Why the NFL Has No Clear Super Bowl Favorite

Knowledge at Wharton

Ask about Knowledge at Wharton

Episode Description

ABOUT THE EPISODE Is this NFL season more wide open than it appears?

Brian Burke, Sports Data Scientist at ESPN, and Kevin Cole, Data Scientist at PFF, use football analytics to examine the season ahead. Burke explains how tracking data is changing the way teams measure pass protection and pressure, while Cole breaks down quarterback projections, team strength, and why the league's top contenders may be closer together than expected.

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 Mastey hosting with two of my longtime co-hosts, two of my three
  3. longtime co-hosts. Eric Bradlow is here dialing in from his office in West Philadelphia, as is Adi Weiner
  4. in his new-ish office. I am in from Austin, Texas, and Shane Jensen is out doing Shane Jensen things.
  5. We'll be back. Some four of us will be, some combination of us will be here. Always are.
  6. We are rolling into the NFL football season this week. We kicked off college football last weekend, first
  7. full weekend, and we did our annual preview show last week. So we're changing gears.
  8. Big, big change here. Now we're going to go NFL football, professional football, and do another preview show.
  9. To do that, we have two guests. We're going to go two guests this week. We're going to start with Brian Burke, who
  10. has probably been here, like Bill Connolly on the college football side. Brian Burke has probably done the NFL preview
  11. show with us every year since we started. This will be our 13th year in a row. Always glad to have Brian along.
  12. And in the second half of the show, we'll have Kevin Cole, friend of all of ours, friend of the show, one of the
  13. best football analysts out there, coming up at the top of the hour here on Tuesday
  14. afternoon. First, Brian Burke. Brian coming in from his home down in the greater D.C. area, if that's reasonably
  15. approximate. He's apparently sold sponsorship for his podcast hits this year. He's wearing lots of logos, no doubt, rolling
  16. in the endorsement money. Brian Burke from ESPN, thanks for joining us. Yeah, thanks for having me.
  17. It's always fun. Always fun to have you, man. So, the perennial question for Brian at the
  18. beginning of September is, what was the off -season project? Because you always take on an R&D
  19. project of some kind and you whittle away in the Brian Burke workshop and then you
  20. emerge in August and tell the world what new thing you learned about football. So, we want to hear that.
  21. But first, you got to tell us something about Burke Racing and the gear you're wearing
  22. and your new television celebrity, whatever that might be. So, give us a quick, Brian, update before
  23. we get into the R&D. Yeah. No, this weekend is a track weekend for me. It's my first competitive event.
  24. So, it's not like a NASCAR race or a road race. It's more like a time trial. It's called a time trial.
  25. So, it's more like qualifying. You just try to put down your best lap as possible within a given session.
  26. So, that's coming up for me this weekend. What type of car are we talking about?
  27. It is a BMW 330ci, but it's been completely upgraded and torn apart. It's pretty much a race car.
  28. And for those of you who don't know, Brian is a former Navy fighter pilot. He's been retired now for a number of
  29. years, but he does like speed and is accomplished at speed. Brian, what was this news about?
  30. We wish it was a time trial, of course. We've heard about your adventures for the last
  31. couple of years as you've waded into this business. But now you've mentioned also a television something
  32. something. Yeah. At the end of last season, we debuted a simulcast, like alternate telecast of Monday Night
  33. Football, including our playoff games that we had. It's called The Playbook. And I can't make any official announcements or
  34. promise anything, but it sounds like the show is going to go forward this season.
  35. It won't be an every week thing, but we're really looking forward to it. So, it's a very analytic, heavy, kind of
  36. smart football, a lot of X's and O's, a lot of real-time analytics on the screen.
  37. And then my role is kind of like the Mike Pereira, sort of. He comes in and he talks about the
  38. replay challenge or the flag, and he tries to explain the rules. And then he's gone.
  39. That's me. I do the same thing, but for decision analysis. So, four towns, obviously, two points and clock
  40. management stuff. Okay. Who are the other people that are carrying the telecast when you're not on?
  41. Well, Seth and I split them about 50 -50. Seth Walder, my colleague, who I assume everybody
  42. knows. And then we've got Field Gates is the host. He does a play-by-play. Dan Orlowski.
  43. It's kind of a rotating crew. Dominic Foxworth, Luke Keighley last year, but I don't think we can get him back this
  44. year, but he's on Netflix, I think, now. But yeah, it was a lot of fun. It was really cool.
  45. It was a stretch. It was a challenge. So, at this stage in my career, to just do something completely different and be live
  46. on air, my palms got sweaty, and it was very cool. Can I just talk about how you can
  47. do, let's call it live feed analytics? We understand building models and coming up with
  48. results and all of that, but what enables you as the game is going on to, you know, is it like some AI-based
  49. engine or is it? Yeah, it is. Okay. Yeah. It's a lot of neural network kind of real-time processing of the tracking data.
  50. And so, we do things like run pass percent or blitz percent, that sort of thing.
  51. Coverage identification, that sort of stuff in real time, you know, identify the formations and personnel,
  52. that sort of thing. It'll be interesting to hear those guys talk about the game.
  53. So, they're doing a somewhat traditional play-by -play and color. It's just going to be a more analytics
  54. -infused, X and O-infused version of it, instead of the more or less schlock that
  55. we get on the main channels. It reminds me, I don't know if you've watched Nate Duncan's basketball podcast, but Nate kind
  56. of does that for the NBA, and he started out just as a serious fan, but it's a much richer experience.
  57. So, it's a more intense experience. It's perfectly, you know, if you're not paying
  58. a lot of attention, it still sounds like play-by-play. It's like, this is just a game.
  59. It's like not, it's not Brian Burke and his spreadsheet, you know, but it's deeper and
  60. it's more explanatory and you come away smarter, I think. I love it. My favorite part of the telecast is the
  61. camera view. So, it's a totally unique camera view. So, we have some alternatives.
  62. You have the regular TV feed and then you have the coaches film, the All 22 video, but the All 22 video is way
  63. too zoomed out for television. So, what we do is we zoom in on the All 22. So, you still see all the players, but
  64. you don't see a ton of green space above and behind them. So, it's my favorite of all the telecasts,
  65. all the different spider cams and Amazon Prime Vision stuff and I like this view.
  66. Okay, you're not supposed to say too much, but one more question. If you had an over-under, if there
  67. was a market over-under on when we would see this next telecast, how long do we have to wait?
