across all these head to head matchups that
  273. they would have. And if you don't allow for that variance to potentially blow up in the end, it
  274. would have been heavily concentrated around like three to four wins that year, but it's a
  275. completely different team. Right. And, and one of the reasons why I did this at like the variance level is
  276. because I don't know what the public college football data, you know, who's, who's playing right.
  277. You know, if I had more information about, Oh, they got this particular transfer and he's
  278. playing X number of snaps at this position. Maybe that actually goes into like the mean
  279. rating for the team of like, Oh, this is the skill of this team directly, but I don't know that.
  280. So I decided let's just chuck it into the variance and see what happens. Right. Well, that's, you're underselling, you're underselling that strategy,
  281. given what you don't know. It's like, but you, what the situation is, you don't have information on the means of
  282. these players. The best thing you can do in that situation is to widen your forecast, meaning the
  283. extent of variance. It's like, you're literally recognizing the uncertainty in the data.
  284. Now, the other thing is, of course, as you said, if you do that at the team level, then of course, it'll say how
  285. much the mean ability estimate will update and how quickly it will update as a function
  286. of the observations in the season, because a tighter prior variance will obviously hold weight for
  287. longer as opposed to a wider variance. And so it not only has implications as you're saying, Ron, for the distribution over the
  288. season, but even let's call it the Bayesian learning that's going on within the season.
  289. Exactly, exactly. Ron, would you ever do this? It seems like you would do this for,
  290. you could, you could do this for coaches. Have you, have you bottled just coaching changes?
  291. I would think that variance would increase with a coaching change or even the degree of
  292. coaching. So it's, yeah, it's a great question. And you know, these, these things are definitely
  293. related to each other, right? When you have a coaching change now, it comes with transfers, right?
  294. And so, and like teasing it apart is difficult, I would say. Now you also do have coaches though, that
  295. their strategy is transfers, right? Dion Sanders is averaging a high number of transfers every year.
  296. Kurt Cignetti, he's averaging a high number of transfers still at Indiana. Now, one of the things though, that was
  297. really interesting of doing this is, you know, controlling for these transfers and still actually getting
  298. a sense of what is going on over this period of time. And is there something unique about in the
  299. NIL era of itself? And what, so I just basically threw it like an indicator variable and those, okay, are
  300. we in post 2021? And what you actually see from that is, you know, holding number of transfers fixed, the
  301. between year variance shrinks in the post 2021 era. And now this is within FBS and people
  302. sometimes forget that the FBS is pretty big, you know, it's over a hundred programs of
  303. the FBS. And so what that kind of is saying is, all right, if your transfers doesn't change,
  304. we have a pretty good idea of where you probably going to be at from year to year now, but if you flip the
  305. roster in some way, all right, now I have a lot of uncertainty about what could be taking place.
  306. And, you know, I, I kind of think of this as sort of in this era, is it a little bit of the rich
  307. getting richer, the poor getting poorer in college football, right? Where maybe there's more variance at the top
  308. of FBS, but, you know, are we really going to see the type of leaps that we might've saw 20 years ago of the,
  309. you know, a Boise state coming into the Fiesta bowl and making a big upset? You know, are we going to see that
  310. type of leap from, you know, a non power for conference team, you know, jumping in
  311. and making a run of the playoff? I'm skeptical of that right now. Given the lines with resources, right?
  312. That's right. I mean, the spread between, um, Power four and G six has never been higher and
  313. people think it's kind of done now that now we are, the positive is on the top of the market is much flatter than
  314. it was for a long time. It was very peaked with one, two, three teams. Now we've got this kind of flatter Maxima
  315. up there, but they just hoovered up all the talent from every stage, just to pipeline
  316. up, you know, and that means Bill, I think Bill Connolly, like he he's done some,
  317. you know, he's one of the best college football writers to follow and the work he's
  318. done with data. And I think he's talked about this of like seeing recruiting mattering less and less now
  319. from year to year of seeing just, and that these teams that could take these strategies,
  320. like Ohio state doesn't need to get many transfers, but they can be very selective in
  321. the transfers that they get, right. They could get the top. All right. So Ron, we're going to run out of
  322. time with you. We want to hear about your work on sports betting as well. Let me give just the slightest introduction.
  323. Ron created a new elective at Carnegie Mellon in the last year. I think he taught it in the last
  324. year for the first time on sports betting. And he did this because he wanted to educate these young people who are, you know,
  325. front and center in the sports betting world. And one of his innovations was to build
  326. and run a sports betting market in his course where they earn extra credit points based
  327. on their performance. And then he did some fascinating things that I'll leave for him to describe, but it's
  328. really the, the top line narrative that comes out of this is spectacular. But the most important thing is out there
  329. educating these folks about sports betting directly. Yeah, it's yeah. Thank you.
