Did the Knicks Change How Champions Are Built?

Knowledge at Wharton

Ask about Knowledge at Wharton

Episode Description

ABOUT THE EPISODE The Knicks just won a championship with a roster construction strategy that nobody saw coming. Sports data journalist and former FiveThirtyEight contributor Neil Paine joins Wharton Moneyball to discuss New York’s historic title run, why their formula may be impossible to replicate, and what it means for the future of team building in the NBA. Plus, the crew breaks down the Carolina Hurricanes’ Stanley Cup victory, the role of analytics in modern sports, and surprising early results from the World Cup.

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.

#NBA #Knicks #Basketball #NBAFinals #SportsAnalytics ----- The Wharton School of the University of Pennsylvania is committed to sharing its intellectual capital through Knowledge at Wharton, the school’s online business analysis journal.

Launched in 1999, Knowledge at Wharton offers free access to:

- Articles, podcasts, and videos highlighting Wharton faculty research and analysis of current business trends - Interviews with book authors from Wharton School Press - In-depth series of curated content like Wharton Executive Education’s Nano Tools for Leaders - A searchable database of more than 10,000 articles covering all aspects of business - A weekly newsletter that delivers Knowledge at Wharton insights directly to your inbox

Visit the Knowledge at Wharton homepage: https://whr.tn/3qUpSBE Subscribe to the Knowledge at Wharton YouTube channel: https://whr.tn/3DEAgAU