  68. It could be three weeks. It could be eight weeks. So, it's still TBD. Yeah. Okay, very good.
  69. All right, Brian, let's talk about your R &D project. Every year, you do something.
  70. It changes and maybe your practices have changed, but for a long time, you would take
  71. a big off-season project and you'd emerge from the lab in August and you'd reveal
  72. something about football and you'd share it through the world, with the world through ESPN.
  73. So, what was it this year, if there was such a thing? The thing I could probably talk most about
  74. is we revamped what we call win rates. So, we have these metrics we dreamed up about eight, nine years ago when the tracking
  75. data first hit and there's past block win rate for blockers and then there's past rush
  76. win rate for past rushers and it's a pretty simple formula. If you get your chips, your tracking chips
  77. on your shoulder pads, if you get them closer to the quarterback than the guy that's
  78. blocking you, then you've won that block. And so, that's essentially how it works.
  79. It's a weak link reliability system. So, it's modeled like you would model some sort of durable good that's supposed to last
  80. a long time. And so, time is an essential element to it. We plot, the metrics plot these full survival
  81. curves. So, every blocker has a survival curve. Every pass rusher has sort of a hazard
  82. curve, as you would say, and reliability. But those curves are too much information for
  83. ESPN consumers, right? We can't put that on TV. So, what we did was we drew a line on the chart at 2.5 seconds
  84. and that's a reasonable number, just shorter than the average time to throw is a little
  85. more than two and a half seconds between two and a half, three seconds. And it was very representative.
  86. If I just give you your percentage win rate at 2.5 seconds, then that's going to be representative of the entire curve.
  87. So, it boils down to basically batting average for linemen, which is really easy to explain.
  88. It's sort of self-explanatory, pass rush win rate. What does that mean? Yeah. And when you say, Brian, that you're looking
  89. at the weakest link, one could imagine like, for example, a multiplicative model has like zero
  90. times anything is zero. So, if I take one minus the win rate or the win rate, if I never
  91. win, and even though people around me win, you could imagine that leading to a very
  92. small success probability. Is that what you guys do with it or something like that?
  93. Well, there's an individual metric for each individual player. Yep. And then there's a team metric.
  94. That's what I was thinking about how to combine it across play. So, Brian, is it schedule adjusted or strength
  95. of opponent adjusted? No, there's no opponent adjustment. Before you have the adjusted version of a
  96. metric, you need to have a good metric. Yeah. And over the long run, it tends to
  97. even out. What we do do is we look at how much help each blocker is getting. So, double team percentage, how often pass rushers
  98. are getting double teamed, and then also chip help. So, chip blocks from tight ends and running
  99. backs. So, we detect those, we report those along with the win rates to just give you
  100. an idea of how, at least directionally, how you could adjust those numbers. Brian, this is, for those who don't recall,
  101. this was one of those winter projects we were talking about. This was an R&D project when tracking
  102. first came out. This might have been the first tracking metric that y'all came up with, if I
  103. recall correctly. I'll tell you a story about, it was very serendipitous. I kept getting out, we got the tracking
  104. data, and only a couple of us knew how to use it or access it. And so, throughout ASPN, they would ask me
  105. questions like, hey, can you go on the tracking data and tell me who the nearest
  106. defender was on this pass play? Because it might not be on the video. And so, I'm like, okay, I can do
  107. that. And I'm a very visual person. I validate things in charts and graphs. So, I would animate the play, I'd animate
  108. the dots, and draw line segments from the closest player. That way I could tell, oh, yep, yep,
  109. yep, it's doing it right, it's working. And this is the closest player to him, this is the closest player to him.
  110. And so, I just automated the entire thing. So, I didn't have to answer anybody's question
  111. anymore. I just automatically told you who the closest defender was. But as I was watching this animation, I
  112. didn't just do it for receivers, I did it for all players. And I was like, oh, these are the
  113. blocks. I'm drawing a line for all these blocks. Oh, this is a double team because there's
  114. two lines going to this one guy. And then I had always thought, I had this theory long before we had the data,
  115. maybe back in 2010, where I was like, you have to model pass protection as a survival system.
  116. And it only takes one blown block or lost block for SAC to happen. What I've also learned is it's not really
  117. about SAC. SACs are great and they're visceral, you see them. But when you get a pass rush win,
  118. you affect the play. So, we know that this is a good metric for a lot of reasons.
  119. But one of the reasons is when there is a pass rush win, and this year we added bull rush win, which is kind
  120. of the big improvement for this year. And so, when there's either one of those wins, the completion percentage plummets, the EPA per
  121. play drops, the yards per attempt drops. So, Brian, let me ask you. If I told you, whether it's at the
  122. individual level or team level, these metrics early on in the game, have you guys shown
  123. that that's predictive of outcomes? Because I've always made the claim here on Moneyball for the last 13 years.
  124. Cage heard me say this. He doubts it every time. Just all I need to do is watch the first series on each side of the
  125. ball. I'll tell you who's winning the line of scrimmage, and then I'll give you my prediction
  126. of who's going to win the game. Have you guys found that predictive at all? And is that why it's potentially valuable?
  127. It's definitely predictive because it's reliable from year over year. So, the individual metrics correlate about 0.5
  128. from one year to the next, which doesn't sound like a super strong correlation, but that's
  129. actually extremely high for a sports statistic. Batting average is actually less than that, and
  130. then MLB for context. It also correlates really well with EPA per play. So, we can estimate that line play in
  131. the passing game is responsible for about 30%, up to 34% of the outcome is determined by that matchup between the pass protection.
  132. Hold on. That's a big claim to make. I haven't heard. I don't think I've quite heard that before.
  133. So, can you say that again? Yeah. At the team level, our pass block win rate correlates with EPA per play at 0
  134. .34. Okay. So, accounting for the solo variance that plus all the covariance that it's interacting with, we
  135. can explain about 30 to 35% of what happens on a pass play based on... Which is obviously fundamental to...
  136. I know these are correlational, and therefore, we don't know exactly how to attribute blame or
  137. causality here, but tell me a little bit more about this because it's a team that's
  138. winning the pass block rate doing so because the quarterback has opened wide receivers and gets
  139. the ball off quickly, and therefore, wins because obviously... Or is it adjusted for all that?