  330. It's a, I honestly think it's probably like the most important thing I'm doing because you
  331. know, sports betting is, it's pervasive. You can't watch a single sports sporting event
  332. of any kind without being bombarded by advertising. And fortunately this is part of, we have
  333. this grand challenge seminar program here in Dietrich College at Carnegie Mellon where it promotes faculty
  334. from different departments to work together, to bring, you know, create new classes to, to take,
  335. you know, big problems head on. And fortunately co-teaching this class with Linda
  336. Moya over in decision sciences. She's a cognitive neuroscience expert. And, you know, I'm bringing the stats data
  337. science perspective and she's teaching like the pathway to addiction while I'm, you know, teaching both.
  338. This is an intro level course. So I've, I've had students in that class last year where this was their first stats
  339. class. So I'm teaching them statistics and probability and building models through the avenue of sports betting
  340. while also teaching them about this industry. Right. And very early on getting them exposed to
  341. the fact that if you are a successful better, you'll get shut out. You'll get limited, right?
  342. You're not going to be able to, you know, continue to place bets. You're going to be, you're allowed to place
  343. bets. It'll be very limited amounts. And so I ran this, this sports book called the Tartan Bet.
  344. And no official affiliation. But yeah. Tell us, most of the audience doesn't know
  345. where Tartan comes from. Yeah. So that's our, the Carnegie Mellon mascot. And Tartans are, a Tartan is actually like
  346. a piece of cloth. It's not, it's not a weird mascot. Yeah. You know, it's CMU and our mascots, the
  347. Scotty dog. Yeah. But yeah, it's, it's one, it's one of the cutest, uh, mascots out there.
  348. Well done, Shane. All right. Unplanned. But yeah, we have, uh, in that. So I have these students where they start
  349. with a, what I call a thousand Tartan bet credits at the beginning of the year, that's equal to 1% extra credit.
  350. And they'll bet over the course of the semester. And I gave them assignments about this and,
  351. you know, Oh, I want you to place so many bets this week. And, you know, eventually we get to the
  352. point where they're, they're learning building models and okay. Now what I want you to do is
  353. you're going to build this NBA linear regression model of score difference based on number of
  354. things, and then I want you to make a decision. I want you to use your model and
  355. the associated uncertainty from that model to decide whether or not you place a bet.
  356. All right. And I got, I had students, their mind was blown. I'm like, wait a second.
  357. So I have to fit this model, but I'm not just like telling you what the coefficient is.
  358. No, I don't care. What I want you to do is using this model, make a decision and tell me
  359. how you made your decision compared to real bets that, and this was the most, I think I've ever worked for a class of
  360. every morning and evening. I'm checking real sports book lines, updating my sports book, putting these bets in and students
  361. are placing bets the entire time. I had students that they placed over 400 bets over the course of this semester.
  362. The, uh, using these fake credits that they had. And I would give them like class level
  363. updates every week. All right. Where are we at as a class? And they were, they were hovering around basically
  364. the break even percentage, uh, the, the typical lines follow at the beginning. And then after spring break, I introduced the
  365. parlay builder. And then that's where things really fell off because then, and this is the, one of
  366. the fun things is as a statistician and teaching statistics and teaching probability theory, betting is
  367. this really interesting kind of fun way to teach these concepts, right? Cause they have to directly think about uncertainty,
  368. right? I'm not teaching them hypothesis testing and you guys, no, I don't have to worry about
  369. that. I'm teaching instead for them to think about it. You're not just giving me a single number.
  370. You're going to be a probability of how you think this could happen. Right. And that's, and conveying that when you go
  371. to the sports book and knowing this whole American odd system, how they price things, don't
  372. even read any of that. Convert into probabilities. Think about it in terms of probabilities.
  373. Right. And them understanding how hard this is, right. Understanding how hard someone like Rufus Peabody has
  374. to work and the logistics he has to manage to handle all of this. The, uh, and we fortunately we had bill
  375. Benter who lives just a few minutes away from CMU, famous horse reading, horse racing, betting,
  376. uh, expert who's a billionaire in terms of, you know, the coverage of how much he's
  377. won. Uh, he came and spoke to the class and he also talked about, you know, I don't bet at traditional sports books.
  378. I've been a pair of mutual betting system where I'm not getting limited. It's, I get, uh, you know, the, the
  379. racetracks take just a cut of how much is bet in this pot. So I don't get thrown out.
  380. I can't, I can't do baseball bets. So he talked about the difference in what he asked to do.
  381. Now the monkey wrench in this whole thing is prediction markets. And that's something that.
  382. Well, this is obviously that, that, that, when you talk about getting limited, this is where
  383. we wanted to go. So you, you conceived of this course and started building this course.
  384. Probably about the time somebody can see the call sheet and the way around all the
  385. regulations. And the world has, I don't know what the volumes have been, but it's, it must
  386. be the case that prediction markets have taken much of the volume away from sports books.
  387. But my question to you is like, you know, setting aside implications for the direct logistical
  388. implications for your class, how do we feel about prediction markets to what extent does it
  389. lessen your mission because they're less parasitic, they're, they're, they're less, um, you know, they're more
  390. straightforward, people don't get limited. It's a, it's a cut, you know, in the same way that the para mutual market
  391. is that kind of thing. So I still think, you know, when we talk about what can lead to addiction and
  392. these downward spirals that especially young men are, we're, we're observing, right, we still have to
  393. be educating them about this and, you know, does it maybe look more fair on the surface potentially, right.