Read the full transcriptHide the full transcript
  1. Welcome, welcome to Wharton Moneyball. Welcome to a full hour of sports analytics
  2. here on the Wharton Podcast Network. This is Cade Massey hosting today with two of my longtime collaborators, Audie Weiner and Shane
  3. Jensen. Our fourth co-host Eric Bradlow is out this week doing Eric Bradlow things.
  4. Some combination of us are here almost every week of the year. This is big talk, given that we're not
  5. going to be on the air next week. Next week is a rerun week, but almost every week, 48, 49 weeks of the year,
  6. we're here. Been doing it for more than 12 years now. Looking forward to the next hour.
  7. We are going to run our usual format. This is Tuesday afternoon. We'll record, put it up on Wednesday morning.
  8. We've got it going in your very favorite podcast locations, but also we have video these
  9. days on YouTube if you want to catch us that way. Our format today, guests in the first half
  10. of the show, open lines in the second. Our guest, our, you know, if anyone gets credited as the OG, the OG Wharton Moneyball
  11. guest, it would be Neil Payne because Neil was there in the beginning when we were
  12. still in the basement of a UPenn building and Neil was in Philadelphia, to which he
  13. will return soon, but first he's going to visit with us. It's been a little while since we've had
  14. Neil on the show. He's probably the leading, most repeated guest on the show. We're always happy to have him.
  15. Neil, if you don't know, a sports data journalist. He really is this mix of good with
  16. numbers and words. He was practically a co-founder of FiveThirtyEight, did a lot of work there.
  17. You can find his work in a lot of places, but especially his sub stack. These days, Neil Payne, always good to see
  18. you. Thanks for making time for us. Hey, thanks for having me back on and always great to see you guys.
  19. Good to see you, man. And your stack O-hats. I'm especially seeing the Orioles and the Dodgers
  20. and the Colts are peaking out today. I'm going to focus on the Pirates. I see the Pirates.
  21. Pirates are in there. We've got some Pirates talk. I think Shane has queued us up for.
  22. Let us begin. We have just, we've got the two longest playoff seasons are basketball and hockey.
  23. They run concurrently. Both are a lot of fun. And this year they were kind of remarkably
  24. parallel and they both wrapped up one night after the other, both with 4-1 wins in the final series and both champion teams
  25. making 16-3 playoff run. 16 wins to get to the championship, of course, with only three losses, both very good
  26. runs in the playoffs. We're talking, of course, about the New York Knicks winning for the first time in, what
  27. was it, 52 years, 53, 53 years? And the Carolina Hurricanes, which 20 years might
  28. sound like a long time, didn't seem like it, not really a long time. 20 years shouldn't be long in Stanley Cup
  29. repeats. That's their second one. I think of them as an expansion team. They're not an expansion team.
  30. They're the former Hartford Whalers. Hell, that probably makes them an expansion team,
  31. but they've been in Carolina since 98 or so. Okay, Neil Payne, open us up. Take us whichever direction you go.
  32. Which of those 4-1, which of those champions, Knicks or Hurricanes, are you more impressed
  33. with? They both had 16-3 runs. They both had really, really good playoff runs. Who are you more impressed with?
  34. Well, I think it's the answer based on just the playoffs alone has to be the Knicks.
  35. They set a new NBA record for the greatest points per game differential, and I think
  36. overall point differential, because they're synonymous with each other in the longest playoff format now.
  37. And that was just incredible. If you think about where this team came from, and how it was built, and how
  38. many years they had been trying, and then they kind of finally get this breakthrough, and
  39. even to the finals, even midway through the finals, even during games of the finals, certainly,
  40. nobody really thought they could do it, but they just kind of kept winning. It kept coming back and put together the
  41. goat comeback as part of that, the 29 -point comeback. And it just felt like if there's ever
  42. been a team almost in my lifetime that was like a team of destiny, capital T, capital D, it felt like this Knicks team
  43. was that. So I was more impressed by them. Now, the other question is, who do we
  44. think is more sustainable going forward? And I think probably Carolina. They've been here for a long time.
  45. They famously had also tried really, really hard for many years to get a breakthrough to
  46. go to the finals, longer than the Knicks did, and have been one of the upper -tier hockey teams for a long time.
  47. And they just have such a sustainable program in place that feels like it can ride
  48. out the chaos and randomness of hockey. And there was a lot of that in this series as well, to be able to
  49. have long-term success. So it's kind of two sides of the same coin, depending on what you're looking at.
  50. But I subscribe to probably the first definition of just like who was most impressive during
  51. the postseason and during the finals and what we just saw. And that was the Knicks.
  52. If anybody had that on their bingo card of like, hey, who's going to break the, I think it was the 2017 Warriors or
  53. 2018 Warriors or maybe the 01 Lakers, they were all in that conversation, 71 bucks record
  54. for highest point differential in a single postseason. I do not think anyone on the planet,
  55. except like maybe some rabid diehard Knick fans hanging around outside Madison Square Garden would have
  56. thought that that was even remotely possible. And yet here we are. Here we are. Neil, it was only a few weeks ago
  57. on the show, I remember saying, well, it's just presumed that the winner of the Western