  140. It takes all that into account. So, it's designed at its core to isolate the individual player performance.
  141. So, it's independent of the quarterback. In fact, we don't even use where the quarterback is.
  142. We use the quarterback spot, like where he's supposed to be, and that actually works better.
  143. So, we look at the dropback and estimate exactly where his feet should be. So, if you could have the exact same
  144. pass rush win, and the quarterback slithers away and throws a touchdown pass, or you could
  145. have the very same pass rush win, and he goes on to get a sack because the quarterback's just not that alert or not
  146. that elusive. So, it isolates that effect. So, it isolates the quarterback and the receivers,
  147. the defenders, and because it is a survival system. Maybe you guys can remind me.
  148. There's a two-name word for it, a KT. Kaplan-Meier. Kaplan-Meier. Kaplan-Meier, right.
  149. My one, Eric. Sorry. Exactly, yeah. So, it's a Kaplan-Meier function. So, it accounts for time to throw and
  150. that sort of thing. Yeah. So, you're basically estimating. You're adjusting for all that information.
  151. So, if your particular quarterback likes to hold the ball for a long time, your offensive
  152. line's pass block win rate's going to be, or at least unadjusted, would be much worse
  153. than a team whose quarterback gets it off quickly. No, no, because we're looking specifically at 2
  154. .5 seconds. Oh, you're looking at specifically at 2.5, only at 2.5? Yeah. I mean, we look throughout the play and
  155. we generate the curves, but given that a quarterback, we can actually see how well would
  156. these players do if their quarterback did hold the ball longer, or how many sacks would
  157. these guys likely give up if the quarterback got the ball out on time. So, it's independent of the throw.
  158. Right. Brian, we're going to have to rush through a few final questions and get you along.
  159. We got a little bit of a short time with you this time. We have another guest coming on, but I
  160. did want to note that you mentioned that you added bull rushes to your analysis, and
  161. this is something you write about. Whenever you debuted this thing, I talked about
  162. adding bull rushes because a bull rush is just smashing the offensive lineman back into the
  163. quarterback space. It's not getting around him with these shoulder pad chips, and you know that it matters,
  164. and so you came up with another way of measuring that, and you say you ran all your analytics and say this proves to
  165. be really influential. It does. It means you can answer a question that people talk about a lot, and that is
  166. whether pressure up the middle is actually the most disturbing to a quarterback.
  167. It has a huge effect. I was really surprised. It doesn't happen that often. Maybe 5% of plays have a bull
  168. rush type win, but when it does happen, it has a major effect on the outcome of the pass play.
  169. If it affects the quarterback's ability to step up, to see the field, read the patterns,
  170. it's going to have an effect. Brian, give us a player that pops out from this analysis.
  171. Did you walk away from this analysis with a new appreciation for somebody on either side
  172. of the ball? No. I mean, it correlates with the previous version. I think it's better now, but there aren't
  173. any big surprises. I would say the biggest surprises or the most interesting things to me were how often
  174. certain pass rushers got chipped, like T.J. Watt gets chipped all the time, and things
  175. like the Rams run these stunts. One of the things it does now is detects twists.
  176. Any kind of defensive line stunt, we detect it and label it and say, oh, he's the penetrator, he's the looper, he's on this
  177. play. Teams like the Rams, they've that more often, twice as often as the team that runs
  178. it the least. Can I just ask one quick, I know we have to run a second quick follow
  179. -up question, Brian, is when you say label it, so how much human coding is done of this or are we at the point
  180. now where all of this can be done just using whether it's chip data or video data, et cetera?
  181. How much of this is scalable or do you have a call, just a whole bunch of interns just labeling and coding stuff?
  182. No, we don't. It's not a traditional supervised learning model that gets trained.
  183. It's a rule based metric. It's physics and engineering at its heart so that we don't have somebody labeling this is
  184. a win or this is a loss and so on. It's strictly rule based. I see. Okay. Brian, let's talk to the fan for a
  185. moment. Let's set the engineer aside and talk to the fan just as we end this interview.
  186. NFL kicks off tomorrow night. Give us something you're interested in and give us something to be following this weekend, this
  187. season, what do you care about? One thing I thought about this year with the Rams and getting ahold of Myles Garrett
  188. and all the press that they're getting, all the love, the really interesting thing to me
  189. is the markets, our model as well, FPI, we don't have the Rams as dominant favorites.
  190. We have them 25% chance to make the Super Bowl, which is very high. It's the best.
  191. It's the highest of any team, but it's still fairly flat from top to bottom this
  192. year. There's no super dominant teams. The interesting thing too is the market's barely
  193. moved with Myles Garrett. Basically, we could determine the effect on Super Bowl chances and it may have increased their
  194. chance to win the Super Bowl from 15 % to 16%, something like that. That was very surprising to me.
  195. I always understood that position players, even elite ones, just don't move the needle that much.
  196. It really is a quarterback league and then next most important is probably the coordinators and
  197. then you start getting into the position players. Well, it's even interesting for me to hear
  198. you talk about, you think the odds and the rankings are relatively flat because to me,
  199. the Rams jump out. It's not the same increment from the Rams down to the Seahawks or whoever you have
  200. number two, as it is from number two to number three. What you're saying is it's not quite as
  201. dominant as the media would lead. It's not like you would think. Yeah, it's not like you would hear.
  202. I've probably said this before on this show, but the Rams might be the best team.
  203. There might be 100% chance that they're the best team. Let's just say that they're so good that
  204. by the playoffs, they have a 70% chance to beat any other playoff contending team
  205. and they get the buy. All right, so 70%, that's really dominant. They probably won't even make the Super Bowl,
  206. right? Odds are against. Much less win it. By the way, Brian, given the division they're
  207. in, if you told me at the end of the season that they ended up third in their division, it would surprise me, but
  208. not that much. You're telling me Seattle and the 49ers could end up ahead of them.
  209. It's not that improbable. This time of year, we always forget how unpredictable.
  210. Our minds just get erased. There's this amnesia we have about how unpredictable the seasons are.
  211. I do remember last year on this preview show, I said, the Chiefs, this might be the year they don't make the playoffs.
  212. That's my prediction. Brian says Rams aren't going to make the playoffs. He says, media's wrong.
  213. That's just taking a stand here. That's not what he's saying. He's saying something very important and it's a
  214. great note to end on, and that is, remember how much uncertainty there is. He made it very concrete by saying, even
  215. if the Rams are guaranteed best team in the league at the point of the playoffs,
  216. they're still unlikely to make the Super Bowl. That's in it there. That's Brian Burke from ESPN.
  217. Thank you for the time, Brian. Always enjoy talking to you and always enjoy this annual kickoff tradition with you.
  218. Yeah, thank you. Absolutely. That's been the first half of Wharton Money Vault. Come back and join us after the break.