  394. But then. Teaching these students the difference between Ron DraftKings and, and the, in FanDuel, these guys are
  395. running these promotions that are targeting people at the right people at the right time to
  396. induce as much gambling as possible. And they're limiting the people who are going
  397. to win. It's like, it's a very different game. I'm not shilling. Oh yeah. I agree that they're actively selecting, right.
  398. Now prediction markets don't actively do that, but when you're making a trade on a prediction
  399. market, right. And an 18 year old college student here, they're, they have no idea that, you know,
  400. the market maker that has a lower fee or potentially no fee compared to them, you
  401. know, cause that's how the prediction markets are making money or fees on transactions.
  402. Right. And they're just taking a price that's available. They're just the, the naive market taker that's
  403. going up against trading firms. Right. They're going up against these organizations that are
  404. operating on a whole different scale from a watcher. Just looking at ads, Cal, she is no
  405. different to me than draft Kings and all these other ones. They're just another odds maker.
  406. So showing me ads, I mean, I know there are differences, but I mean, just because
  407. they advertise doesn't mean their behavior is the same. And there's like, I mean, if you, if
  408. you look at the behavior of the fan duels and the, and the sports books, they are so much more aggressive.
  409. On, on, on, on exploiting exactly the most addicted, the most vulnerable, but aggressively so programs.
  410. Yeah. And then, but now the caveat also to prediction markets is the scale at which we
  411. have these markets now. Right. And you know, we had the, these incidents last year, that's a great Superbowl example of
  412. did Cardi B perform at the halftime show. And one of those was both on call. She and Polly market and Cardi B was
  413. on this stage at the halftime show, but she didn't sing. And so one of the markets said, you
  414. know what, we'll, we'll kind of split the money both ways. Uh, we don't know what happened.
  415. And the other took a stance, right? She was there. How do you define this outcome?
  416. Right. And then you, of course you have the insider trading scandals. Are you, we have insider information.
  417. So I, I think there's still the value in both teaching this, communicating this, and, you
  418. know, and I have to figure out a way of like, yeah, conveying lessons for prediction
  419. markets more directly too, because when I, yeah, when I pitched the idea for this class,
  420. there were no sports event contracts that didn't exist. Now the caveat to that too, is this
  421. could all change, you know, depending on what nine people decide at some point. So, you know, I, I have no idea
  422. what's going to happen from like the legality of it and these years ahead. Okay.
  423. Why don't we leave it there? Give us one thing you're thinking about on the field right now.
  424. Like either from last weekend or this coming weekend, give us one thing. Top of mind, Ron Yurko, the analyst and
  425. or a football fan. College or Crow? Either. One thing. Where's your heart? Where's your head?
  426. I, I, one of the things I'm thinking about is because I talked about the between
  427. year transfer portal dynamics. I kind of want to extend that to within the year, right?
  428. Like, okay. And this cohesiveness of a roster and I'll use Penn state as an example.
  429. You have like Penn state and Iowa state kind of getting thrown together into this single
  430. roster. And in these years ahead, you know, what can we learn about? Like, how long does that process really take?
  431. Can I can, and like, how does it differ between if you have more offensive transfers
  432. versus defensive transfers and coaches from the different systems, you know, how do we model that?
  433. And how did that move together? That's something I've been thinking about at a high level that I don't have an answer
  434. to though. I thought I was going to get something about Aaron Rodgers. I don't want to talk about that.
  435. Yeah. Okay. All right, Ron, thanks, man, for being here. Always a pleasure. Best of luck to you on all the
  436. work you're doing and especially the Carnegie Mellon sports analytics conference is coming up in just
  437. a couple of weeks. Yeah. Thank you guys. Thanks. Thanks for having me. And yeah, hope people can register.
  438. Attend the conference is coming up here, October 23rd, 24th. Great lineup, variety of sports.
  439. It should be, should be a great time. So thank you. Thank you. Ron Yurko, long time in front of the
  440. show, professor, teaching professor at the Carnegie Mellon statistics and decision sciences group helps run the
  441. sports analytics group over there. Data science, data science. Statistics and data science.
  442. The SDS is confusing at Carnegie Mellon because there's a very famous, very famous group, also
  443. SDS, social and decision sciences. All right, guys, that's been the first half. Of this week's show.
  444. We have another half to go. Come back and join us after the break. Welcome back.
  445. Welcome back to Wharton Moneyball. Welcome to the second half of this week's show.
  446. Cade Massey hosting with Eric Bradlow and Shane Jensen. Adi Weiner is away in the classroom at
  447. this hour. Sadly for us. Good for his students. Good for the university. Just off the line with Ron Yurko, long
  448. time in front of the show collaborator in many ways, Ron and the Carnegie Mellon world,
  449. as he pointed out, it is a bit of a Carnegie Mellon run. We're on, which is a good thing.