  58. Conference is going to win the title, right? It's who's going to win, thunder or spurs.
  59. Here's a Western final for the whole, for all she wrote. That's kind of where everybody was.
  60. What do we take away about, I mean, what, I guess the main question is like roster construction.
  61. It's always a mystery in the NBA. What is, there can't just be one way, but what do we take away?
  62. Because nobody expected this team to do it. They didn't do it in following any pattern
  63. that other people have been following. So what do we take away? Yeah, that's a, that's the $6 million question,
  64. right? I saw some tweet where somebody was like, the hilarious thing about this Knicks run in
  65. the finals is that like, nobody can really copycat anything off of it. It's impossible to replicate the specifics of like,
  66. okay, go out, get an undersized guard that nobody thinks can be the best player in
  67. a championship team, sign him to what seems like an above market deal, but ends up
  68. being wildly below market. Then go out and get a bunch of guys who played together in college, but trade
  69. one of them away and get a guy that was considered soft and incapable of stepping
  70. up in the playoffs. Then fire your coach after making the conference finals multiple years and get a guy who
  71. had been kind of floating around as a coach for many years without winning the championship
  72. as the head coach, and then put it all together. And lo and behold, you have the best
  73. playoff team of all time. There's just, how do you replicate that? But I do think maybe it should just
  74. expand our horizons. And if that hadn't already been true, when we had previously seven different champions in seven
  75. years, now we have eight in eight years. Whether it's the second apron, you guys know
  76. all the theories abounding about why the NBA is potentially quite different now than it was
  77. in its entire history before this. But I think maybe we just need to expand our horizons around what a championship team
  78. looks like and how to build one, because it's like a totally different paradigm than the
  79. one that we were used to 10 years ago when the Warriors and Cavs were going to the finals every year.
  80. And basically the championship hinged. You could almost book it on the day that the free agency tweet was made by
  81. Shams or whoever, Woj, about where KD was going and where LeBron was going. You basically say like, okay, well, that's your
  82. championship in the summer before the following season. You can't do that anymore, in part because
  83. there isn't that crazy free agent frenzy because of the restructuring of financial rules, but also
  84. because it's just a lot more chaotic and random than it used to be. And a lot more players have led teams
  85. to championships than the types of players and the small number of players that had done
  86. that in previous eras. So Neil, you've gone from roster construction, gone to team-level analysis to league-level analysis.
  87. And I want to come back to roster. Acknowledging that eight unique champions in eight years
  88. is cool. I think this is something that most leagues would have striven for, right?
  89. And so that's great. But let's come back to roster. And I wonder if there is, you know,
  90. one way you're saying it is no one knows, you can't say, you wouldn't replicate this
  91. what they did, like in terms of which type of player you get, which type of coach you get.
  92. But is there, you know, think about it like a feature engineering exercise. So if what you just described was the
  93. data, if this was a big data set and we were going to then do some analysis on it, we would engineer some features
  94. out of that raw data, some higher level features. And could we think in that, like, can
  95. we abstract up one level from 6-1 guards to what qualities the roster provides you?
  96. Might that be a productive avenue for thinking about what was this, what were the key
  97. features in this team? And again, we're acknowledging there's not one way to get it done, but is there something
  98. we can learn? And maybe at that next level up, as opposed to the fundamental, you know, player level
  99. stuff. Well, I think it's probably true that like their fifth best player, whether that's, you know,
  100. Josh Hart or Bridges or someone like that, is probably much better than sort of the
  101. median fifth best player on an NBA champion. And I think that that is maybe one of the things that you could take away
  102. from this is they really invested, yes, they invested in Brunson, but they also invested a
  103. lot, arguably more, because I think Ananobi and Bridges make as much or more than Brunson
  104. does, and certainly Towns does. They invested heavily in the supporting cast, like
  105. the rest of the starting five, and they all really complement each other well. They're very versatile players.
  106. We saw each of them become like main characters at different times. I think we all recognize that Brunson was
  107. sort of the engine that drove everything, but there were times in which like Bridges got
  108. hot and kind of carried them early in the series. Ananobi, of course, hit probably, people are calling
  109. it the hand of God, you know, maybe that's a soccer reference kind of coming in as well.
  110. But I think that tip in was one of the biggest shots that anyone ever made in the history of the NBA finals to
  111. kind of cap things off in game three, or game four. So really, to me, it's just maybe the
  112. lesson is you know who your kind of offensive creator is as the key centerpiece, and
  113. it's okay to build around a guy that maybe has defensive shortcomings. He literally has shortcomings height-wise.
  114. If you get a good enough, versatile enough, and kind of just all-around strong core
  115. around him, especially the rest of your starters. But you could get a lot of false
  116. positives doing that, too, because we've seen a lot of teams be built that way that