  219. Welcome, everyone. Welcome to the second half here of Wharton Money Vault on the Wharton Podcast Network.
  220. Cedric Bradlow, along with my co-hosts, long -time, almost 13-year collaborators, Cade Massey and
  221. Addie Weiner, some combination of the three of us, and Shane Jensen are here every week
  222. on Wharton Money Vault. Kevin, it's great to have you back to Wharton Money Vault, beginning of the NFL season.
  223. This is our, I guess Cade had said earlier in the show, this is our 13th year of doing a preview show and it's
  224. great to have you again as being part of it. People know Kevin from his sub-stack Unexpected
  225. Points. He was at PFF, which is one of the companies that was at the forefront of,
  226. I'll call it, unique and interesting data collection, which allowed for a lot of the advanced
  227. analytics that we have in football. So first, Kevin, let's start with what are you working on now and what's exciting to
  228. you about the upcoming NFL season? Yeah, yeah. Well, thanks. Thanks for having me again.
  229. It's an honor to be on here as a listener for, I don't know about all 13 years, but for quite a ways back,
  230. I've been listening. So it's always an honor to be on here, along with some of the luminaries that
  231. you have in the sports analytics world. Yeah, so I'm working on so far this season, just some of what you may call
  232. the basics, setting the stage for what we're seeing, what's going to happen, we're predicting is
  233. going to happen this season, whether it comes to how strong the different teams are, how
  234. much they've improved or haven't improved based upon offseason acquisitions, how much the draft has affected
  235. what teams are doing, and then looking specifically at quarterback play being, you know, mostly the
  236. driver of the offense and the offense being mostly the driver of team performance or something
  237. we can count on for a particular individual player more than any other doing predictions there.
  238. And then of course, a week or two into the season, scrapping all that and giving
  239. our next best prediction, which is really, which is another highly educated guess, but something that's
  240. highly variable at the same time. So Kevin, let me ask you, is the foundation of the system you work on, and
  241. please correct me if I'm wrong, is it individual player based? In other words, there are, let's call it
  242. qualities of players, and some are on the offensive side, some are on the defensive side.
  243. And then is it as simple as I'm sure it's not, but like the whole is equal to the sum of its parts.
  244. Can you give our listeners a sense of how you go from individual level assessment to
  245. let's call it team strengths, and then eventually win probabilities? Yeah, yeah.
  246. No, I mean, the foundation is really the team level performance, but then I do make
  247. adjustments based upon individual player additions, subtractions, injuries, things of that nature.
  248. So I have a kind of, you know, after the fact post hocs, we're looking at player value, and then projecting these player values
  249. at different positions, based upon how they performed over the last couple of years.
  250. And then so I have this, where really, the foundation of it is like an on off splits, which we know are very, you
  251. know, very unreliable as far as looking at individual players. But if you aggregate enough of them, and
  252. then you smooth them out in some sort of way, you can get a little bit of a value looking at individual players there.
  253. So I have these projections, I have the ability to then use past performance to project
  254. forward for players. And I essentially look at the rosters. And I say, okay, if we had the
  255. end of last season, which would be my most recent team level projections that I had
  256. last year, based upon the rosters, how would I predict that roster would perform this season,
  257. but now I'm going to make these changes to these individual players. And then even beyond that, if there are
  258. unit level differences, or missing players, or within if you have more than two different changes
  259. within a particular unit, that can also be a bit of a ding up or down based upon that.
  260. So it's really adjusting off of the team level predictions rather than aggregating up from the
  261. player levels. I see. Let's talk because good thing. Thank you, Kate, for putting it into the
  262. chat as well. Apparently, you do a lot with quarterbacks, and we would all expect that's where the biggest
  263. movement comes from up and down from team predictions. Can you talk a little bit about your
  264. quarterback model, and like how you build quarterback, how you build a quarterback model?
  265. And then I wouldn't mind if you tell me who you think the top quarterbacks in the NFL are going to be.
  266. Possibly, do you see any degradation of Patrick Mahomes? Where do you see him in the mix?
  267. But how do you see our hometown hero, Jalen Hurts? A, what is the quarterback model?
  268. How do you build it? And where do you kind of see the ranking of quarterbacks right now?
  269. Sure, sure. So I do something where I've found it's a Bayesian model where it uses what they
  270. call the normal normal distribution. So it assumes, which is roughly correct, that
  271. there is a normal distribution of performance when we look at a quarterback level for their
  272. efficiency. So I'm looking at their expected points added per play. So that includes their dropbacks and the other
  273. plays that they're primarily involved in. So that would be designed runs where the
  274. quarterback is the ball carrier. So looking at those efficiency numbers... Okay, so just to be clear, across quarterbacks,
  275. if I did a, you know, just for our listeners out there, especially for the students
  276. out there, if I did a histogram of expected points added, I guess, per efficiency per
  277. play, you're saying across quarterbacks, that would roughly, or the assumption is that that's normally distributed.
  278. Correct, correct. So I'm looking at what I call franchise quarterbacks or those who have had at least
  279. a thousand career dropbacks. I mean, I would like it to be more, but honestly, it's very skewed in the
  280. directions of those that continue to play on versus those who, you know, wash out after
  281. a season or two. So it's roughly those career numbers are roughly a normal distribution.
  282. And then if you look at a play level, it's not quite a normal distribution. You have to do some adjustments to the
  283. distribution there because you have so many plays where like an incomplete pass, for instance, is
  284. going to be the most common EPA figure you're going to get is going to be negative in that incompletion.
  285. And then when you do have a positive play, there's a much higher range of different
  286. expected points you can have added there. So you have those two, and there is essentially a model you can build off of
  287. that. What's the second normal? So I understand one normal is, let's call it the heterogeneity distribution across quarterbacks.
  288. What's the other normal part of the normal normal? So on the play level, so the expected
  289. points added on a play level. So you're able to use both of these. And then also, which I like about this
  290. model is built into it. What's built into it is that as you're getting more evidence, you have more confidence.
  291. So essentially you have a, you know, you have your assumption going into it for the
  292. distribution of kind of all quarterbacks based upon draft position that I have. You have your evidence and then you have
  293. your updated distribution, your posterior, which is now your new prediction. So while a lot of the stuff that
  294. I'm putting out there, when you have quarterback rankings and things like that, we're looking at
  295. the mean or the best guess is what you can call it for what their efficiency will be.