  450. We talked a fair bit about their conference, or at least we mentioned it multiple times.
  451. They have a big annual conference. We have a sports business conference every fall.
  452. That's coming up this weekend. The Wharton Sports Business Summit this Friday. No, that's not this Friday.
  453. I'm talking about Wharton People Analytics is this Friday. The Wharton Business Summit is another week out.
  454. So coming up soon. Wharton Business Summit, lots of interesting speakers. Keep an eye on that.
  455. Join us in Philadelphia if you can. All right, guys, we've got a little time to talk about current events.
  456. Some of you, neither of you look very happy about current events. My observation about baseball, when it hits the
  457. postseason, man, it is going fast. Those wildcard series, bam, bam. I think three of the four were done
  458. in two games. It's like, don't mess around with the wildcard series. And now you blink and the Yankees are
  459. down to nothing. What do you think about the last week? It's been an active one.
  460. They blinked a lot while fielding last night. I feel like again, again. I don't know what was up with this.
  461. It's so interesting to me. And this is again, not really a very analytical statement, but I feel like when I
  462. first came up as a baseball fan in the early 2000s, late nineties, early 2000s, the
  463. Yankees were obviously very, very good at that time. But part of it was they just, you
  464. know, they were, they were locked in defensively. Like, you know, that team was like kind
  465. of famous for always making the big defensive play and all this stuff. And I just feel like this is such
  466. a weird ethos that the Yankees can seem to shake over the last few years. I don't know what it's all about, whether
  467. it is just mostly bad luck or whether there's something cultural. I don't know, Eric, if you've got any
  468. thoughts. Yeah, I mean, I don't remember ever seeing, in my view, there were at least five
  469. errors in that first sitting. They counted three, but there was a double play ball.
  470. It should have been a double play ball. There was a missed throw, which the guy advanced, which they didn't call an error.
  471. I mean, I don't ever remember seeing such a horrific defensive set of plays that actually,
  472. I don't want to say it costs them the game, but it, you know, they've been behind.
  473. I mean, you don't remember two years ago in the World Series when they had a very similar type of, like, like I do
  474. feel like it's, I mean, it's true. Last night, I do think said almost like a, you know, like was historic just for
  475. the number of miscues that just kind of kept coming time after time. I do kind of, I don't know.
  476. Yeah, but it also, it also shows you the value of pitching in the postseason because
  477. look, the Yankees look like world beaters. Okay. Well, the Red Sox starting pitching's not that
  478. great. And you know what? The Rays on the other hand, and by the way, it's not like Garrett called it
  479. anything wrong. What'd he give up? He gave up one run. How much did Slitler give up earned?
  480. Was it one? I don't know what he gave up. It was tiny. So, I mean, it's not like the Yankees
  481. starting pitching hasn't, has done anything particularly wrong, just that they're up against a team that
  482. makes rarely defensive mistakes, runs the bases well, uh, takes a lot of pitches.
  483. Even if Slitler had the Yankees were winning, he wouldn't go on more than five innings.
  484. He was pitching 20 to 25 pitches an inning and they were doing a good job of staying away from bad pitches.
  485. And so I think the Razor, you know, I think they're the better team and especially
  486. without Judge and basically no Stanton. I mean, he hasn't played yet. Um, maybe pinch hit once in the, uh,
  487. Do you think, um, like, I mean, this is kind of, obviously we're just talking counterfactuals
  488. here. Do you think somehow, like, uh, like this, this team with a healthy judge, is that
  489. enough to kind of, cause he does seem to have that super additive or kind of interactive effect where it's like, not just his
  490. great bat that you're adding in, but like somehow the people around him get better.
  491. I don't know. Yeah. I think, I think it's possible that if Aaron judge were fully healthy, Aaron judge, and,
  492. you know, I don't know, maybe Stanton were healthier or healthy. Uh, Stanton, you throw those two guys in
  493. the lineup. You got Ben Rice, you got Goldschmidt. Uh, all of a sudden now you've got,
  494. I forget who the other guy, the guy, uh, whose father was also a baby, the guy that second for the Yankees.
  495. I forget who his name, but, uh, also another really good player. Oh, Garcia. No, no, no, no, no.
  496. Um, I'll think of it in a second. But all of a sudden the top five, six players on the Yankees can actually play,
  497. but you take out judge, you take out Stanton and, you know, they have one or two guys that can hit well in the,
  498. uh, this is what Adi and I've been talking about for the last 10 years. They have a couple of guys in the
  499. lineup that can hit the other seven. Not particularly, no one scares you in that
  500. lineup. And as opposed to the, to the Rays who have a bunch of guys that are basically all contact hitters and they just, you
  501. know, make pictures pitch. I mean, Garrett Cole talked about, he was exhausted after five minutes.
  502. He was exhausted because he's like, they weren't, they weren't going after bad pitches.
  503. Uh, he was having to throw everything on the corners. He was just saying he was just physically
  504. exhausted. And I thought that was a pretty telling type of situation. It's not over for the Yankees.