  117. kind of hit a wall, and they can't break through. I think that's one of the interesting things
  118. about the Knicks is they were hitting that wall, too. They had the most, or I think second
  119. most, playoff wins in a three-year span among a team that didn't make the finals
  120. to then go on and make the finals the following year. The other team in that category was the
  121. 1991 Chicago Bulls. So, you know, you think about the Bulls fighting against the Pistons for so many years,
  122. finally getting that breakthrough. We don't think of the Knicks in that conversation.
  123. Maybe we should because they really also tried for many years to get through, and most
  124. of the teams in that category that win a lot of playoff games but don't make the finals, they fall by the wayside in
  125. that fourth year because things change. The team gets blown up. There's just, it's impossible to sustain that with
  126. this sort of failed proof of concept or seemingly failed proof of concept, but they stuck
  127. with it, and they were rewarded for doing that. So maybe you'll actually see more patience in
  128. some ways as well. They changed the coach. They did change coach. That's a big deal.
  129. It's easier to change the coach than it is to change the players, and especially in
  130. this case because they were kind of capped out, and the players were almost borderline untradable
  131. in some ways. So they pulled the only lever that they had, and maybe they kind of backed into
  132. the right way to do it as a result. Right. Guys, as statisticians, what summary statistic would you
  133. like to conjure to illustrate how unlikely the Knicks win was given the number of massive
  134. leads that the Spurs had over various, in what, four of the games? So for example, a colleague sent me, Adi,
  135. hold on. A colleague sent me a stat saying this is the most amazing stat I've ever seen,
  136. that the Spurs led 72 percent of the time. I think that's impressive, and I'd like to
  137. see that compared to other champions, but that doesn't capture the size of the lead, the
  138. magnitude of the lead. So what could we do? What could we conjure to capture that?
  139. Adi, sorry. So first of all, that's very hard because assessing a statistic after a game is over
  140. is subjective, subject to selection bias. So actually, I gave a talk today on the maximum win probability of the losing team,
  141. a paper that I wrote with my grad student Jonathan Pippin, and one of the main points is that big, essentially, blown leads happen
  142. much more frequently than you think. And actually, the simple point is that two teams that are equal in talent, the team
  143. that loses is about 50-50 likely to have a big lead at some point in the game, and that's counterintuitive.
  144. And also, the point which I've made many times, typically, one team will often go out
  145. and lead most of the game, and that's the arc sine law. If you model a basketball game as a
  146. random walk, which is a reasonable thing to do, you get those sort of funny statistics,
  147. and most of the public doesn't realize the actual behavior of these random variables.
  148. But what the Knicks did is ridiculous. I mean, we just know that intuitively. I mean, they came back from a, what
  149. was it, 29-point deficit at some point, 20 points in the fourth quarter. Almost every statistical model suggests that's one in
  150. several, I mean, it's a thousand event or approximately. Yeah, I think so, 99.6. I mean,
  151. again, I mean, I'm kind of with you, Audie, what you were first saying. I think lead shifts, I mean, I would
  152. like to know what the probability is, like, of, you know, a blown, I know it doesn't have, it's notable because it happened in
  153. the finals. Right, of course, it's notable, but I will say one thing about the- I feel
  154. like we're underestimating the denominator of these kinds of occurrences, like in basketball games.
  155. It seems like the swing can be dramatic. The simple point that having a lead of that size generally isn't blown.
  156. I mean, that's just the definition of a win probability that size. But I will say something about NBA win
  157. probability models, at least the public ones, and that's the only ones that I have access
  158. to. Our work does show that the NFL win probability models are much more calibrated than the
  159. NBA. The NBA ones are bad. And most, particularly where they're bad, they give way too much weight to a big lead,
  160. and particularly late in the game. Those things happen, I think, far more, much more frequently than the models undertake.
  161. And I think a lot of people's intuition, we almost, you didn't present it this way
  162. at all. You didn't do this, but like, you know, like conditional, and maybe this is what you
  163. were getting at, Cade, with your kind of question of the general is, you know, conditional
  164. on there being one big point lead in the series, one game going that way, do we then just apply every other game as
  165. an independent event to multiply out these very rare events? No, I mean, maybe something- You can't.
  166. Maybe these aren't correlated with each other. Whatever is systematically happening makes it more likely
  167. to happen. I think what's important, we got to bring this front and center. And we've been talking about the Knicks the
  168. entire time. Can we talk about the Spurs? I mean, the greatest, frankly, the greatest ex
  169. comment I've ever seen in my entire life was the Spurs are really great at taking
  170. the lead, but not so great at holding the lead. And the really funny thing about the lead
  171. is the important part is the holding. That, of course, is a riff on Seinfeld with the rental car reservations.
  172. But can we talk about the Spurs? That's the thing that I don't quite understand.
  173. If you flip it over, are they really part of this outcome? Did they blow it in an epic fashion