  296. The distributions are also different. And I go across to, you know, I make a point of noting that for instance,
  297. if we're looking at the Atlanta Falcons, if you're going to have, you know, if Tua
  298. is going to be their starting quarterback, or if we're going to have Penix be their
  299. starting quarterback, well, you know, there may be a slight upside in terms of your best
  300. guess for what Tua is going to provide based upon what he has done. But then again, we know a lot more
  301. about him. So there, so the distribution is more narrow. So if you're looking for really that upside,
  302. so someone to, you know, much higher record, someone to make a leap into the next
  303. type of category of performance, you're going to find that more so with the quarterbacks who
  304. have, who have less evidence. So, so, so I have that for all of these quarterbacks careers basically.
  305. And I could, you know, I can't update it and I've done this, which people, I've done this as an illustration where it'll show
  306. just actual dropback pick dropback, how it moves throughout, throughout time. And it's, you can really see there are,
  307. it does move up and down quite a bit. There's probably more of a reaction to early
  308. season quarterback, early career quarterback performance than people think. I think people tend to have a asymmetric
  309. view of quarterbacks. They say, if they're really good, then they're going to be fantastic Hall of Famer.
  310. And if they're really bad, then, well, it didn't really matter. And we'll just go and see what they're
  311. going to do their next year. And of course we have to, you know, both of them have to matter.
  312. So, so all of that goes into it, but that being said, you know, all models are wrong.
  313. Some of them are useful, right? So I do think that when people nitpick, you know, the results there, or they look
  314. into the results there, you know, I agree with some, with some of the ideas that
  315. they're, that's things like Patrick Mahomes, we can talk about, we can talk about him, that
  316. maybe he's a little overrated in the model because of how outstanding his earlier performances, even
  317. though I'm decaying that earlier performance, I'm making it worth less than what happened in 2025.
  318. So, so maybe it is over, overweighting and based upon that, but I think it's more
  319. important to just be consistent with how you're looking at it and to try to get the best results.
  320. And as long as you know how it's calculated, then you can kind of make those mental adjustments, educated mentally.
  321. A couple of things. First, I mean, if you, if you start decaying, you're going to inflate the variance, right?
  322. So that's, there's a consequence to doing that. Which, which you're, you're treating it as symmetric
  323. because of the normality, right? So their upside and their downside are equal. Are you doing empirical Bayes or are you
  324. using a subjective prior? What's, what are you doing? Yeah, I mean, this is, this is all
  325. based upon, yeah, what we have for career, like I said, what I'm calling a franchise
  326. quarterbacks is what the prior is based off of. And then for the, for the distribution.
  327. Yeah. So, so that's, that's really what you're doing. You're doing subjective Bayes.
  328. You're, you're, you're imposing a prior. You're not trying to estimate it based on
  329. over dispersion. Okay. Yeah. So this is interesting. Let me ask you a question. So how well, like for how long does
  330. it take? We, we talk about this all the time. Like you mentioned draft position and early performance.
  331. How long does it take for, in some sense, that to kind of wash away? Like how long do I not care?
  332. You know, I don't know. Let's take it. Sam Darnold was, I forget if he was one or two, two, three, three.
  333. Okay. How long do I not care anymore? That Sam Darnold was number three. And how long do I not care that
  334. he performed horribly with the jets and whoever else he was with after that? Like, um, when, when do I start, like
  335. how quickly do things update and how things do quickly wash away? Well, I mean, for, for, for Darnold, and
  336. this is an adjustment that I've had to make where I wasn't doing this before, where
  337. if there's a change of team, and then if, so that's like an additional kind of decay on prior performance than what you've just
  338. have seen, because, you know, with the jets, he, he had a lot of volume in those years.
  339. It wasn't as if he was just, you know, riding the pine the entire time. And we didn't see a lot of them
  340. where he got benched very early on. So we had a lot of volume. So there's a lot of evidence that that
  341. probably weighed into a bit, a bit more. So I have made that, that sort of adjustment.
  342. But what I'll say is draft position, it goes away fairly quickly. And I think it's for a couple of
  343. different reasons. One is just that it's not as important as people think then performances, especially even after
  344. a rookie year, after one season, draft position, it does not have that much performance, but
  345. it's really based upon the fact that if you've been out there that long, there is
  346. like, there's, there's evidence in volume, essentially, there is, if a quarterback has enough volume, a
  347. full rookie season of playing on when it comes to a lower draft pick, it overrides
  348. the signal that you would get from from that fifth or sixth round or fourth round
  349. draft position for the fact that they've been out there that long already puts them into
  350. into a new category. So it's a combination of those things. I think really, after a full season, you
  351. don't care about draft position that that much, but especially after two years in the NFL.
  352. So let me ask you, I'm thanks to Cade for putting your Bayesian quarterback rankings into
  353. the rundown. I'm surprised. Let me just tell our listeners the pattern that I see in this data table.
  354. Let's just take the top predict pick here, which is Patrick Mahomes, which may not be
  355. that surprising. In terms of season efficiency ranks, 2020, number two, 2021, one, 2022, one, 2023, five, 2024,
  356. 11, 2025, 11. So how does he come out at number one for 2026? What is the, you know, what's the data,
  357. whether in the prior or in the, you know, likelihood or the data that's going to
  358. come that would get Patrick Mahomes to number one? I guess it would be based on the
  359. prior at this point. We have no data yet for the season. Why would he be number one?
  360. Yeah. Yeah. I mean, this is, this is also going back to, you know, even further 2019, 2018,
  361. number one, number one in those two seasons. And it's also like the quality of the
  362. number one, where if you're just looking at, let's say from something, you know, something like,
  363. like standard deviation, I mean, he's number one in a way where it's so far off of out to the, to the right, kind
  364. of the high end tail of what you're looking at that. And this would be one of my own
  365. critiques of, of perhaps what the model, what the model is that even though you're decaying
  366. that old performance, he's getting so much credit for those old seasons because it is so
  367. extraordinary versus anything we've seen. I mean, he had and may still have, I'd have to check it against.
  368. I also did like a career QB value of all time sort of rankings, but I mean, clearly up until even through last year,
  369. he had the greatest start value accumulation of any quarterback in NFL history. So when we're talking about that type of
  370. performance, not just number one, and it extremely high volume. Now that's another question with the model.
  371. You could say if we're waiting, you know, every pass is equal to every other pass
  372. essentially. So if Patrick Holmes is throwing the ball, you know, 400 plus 500 times, something like
  373. that, that can also give him even more credit on those outstanding outlier seasons.