  505. I mean, look, you've got, look, the part that's a little surprised me. If you told me at the beginning of
  506. the series, you know, I know the Yankees Cole, Schlittler, Freed, and Rodan. It's hard to imagine that team getting swept,
  507. but they may get swept. But on the other hand, Freed and Rodan could also win at home at Yankee stadium
  508. and it could go to game five and then Schlittler's pitching again, or Cole's pitching again,
  509. and then anything can happen. Yeah, no, I think it's, you know, it's, it's, it's definitely not over.
  510. I just, you know, I guess maybe I'm, I'm, I'm kind of, uh, channeling a little bit of what Cade started out to talking
  511. about that these, these early playoff series, they can, they, they can get, they can really
  512. slip away from you quick. They go, they go fast. So it's kind of certainly, uh, you know,
  513. uh, um, I guess it'll tell over the next couple of days, but it's, uh, I'll be willing to, I look, even the Yankees,
  514. I don't know if they're coming back or not, but I would also be happy if the Dodgers got eliminated.
  515. It looks like the Braves have a shot. I mean, there's no, that's been by far, I think the most competitive division series so
  516. far. I mean, it's the only one that's not, you know, already had elimination games.
  517. Um, yeah, no, that is a fascinating one. I, you know, I, I, and I can't help, but feel like whoever makes it out
  518. of the NL is going to be a pretty strong favorite. I think in the world series, all, you
  519. know, the top two, y'all, all three of the three or four, the three, all the remaining teams, basically in the NL would
  520. be playing better baseball. It looks stronger than at least three out of the four teams in the AL.
  521. How do y'all feel about a five game series at this point? It's surprising to me that it's a five
  522. game series. Hockey is like nothing but seven game series. Right. And here we are already, I mean, we're
  523. still, we're still with the short series, essentially, even though we're at the division series level.
  524. I mean, Kate, eventually they have to stop playing baseball. Right. I mean, you know, maybe we should get
  525. rid of 162 games and then we, then we could show, then we could lengthen the playoffs.
  526. Look, I mean, yeah, it goes three, then five, it's three, five, seven, seven. Given how long the season is already.
  527. I don't think I have a problem with five versus seven. Yeah. And I wouldn't, I would, I would, I
  528. would sack, I would go down to like 156 games, like regular season games. If we ended up and just have like
  529. a more, you know, if they're going to expand the number of teams in the playoffs,
  530. expand it to like best of best of seven kind of series, like they do. It just feels, it feels so, it feels
  531. so. Go back to one 54 Shane. So we can go back to 1930. Yeah. 1930. Why not? Yeah.
  532. Okay guys, quick question. You might know the answer. 93 times a team has been down two
  533. games, oh, and two in a division series or five game division series. How many of those teams came back to
  534. win the series? Well, let's just do some base rates here. So if they were equal, which they're not,
  535. so there's no way any sane person would guess more than one eighth of the time, which would be a half to the third.
  536. Right? That's equal on independent. But let's just say, let's say a base rate. So that's the upper upper bound and no
  537. one would ever guess that. Um, let's just say that team is a 60 40 dog. So that's point once that would be 6
  538. % of the time, 6.4% of the time. So I would say an expectation, the answer should be somewhere around six.
  539. Um, with a standard deviation of somewhere around pretty small. So I'd say my confidence interval somewhere between
  540. three and eight. This man just did a real time full analysis, right? Well, I'm doing a binomial calculation.
  541. NP square root NP one minus P. Shane's got a more parsimonious answer. Shane, I'm pulling the string on the back,
  542. on your back. I'm pulling, I've pulled your string. What's, what's your answer? Oh, uh, the one eight, one eighth, one
  543. eighth. 12.5, 12.5. Right? So that's one eighth is 12.5 and the answer is like 11.
  544. And so it's much closer to the coin flip than you might've thought, Eric. And it's very, it's a nice little affirmation.
  545. It's consistent with Shane's model of baseball playoff. Well, I forgot my most important theorem of
  546. all of sports is momentum. Cause then by, I shouldn't have done 0 .3 to the third.
  547. I should have done 0.3, 0.3, 0.9. Cause we know after winning the next two, that team has got momentum.
  548. Um, okay. Um, any thoughts on who's going to come out of the, I guess we're just flipping
  549. coins, so who knows, but the Brewers, they got that, Oh my God, that luck off the roof on the, on game one against
  550. the Braves hit. That is unbelievable. You have to feel so bad race. I'm sorry. Yeah.
  551. Padres. Right. Yeah. That's a home run and they're going to lose that game and all that.
  552. That's terrible. But look, I think we deserve a Dodgers Brewer series, right? I think I really want to see that
  553. series. And, um, but I wouldn't mind as long as it's more competitive than the last time
  554. we did this, cause that was like a snooze fest, right? I mean, I would love, I would love,
  555. I'm kind of pulling for the Brewers kind of to go, go all the way through at this point on the NL side, but
  556. they got to obviously give the next series, you know, they got to get to the next series and have it be more competitive.
  557. I think there's a good chance that they'll, they'll face the Braves instead of the Dodgers
  558. though. Certainly. It's not about momentum. Do we, don't we agree though, on the
  559. American league side that the white Sox have, you know, some, let's call it non-stationary
  560. period going on. You don't, you just think they're playing mediocre teams and I'm not paying enough attention to
  561. know. I'm just shocked that they're still here. And one game away from the ALCS.
  562. Are you kidding me? I'm going to, I'm going to move. People shouldn't listen. I still believe in the coin flip model,
  563. but that central team is going to get destroyed by whoever wins the AL East and the, and the ALCS.
  564. That is, that is my, that's my non -coin flip little admission right there. Okay.
  565. He's not impressed cause they still haven't beaten anybody. Got it. Unimpressed.
  566. Yeah, they still, you know. Guys, let's talk a little, let's talk a little NFL four games in, you start feeling
  567. like you might know what teams are about. Where are you on what's going on in the NFL?
  568. And our Phillies, our Phillies, our Eagles dropped a stunner to the Rams. I mean, Rams are a big team, but
  569. they were up big late in that game and somehow managed to lose it. Eric, how much were you suffering on Sunday?
  570. I was suffering. I mean, they were up 20 to 10 with nine minutes left first and goal at
  571. the two. And so they were 0 for 12 on third downs in the game. Oh, for 12 on third down.
  572. Yeah. And I don't agree. I don't agree with the pundits were saying they should have kicked the field goal.
  573. No, they shouldn't have kicked the field goal. They should have gotten it in from the
  574. two yard line for four chances. Is this team like, how is this team not automatic?
  575. Like that's their like, right. Name of fame is they're automatic from two yards.
  576. And by the way, they're in real trouble. I don't know if you guys, so while we've been on the air talking to Ron,
  577. Lane Johnson announced his retirement today. Yeah. And so probably the best offensive lineman they
  578. have. Matter of fact, their record is markedly worse. Like they're a sub 500 team without Lane
  579. Johnson playing. Why is he retiring mid season? I didn't see it. Just, I just saw, you know, Adam Schefter
  580. reports, Lane Johnson. He missed the game for personal reasons. I assume it has something to do with
  581. some personal reason, which I don't know. But you think about it. Um, Saquon Barkley's injured, right.
  582. And there his backup tank Bigsby is injured and Devante Smith is out. And like, literally I could start naming 10,
  583. 15 players on the team. The vibes are just absolutely terrible around. Yeah. And so they're, they're struggling.
  584. And if you remember the end of last season, that sucked too. And so, you know, you can look at
  585. overall record and say they won the super look. Ron talked about this. There's a lot of season to season variants.
  586. They're very far away from the team that won the Superbowl just two years ago. It does seem like kind of a, of
  587. a long lost dream. Now that 20, so they're, they're going to, they're playing Jacksonville this weekend in London.
  588. So neutral side game and the line on that game, the line of that game is six and a half Jacksonville.
  589. Jacksonville is favored by six. It's stunning to me. Um, yeah, I mean, they just absolutely destroyed
  590. the Patriots a couple of weeks ago. I mean, they've been kind of rolling through
  591. everybody. So. I mean, I guess Denver and them played a very close game, but they've been sneaky,
  592. very good. It'll be an interesting test. That's a Sunday morning game. Fun, fun.
  593. Sunday morning, wake up to Philadelphia and Jacksonville. 9.30 AM Eastern. I think the one team that's probably surprised
  594. me the most, I know I should not even say this. Wasn't Kansas city not that good last year,
  595. even when Mahomes was playing. And now all of a sudden you added, what's that Kenneth Walker, the, you know, all
  596. of a sudden you add him and now the offense looks like the old chiefs offense and, you know, they seem to, as before
  597. they seem to know how to win every close game and, you know, I just, I'm not saying they're going to win the Superbowl,
  598. but what I'm saying is it looks like the AFC is going to go through Arrowhead again.
  599. No, I think they're my pick for the, for the Superbowl. At least there's a particular point just because,
  600. you know, some, most of the other kind of, you know, and AFC teams look, look weaker, you know, like seem to have, uh,
  601. have got more weakness. The team that's kind of surprised me. Well, one second, one second before you, Shane,
  602. before you go, I just want to say we talked prediction markets with, um, Ron in the first half of the show.
  603. Quick and easy to go to prediction markets on Superbowl winner this year. Are you surprised that the Ravens are tied
  604. for fourth? Essentially? I agree. I agree. They've been one kind of a surprising and
  605. impressive team so far, but the markets have the Rams at the top at 11 bills, 10 Niners, 10, and then Ravens chiefs and
  606. Seahawks at eight with the Jags. We were just talking about at six behind that. What do you think about that?
  607. How does that not include the bears in there? Bears are five. Oh, did it was the, okay.
  608. I was going to say like, cause that's the team that's impressed me the most. I mean, I, I think, um, yeah, I
  609. mean, I, I, I don't, how fast do prediction markets kind of like the Rams kind of being the top pick seems to me.
  610. Increasingly questionable as time goes on. I realized they did just have a victory.
  611. They did win last week, but it was, you know, not, not, not a reassure, not the type of win that reassures the market.