  374. So I see the critique of Josh Allen perhaps being better, but it's really about just
  375. how far out he was in advance of everyone else in the past. And in some ways, I think, I think
  376. it could be a little bit of a mistake, but then the one, the one example I point back to, and I think where
  377. this model was helpful doing going into, I guess it was the 2020 season was it had a very high prediction on how well
  378. Aaron Rogers was going to do after he had a couple of pretty poor seasons. I mean, even, even worse seasons than what
  379. Mahomes has had the last couple of seasons, but it looked back to those outstanding seasons
  380. he had in the past where in a similar way, he was kind of head and shoulders above over the previous five years type
  381. of stretch than everyone else to say he has the potential to jump back. And then of course he, you know, he
  382. went on to win a couple MVPs after that. Let me ask you another question. How many times, like when you look at
  383. these trajectories, if you look at Mahomes, let's assume, as you said, he had the greatest
  384. performance in the first seven, eight years of his career of any quarterback. And now there has been some degradation in
  385. his performance. If you looked at these trajectories over quarterbacks over the last 25 years, has there been
  386. a situation where someone's been high kind of come down and then gone back up? I'm just wondering if these, like, if I
  387. just had your efficiency rating on the X on the Y-axis and time on the X-axis, do people kind of have slumps
  388. or like couple year bad periods and tend to go back up again? Or should we just make the assumption that
  389. no, maybe Patrick Mahomes statistically from an efficiency point of view is going to be somewhere
  390. between eight to 12 for the next four or five years. Do you end up, do you see these
  391. upturns again? No, you definitely see the upturns again. I mean, that was part of the exercise
  392. of looking back at quarterbacks careers. Maybe it's a little difficult to see even,
  393. you know, looking back, but I mean, it was a bit of a shorter window, but for instance, if you look at Peyton Manning,
  394. which may be the most similar kind of quarterback to Mahomes when it comes to how
  395. their career started, his fourth and fifth seasons were actually not that outstanding.
  396. There was some talk about him at that point, whether or not, you know, he'd been
  397. figured out and things of that nature. And then he jumped right back after that to win a couple of MVPs the next
  398. season and be, you know, first team all pro for several seasons after that. So it wasn't, the timing wasn't exactly the
  399. same, but what I saw quite often when I was looking at the career arcs and I've kind of plot them out over time
  400. for all of these top 50 quarterbacks of all time is how much they jump around. It's a little bit surprising.
  401. You wouldn't think it would be the case that that much, but it has been the case for a lot of different quarterbacks.
  402. I mean, what's different about the past to today is you had a lot of who started off slow potentially for several seasons and
  403. then took off where we don't see that as much nowadays, but you also see those mid-career slumps that you wouldn't expect to
  404. see as often as you do. Before I get to the team level, let me ask you a couple other things.
  405. Just looking at this list here, I'm a little surprised, but maybe not entirely because statistically
  406. people seem to love him that Brock Purdy is at number five. And to be honest with you, I've seen
  407. this guy shred a lot of people that Matthew Stafford's at number nine. I mean, I saw him almost single-handedly
  408. take the Eagles down, whatever it was two seasons ago in the snow game where if it hadn't started snowing, let me tell you,
  409. Matthew Stafford single-handedly was beating the Eagles. The Eagles were not going to go to
  410. the Superbowl and rout the Chiefs. The Rams were going to go to the Superbowl and rout the Chiefs.
  411. So just give our listeners a sense. Why is Brock Purdy so high? But if that's what the data says, the
  412. data says it. And why is Matthew Stafford maybe not as high as some of us might think?
  413. Yeah, for Brock Purdy, it's really what are we measuring? What are we predicting?
  414. So we're predicting his efficiency, his EPA per play. I mean, I make some adjustments to that
  415. also. But so if we're projecting that, there's really not a lot of arguing that he's been
  416. outstanding by that metric. I think the argument with Purdy is whether or not he's as good as his numbers
  417. point to. So you could have a high projection on him and not necessarily be in disagreement with
  418. the fan or film watcher or analyst opinion that he's not necessarily as good as his
  419. numbers are. So those two things can go together. When it comes to Stafford, I mean, if
  420. you go back far enough when he was with the Lions, and you could say, well, it's a function of playing on the Lions,
  421. and that's why he had a problem. I mean, he wasn't that efficient necessarily early
  422. in his career. He had massive volume. So the numbers have been really, really huge
  423. on him when it came to yards, touchdowns, things like that. But in a simple stat like yards per
  424. attempt, he was never really that great. He had a little bit. He wasn't necessarily that outstanding for his second
  425. and third seasons when he was with the Rams. And what's interesting is I think he's a
  426. little bit of a high point right now because even if you go by sentiment, and one of the gauges I like to use
  427. for sentiment is Mike Sandoe of The Athletic. He surveys 50 NFL insiders. So a combination of front office people, scouts,
  428. coaches, and others. He surveys them and he says, okay, is this guy a Tier 1, Tier 2, Tier
  429. 3, or Tier 4 quarterback? So when we look at Matthew Stafford, so he had 48 out of 50 people said,
  430. 50 of these people said he was a Tier 1 quarterback this offseason. So after last season, if you go back
  431. just a couple of years, it's only 26. It's only half of the people seeing him as a Tier 1 quarterback.
  432. So that's just a huge shift in opinion over the last two years. So I think generally my projections are going
  433. to be more in the middle between where we're going to see some of these swings and projections.
  434. And you would think those swings in perception, and you would think it may not necessarily
  435. be the case, but you would think those shifts in perceptions amongst NFL insiders would be
  436. less than amongst the average fan. But maybe they actually track what the average
  437. fan thinks about these quarterbacks. Well, all I can say is good luck to the entire AFC because the top four
  438. you have up there are Mahomes, Allen, Burrow, and Jackson, which by the way, while a
  439. lot of NFC champions have come out indeed of the last couple of years out of the NFC, looks like quarterback play, you would
  440. agree, at least at the top end, appears to be stronger in the AFC, which I think most people would agree with.
  441. Yeah, yeah. I mean, Joe Burrow, he's got to stay healthy also, but yes, I believe that's true.
  442. Could you talk to us about the team rankings? So, you know, everybody, you know, hope springs
  443. eternal. Any team could go 17 and 0 right now, while the Dolphins can't. But, you know, so where do you see
  444. team rankings right now? And could you talk to us about how much, whether it's flatness or do you, is
  445. this a year in your mind, Kevin, where you see great separation between teams? Are there the obvious greats, the obvious bad
  446. teams at a lump in the middle? How do you see the distribution of teams and who's kind of at the top?