  612. I feel like, um, and so like, you know, I, I kind of feel like how much are we dealing there with kind of
  613. preseason that preseason prediction versus like what's happening week to week. And I mean, I'm not saying we should
  614. be reacting too much week to week, but it's probably, you don't have faith. You don't have faith in the markets.
  615. What's your level of faith in the markets is one question. Like, can you do better?
  616. Can you do better than call? She, um, no, I don't, I'm not sure I can. I just, when, when, when the market, when
  617. you're telling me the market still, the Rams is the Superbowl favorite. Is that still like, kind of like, and
  618. is that the instantaneous market or is there like a lot of volume, you know, like
  619. this is, this is high, this is high volume. This is, it's a, it's a, it's a good market read.
  620. Um, I w I would say it's a good market read. I mean, the volume of this market is
  621. 165 million. Just hilarious. Um, by the way, what do we have on Monday night? Buffalo at Los Angeles Rams, the top two
  622. teams in the call sheet market. The line right now Rams by two and a half, which is essentially home field.
  623. I feel like that's the whole NFL. I feel like all the lines are between two. I'm not, yeah, the majority of lines are
  624. like two and a half to three and a half, which is approximately home field. I mean, it's basically saying, yeah, these guys
  625. are roughly equal. I mean, it's just, it's amazing how smashed up and then you see it play out
  626. teams win who you'd wouldn't think would win. Here's the way I would frame it. Maybe in the language that Ron used in
  627. our first, you know, 30, 40 minutes. I feel the chiefs are beatable, but they're
  628. extraordinarily consistent. I think what could happen is it wouldn't shock me if the Rams in the playoffs
  629. played a bad game and lost. Wouldn't shock me if the Ravens had a bad game and loss.
  630. Wouldn't shock me if the Bills had a bad game and loss. Wouldn't shock me if the Bears had a
  631. bad game and loss. It would shock me if the chiefs had a bad game and lost. It would not shock me if they lost.
  632. I mean, the last, the last time we saw them, the last playoff game we watched, they had a real bad game in the
  633. playoffs. So you're kind of, you're, you're, you're, you're kind of making a projection from this season,
  634. I think there, right? Yeah, making a projection from this season. I think their level of play appears to
  635. be quite consistently good. To have added one of the best running backs in the league, if not the best
  636. running back in the league, to that offense is a great way to get some consistency,
  637. right? I mean, that's a whole nother dimension that they've not had in the time we've watched
  638. Mahomes. Guys, um, other, another game that really has my eye next weekend is the Sunday night
  639. game. Baltimore's going to Atlanta, Atlanta. One of the surprises so far, Pennix is
  640. kind of taking hold of that position and Bijan, my boy, my Longhorn has been just
  641. so much fun, his first carry, first carry last night. Anyway, um, Baltimore, MVP leader, in my opinion
  642. right now, what's that? MVP leader. Yeah. Um, what did the amazingly, well, Lamar is
  643. expected to be out. And so that's the big thing. Cause the line on this game is Atlanta
  644. by three, who would have thought that the Baltimore Atlanta line would be three, but a
  645. lot of that is Lamar. It's a real shame. We don't get to see the Ravens. Is it still a Hundley?
  646. Is that the backup for the Ravens? I believe so. Yeah. So yeah. That's, I mean, that's one of those games
  647. where we're, you know, let me say the only good news for Baltimore, obviously they want
  648. Lamar in there. We're going to see how good the rest of the team is in the sense of
  649. if their defense is strong, running game's good, special teams is good, et cetera.
  650. They can win that game, but we're going to find out if the rest of the team can kind of, in some sense, compensate
  651. for the loss of a multi-time MVP. Cause let me just say, let me give an example.
  652. This shows you, I think right now, to me, the coach of the year, the coach of the year right now is Jim Harbaugh.
  653. Because what John, John, John, John, John, John Harbaugh, sorry, John. No, Jim's getting fired, Jim's getting fired.
  654. John Harbaugh, because Jackson Dart went out, they've lost all kinds of players and he's found
  655. a ways to win games. They're three in one. They're leading the division. I'm not saying they're going to win the
  656. Superbowl or anything like that, but he's found a way to win with the players he
  657. has playing right now and let's see. Who do they beat the last couple of weeks? Titans and Cardinals?
  658. I think what Chicago has been doing, I think is even more impressive. In the face of being their quarterback, I
  659. think what Chicago has been doing is kind of more impressive because they haven't just been
  660. kind of winning, but they've been getting, you know, elite looking performances that, well, maybe not
  661. as much last week, but whatever case Keenum did like a couple of weeks ago. I was incredible.
  662. That was the most incredible thing I ever, they made him look like a Pro Bowl quarterback and he looked.
  663. And, and I guess, you know, I think the part that makes me kind of put, put Chicago higher than New York, the Giants
  664. in my mind is that, you know, the, the prospect of Caleb Williams still coming back
  665. this season, I don't think that's still like a, a season ending injury in the same
  666. way that Jackson Dart was. So, you know, the Giants. Well, we know of course the Giants have
  667. my guy, Jameis. So they'll be fine. Well, it'll be exciting. It'll be exciting. I think, didn't he have another like.
  668. Oh, 400 yards, three touchdowns, three picks. That's classic Jameis. Classic Jameis.
  669. You're going to, he's going to score you, he's going to score 30 for you and 20 for the other team.
  670. The variance machine, that guy, the variance machine. Okay, guys, before we go, I have to
  671. call out a pretty good slate on the college football side, including one of the games
  672. of the year. So the night game, one of the two big night games is Georgia at Alabama.
  673. Alabama has been a bit of a revelation in recent weeks, had a big win this past week, all of a sudden we're finding
  674. them at the top. I think they're at the top. They are at the top of Andrew Percival's
  675. metrics consensus, which is 10 different power ranking regarded, highly regarded power ranking systems.
  676. They, they are, it's the first time in a couple of years, we've had an SEC team back at number one.
  677. So Alabama is hosting Georgia. They're one and a half point favorite. So just spectacular lineup at the top of
  678. the, at the top of the country, top of the SEC. Texas OU is this weekend. This is one of the great rivalry games.
  679. It's in Dallas, always on a neutral field. That neutral field is probably the reason more,
  680. as much, if not more than any other rivalry is one of these chuck the record books out.
  681. Texas is an eight and a half point favorite, but Texas has won so many of these games over the years.
  682. They didn't deserve to win. They were not the better team. I can't help it. Be scared about being an eight and a
  683. half point favorite. Is that like karmic, karmic retribution in store or something like that?
  684. Not a night game also? Night game also? No, that's a 2.30 game, which is better than, for years it's been a 2
  685. .30. 2.30 for you. I think 3.30 for us. 3.30 game. Usually it's noon, but they've bumped it into
  686. that prime spot in the middle. A couple of the games, keep your eyes on, sticking with good teams, UCLA is going
  687. to Oregon and UCLA has been a surprise this year too. So UCLA has been down for a while.
  688. They're playing entirely decent ball. They're rated number 21. They're going to Oregon, big dog, 11 and
  689. a half point dog, but it'll be a good proving ground for UCLA and Oregon's still
  690. trying to get traction after a rough start. We've also talked about, you know, the college
  691. football, isn't the standard deviation, something like, I forget, is it 14 or something?
  692. It's at least 14. So even if you're an 11 and a half point dog, it's not like you're a
  693. 10% chance to win the game. You're more than that. Oh, yeah, that's right. Substantially more than that, I would say, because
  694. that's well, well, less than one standard deviation. Let me give you one more. And it's a sad game, but it's like,
  695. okay, somebody's going to lead in bad shape. And it's USC at Penn State. Penn State this past weekend had one of
  696. the biggest losses of the season. They went to Northwestern. Northwestern kicked off their new stadium.
  697. They did this thing where they built on top of, they raised the other one. They're up there in Evanston on the banks
  698. of Lake Michigan. They built a smaller stadium. And this is a movement in college football.
  699. Like SMU did this, they built a stadium on campus, small, a few years ago, 35 ,000 people, they kicked it off against Penn
  700. State last weekend and beat the hell out of Penn State. And all of a sudden Penn State has
  701. two losses. They lost to Wisconsin the weekend before. So Penn State is suffering under new coach,
  702. Matt Campbell, and things are unhappy, but here comes USC who are definitely suffering under hot
  703. seat coach. And we're going to find out whether, which of these two teams can survive.
  704. It's the whiteout. It's a night game. Oh, that'd be a lot of fun. The whiteout at Happy Valley.
  705. Yeah. So that's some good stuff to keep your eye on, on Saturday as well. So great weekend.
  706. That was a great weekend of football, Saturday, Sunday, and Monday. Guys, that has been a full show here
  707. on Wharton Moneyball for the whole crew, for our guest, Ron Yurko, for Audie Weiner in
  708. absentia, Shane Jensen, Eric Bradlow, and Cade Massey. Thank you guys for listening.
  709. Many thanks to our producer, Jacob Grodnick, and the boss lady, Dee Patel. Wish you guys the best.
  710. Enjoy the weekend between now and next show. Enjoy your sports.

Brief Summary

Ron Yurko discusses sports analytics, the Heisman collaboration, and current sports events on Wharton Moneyball.

Key points

  • Heisman Collaboration. Carnegie Mellon partners with Heisman to modernize the award and include analytics.
  • Fan Vote Impact. The Heisman fan vote now officially counts, influencing the final decision.
  • The Importance of Metrics in Awards. Discussion on why metrics should be applied to awards beyond the Heisman Trophy.
  • NIL Era Changes College Football. Exploration of how the NIL era has transformed college football dynamics and player transfers.
  • Innovative Sports Betting Education. Ron created a new course on sports betting to educate students about the industry and its complexities.
  • Prediction Markets and Gambling. Discussion on how prediction markets differ from traditional sports betting and their implications.

Episode highlights

Tension

50/ 100Some tension

Related episodes