  447. Yeah, I think the distribution is a little bit maybe tighter in the top half than
  448. you would see in some other years, only because if you're going to take like the
  449. top two teams that I have, and I think the market is fairly in alignment with it, is saying that the Rams are number
  450. one and the Seahawks are number two. I mean, these are not necessarily teams that
  451. you would have seen in that projected to be quite so high in previous years. And then when we have the Chiefs who
  452. are maybe a little bit outside of the, of the top five at this point where they would be almost the perennial number one
  453. in those numbers, the Ravens, maybe outside of the top five or six, they would normally
  454. be, you know, a top three or top four type of team. So, and the San Francisco 49ers, which, you
  455. know, maybe going to last year, they weren't quite as high, but in the past they
  456. had been a team that had been in that top five ish sort of range for a while and they've fallen quite a bit.
  457. So I think there's been a, there's been a shake up there. Things are a little bit tighter than they've
  458. been. And also the alignment of those quarterbacks that we talked about that I'm projecting as being
  459. the top quarterbacks in the NFL. You know, they're, they're, they're not also aligning
  460. with saying these are the one, two, three, four, five best teams in the NFL. And it's not just because of, of, of
  461. gaps in defensive performance. I think it's also because the quarterbacks, and this is part of my projections have, there's
  462. a little, a little bit of a lesser degree to how important they are because you're
  463. seeing the raised with what the performances you can get from others. How can give our listeners here on Morton
  464. Moneyball. And again, we're talking to Kevin Cole. Kevin is his, you can look at a
  465. sub stack unexpected points, formerly of PFF. Kevin, if, if I told you, you have
  466. your 14 teams that you predict to make the playoffs, let's say seven NFC, seven AFC
  467. based on your predicted rankings, et cetera. What, I mean, what's anybody's expectation of what
  468. good performance is and predicting how, how many of those teams are going to make the
  469. playoffs? I think we, we had a stat, maybe we said a couple of years ago that I knew for, there was a five-year
  470. stretch where roughly half the teams that made the playoffs in year T did not make
  471. it in year T plus one. So what, what would be good? Like if I told you right now, 10
  472. out of the 14 teams you predict, would you be happy with that? I mean, 12, I don't know what, what's
  473. a number that would represent a reasonable degree of predictive accuracy. Oh, I don't know if I mean, I
  474. mean, again, I'm just a little bit of a guesswork going into this, but 12 out of 14, I think I'd be ecstatic.
  475. It's something like that. When you consider injuries and other things, I mean, I mean, you know, the chiefs did
  476. not make the playoffs last year, right? So some, some crazy things can happen. I think even, even 10 out of 14
  477. is probably a pretty solid number that we're talking about there this year. It could also be a little difficult.
  478. I mean, I'd have to, I haven't run the numbers on it exactly, but we do have some divisional like the NFC West, for
  479. instance, we have a lot of divisional strength in these places where, so I think that's
  480. going to hurt quite a bit, the correlation between team ranking and maybe probability of making
  481. the playoffs generally. So if that's the case, team in that division could end up being the fifth or
  482. sixth or seventh best team in the NFL and not make the playoffs. Correct. Correct.
  483. And I think, I think for the, there's also to some degree in the, in the NFC North, you see that too.
  484. So when you, when you have that, that happening, then, then you're talking about higher playoff
  485. probabilities for teams who are in kind of the thicker part of the distribution of the
  486. NFL. And in that thicker part of the distribution, the differences between these teams are smaller.
  487. So therefore they're more difficult to predict. So you bring all that together and it
  488. makes it a, a difficult game trying to predict exactly who's going to make the playoffs.
  489. Well, what I love it for, thanks again, Kate, for putting the team predictions in the
  490. strengths in the rundown. For my NFC South fans, everyone knows I'm a Buccaneers fan.
  491. As far as I can tell here, at least what Kate put in there, they're the last of the teams that are listed.
  492. Is the NFC South that frigging bad? Like the Bucs are minus 0.13, so a below average team.
  493. But as far as I can see, I don't see the Panthers or the Falcons or the Saints in there.
  494. Are they the best team in the NFC South, but they just stink? Yeah. I mean, I, I have them as being
  495. a slightly negative team. And so does as far as the, you know, and again, this is, this is, this
  496. is expected point differential against an average team on a, on a neutral field.
  497. So the, according to, if you, if you kind of, if you, if you apply the first couple of weeks of the, of the
  498. betting markets and then you, and you build that out in a similar manner, that also
  499. has the Bucs as being slightly negative. I think what happened last season is the
  500. Bucs were the best team in the NFC South, but they didn't make the playoffs. The Panthers had an, sometimes it's about how
  501. your season ends maybe for impression wise. Of course they lost to the Rams in the playoffs, but they didn't embarrass themselves.
  502. So maybe people think they're going to be a little better than they are. My numbers had always been more down on
  503. the, on the Panthers than even the market this year. They're actually a little bit higher than the
  504. market. I would get pretty much a weekly message from friend of the front of the show,
  505. Eric Eager saying that he was, he always enjoyed opening up my power rankings on a
  506. weekly basis to see whether or not they were above the Saints. That's all he's like, that's all I care
  507. about. Can we please be above, above the Saints above the Saints that eventually that eventually did
  508. happen. But, but that was like, that was the limit of his aspirations, you know, not, not
  509. being seen as, as a solid playoff team. So yeah, I think they, it really was that bad this year.
  510. I mean, last year. And of course that's going to carry forward to this year where there isn't an obvious
  511. reason for why a team could get better. Although when we do talk about the Saints,
  512. I guess they have more potential for an outlier bump at quarterback than, than the other
  513. teams. Kevin, how accurate are the, are your, uh, your EPA or points above expected on, um,
  514. in so far as how well do they pick the final season at the end of the season, you refit.
  515. What's the, what's the, how close are they? What's the RMSE or whatever you like to
  516. measure your favorite predictive measure of accuracy? Like when you say something that Tampa Bay
  517. is a half or whatever it is, minus a half, what does that mean to me? Plus or minus what?
  518. Um, well for the number here, let me, let me pull this up here. So on, it looks like the it's, it's,
  519. so roughly we're talking about a point and a half off is, is, is roughly what, what we're talking about in the, in that
  520. sort of measure. So that would be the plus or minus. And we're looking at a distribution that starts
  521. at, you know, the roughly the start of the season that goes, that this may be the total distribution from top to bottom is
  522. about 12 points. So, you know, again, in the thick of the distribution, we're talking about, you can expect
  523. to be off by about, you know, five or six spots in, in, in either direction if you're, if you're in the middle, but
  524. once you start to get to the top or to the bottom, then we're talking about more like, you know, three, four spots at
  525. a maximum of what you're looking at. That's pretty standard result for our listeners on
  526. Morton Moneyball, but in the middle of the distribution, there's, you know, the ranks are going
  527. to move around. Yeah. The ranks are going to move around quite a bit as compared to in the tails
  528. of that. That also becomes a, uh, I mean, well, it's a fun thing for me, but maybe
  529. it's not as fun for the fans as when I send out my rankings on a weekly basis, you know, a big deal is
  530. made of, uh, where the, you know, we were the 21st team and we should be the 13th team.
  531. And it's like, really there's not, there's not much of a difference there. So that, and of course, you know, we
  532. beat team X and they're higher than us. How could that possibly be true? So those, those are the two things that
  533. I think, obviously the second thing is a little bit easier to understand, but even the
  534. first thing we're talking about very narrow margins between, between the teams and team strength.
  535. So Kevin, maybe in the last minute or two that we have, could you tell us about a team that kind of surprises you?
  536. Like which team is maybe even when you, you know, you, you hit the button, the model runs, you get some predictions.
  537. I have one just looking at the numbers, but I'd rather hear yours first. Like one team that surprised you in a
  538. positive direction. And one team that maybe surprised you was in a negative direction.
  539. Well, okay. So, I mean, my, my numbers are highest on, at least when we're talking about top
  540. teams versus the market on the Houston Texans, but I think team I would have picked
  541. out. All right. But I also, I mean, I think the Jaguars are maybe a little bit more interesting.
  542. They have a similar upside in their projection. And I think when it comes to, to
  543. their performance, they've been a team that is consistently underperformed in wins.
  544. What my adjusted numbers would have said that they should have had. And Trevor Lawrence has been a quarterback who
  545. has consistently underperformed what my adjusted EPA numbers are. So then we start to get into the
  546. question of, well, is there just something wrong with this guy, right? Is like, is, is, is, are, are these
  547. adjustments not capturing the fact that he's making, you know, some, he was, he, I mean,
  548. he, he had a tendency to make some turnovers on close to the goal line or in certain situations and things like that, that,
  549. that they had, that they'd gone through. So I think the Jaguars are maybe a team I'm a little bit higher on that
  550. than most are on, on an annual basis. And that actually ended up working out somewhat
  551. last, last year. And they could have even won that playoff game against the bills quite easily.
  552. But the other one would be the Texans. And I'm a little bit surprised by the Texans because my numbers are not that high
  553. on CJ Stroud, who I don't even know where the perception is on him. I was much lower than the perception was
  554. generally going out of his, it was rookie year, but obviously people got very down on
  555. him for some, some poor performances while the team was still winning last year.
  556. Well, we're going to find out a little bit this week because the bills of Texans
  557. is a, well, there are a lot of very interesting games this weekend, including just starting
  558. tomorrow, Patriots Seahawks, then 49ers Rams. The bills of Texans would have to be
  559. one of the highlight games of the week as well. So Kevin, maybe in the last, last question,
  560. last thoughts, what are you expecting for this season? What are your thoughts about the overall league,
  561. the competitiveness of the league? Any surprises? Do you expect Mahomes to come back?
  562. Do you expect the Chiefs to actually, you know, turn around and be one of the elite teams?
  563. What do you think? Yeah, I mean, I guess that would be more, more of my expectation would be for
  564. these teams with the elite quarterbacks to perhaps start to solidify themselves a bit better than
  565. what we saw in particular last season. So as you ran down, whether it's the Chiefs, whether it's the Ravens, I mean, the
  566. bills have been a pretty consistently good team, although they have, they were a team that
  567. people were quite down on, especially two years ago versus what their performance are.
  568. So I think it's for, for those teams to maybe for those quarterbacks to reassert themselves
  569. a little bit more. I mean, Jackson has kind of had some injuries off and on. He's had his playoff struggles.
  570. Of course, Mahomes is coming off of an injury, so we'll see how that performs and
  571. he's had a couple of down seasons, but it would be for those and including Joe Burrow and the, and the Bengals to reassert
  572. themselves. So maybe when we're looking at the playoffs, it'll look a little bit more like it
  573. did a handful of years ago, where it's, whereas Burrow, Jackson, Mahomes, and Allen, and, and
  574. what we're going to see from that, whereas last year, I think we just didn't have
  575. those marquee quarterback names in the same sort of way that we were talking about.
  576. So I would look for, for, for that to reassert itself this season. Well, I'm definitely excited about the NFL season.
  577. I think there is a significant amount of uncertainty this season and it's good to have
  578. you on. So we've been talking to Kevin Cole. You can take a look at his sub -stack Unexpected Points.
  579. And Kevin, thank you for joining us here on Wharton Moneyball. All right. Thanks for having me.
  580. Well, that's been a full hour of Wharton Moneyball on behalf of myself, Eric Bradlow, Cade
  581. Massey, Adi Weiner, in absentia, Shane Jensen. Thanks to our producer and sound engineer today,
  582. Alex Grodnick. Thanks to the boss, Marissa Renna and Deep Patel. Between now and next week, enjoy your sports,
  583. enjoy your analytics, enjoy the beginning of the NFL season, and we'll see you next week
  584. here on the Wharton Podcast Network and Wharton Moneyball.

Brief Summary

Brian Burke and Kevin Cole discuss NFL analytics, player metrics, and predictions for the upcoming season.

Key points

  • Brian Burke's Competitive Spirit. Brian Burke shares his excitement about participating in a time trial this weekend.
  • The Playbook Telecast. Brian discusses his new role in a simulcast of Monday Night Football, focusing on analytics.
  • Predictive Metrics in Football. Brian reveals that their pass block win rate correlates with play outcomes, showing its predictive power.
  • Bull Rush Analysis. Brian explains the significance of bull rush wins and their effect on quarterback performance.
  • Quarterback League Dynamics. The discussion emphasizes how crucial quarterbacks are in determining team success.
  • Brian's Bold Prediction. Brian predicts the Rams might not make the playoffs, challenging media narratives.

Episode highlights

Tension

45/ 100Some tension

Related episodes