  174. based on psychological shortcomings, inability to play when it gets- all these things that we
  175. - It's not as easy, I feel like, as football where you can be like, oh, well, they stopped running the ball or whatever.
  176. It's hard, at least for my- I'm with Adi. It's hard for, I guess, a little bit more of a naive person to be like,
  177. are there signs that they take their foot off their gas in some terrible way or whatever?
  178. Well, they did make some mistakes late in those games, for sure, like, you know, in
  179. terms of shot selection. And then especially you think of like, you know, potentially dribbling out the clock in one
  180. of those possessions, the one where Ananobi came down and blocked the shot anyway.
  181. A 29 point blown lead hinges on like if any one single- It was similar to the 28-3 with the Patriots and
  182. the Falcons. If any one play in either of those games goes differently, the leading team still wins.
  183. So like every single play mattered in that. And if you make a kind of boneheaded
  184. decision at any point, it can kind of open the door for them to get back in. Now, I saw some people saying like the
  185. Spurs were one of the youngest NBA Finals teams. We know that experience matters in these things
  186. and that maybe you could chalk up some share of the collapse to that. It certainly was a collapse.
  187. I mean, they had these huge leads. They blew a 14-point lead in Game 1. They obviously blew the 29-point lead in
  188. Game 3. And so it just is like, there's something there to that. And I don't know if it's just more
  189. time on this stage together. Because even Wimby, you saw some of the plays where like he passed it to a
  190. guy who wasn't looking at the ball to turn the ball over and kind of give another possession back to the Knicks.
  191. I do also think there's a New York media bias, though, to kind of make it be like, oh, the Knicks won this more
  192. than the Spurs blew it. Because if the shoes were on the other foot, we absolutely would be talking about, and
  193. think back to like John Stark's shooting performance in 1994, where they would be killing the
  194. Knicks for losing these games in the media if the shoe was on the other foot. But because it's like, you know, relatively small
  195. market San Antonio, we've seen them win before. The presumption is that Wimby is going to
  196. win a bunch of championships in the future. And this is just part of the growing
  197. process that I think we're maybe letting them off the hook a little bit compared to
  198. how we would approach it if it were another team. And it's an important question because it speaks
  199. to how repeatable this Knicks experience is for next year. I mean, everything went their way in this
  200. little run, I feel like, to the championship. Are they even the betting odds favorites for
  201. the champion? No. They're not even second. Yeah. Yeah. So I guess a lot of people are
  202. attributing at least a lot of this to kind of the breaks going their way. Well, the ESPN was, or whoever was broadcasting
  203. it, was good about showing field goal percentage, especially I think in that last game.
  204. Before the game started, and then it's bore out again, San Antonio's field goal shooting percentage
  205. declined monotonically from first quarter through fourth. And dramatically, monotonically and dramatically, it was almost
  206. half as good in the fourth quarter as it was in the first. You want to give the Knicks some credit
  207. for that, and presumably analytics would help us parse it in some way, but there were
  208. plenty of shots that looked like good looks that they just weren't getting down.
  209. And sure, I think they deserve a lot of... And this isn't a trend that they showed
  210. at all during the season, right? This was something unique to the finals. Right.
  211. Good question. That would be... Right. Well, what's even the base for that? Yeah.
  212. I mean, maybe if that's just something that they had in their general kind of characteristic,
  213. it made them even more vulnerable, I guess, to the Knicks. Well, and I don't think you win 62
  214. games by having that tendency. It's so tough to kind of put, especially in the NBA where most games kind of
  215. don't matter throughout the year. You have this massive sample size of games that only a handful of them matter.
  216. Maybe it would be interesting to go back and look. I've kind of forgotten how the NBA Cup...
  217. I mean, it's silly to mention it, but that was sort of a previous example of these two exact teams facing off on kind
  218. of an elevated stage. And the Knicks ended up winning that one as well. It's so tough to figure out what is
  219. signal amidst the noise of the long and mostly meaningless NBA season. Can we bring this back to the...
  220. Sebastian, you should call back to... Go ahead, Odd. Just want to bring it back a little
  221. bit to the hockey and the Hurricanes, not that they... We'll come back to them. But as sports analytics types, I mean, we
  222. typically think of the Hurricanes as one of the essentially sports analytics-oriented team from ownership
  223. to management to their whole process. And this, their victory and their history vindicates
  224. that approach. I don't think the Knicks are in that camp at all. I mean, I think this is...
  225. We can't even... You couldn't even articulate why they won. I mean, it's almost like the anti-analytics.
  226. So is it... I mean, how do we square this? I mean, we statisticians who believe that this
  227. is the answer. What do we do with these two data points? Well, the null hypothesis is it was luck,
  228. right? Yeah. Or just that sometimes it's rare. And I think maybe we can effectively most
  229. years treat it like it's almost impossible for a team to catch us by surprise, whether
  230. we're looking at our metrics or talent or whatever we think the betting odds going into
  231. a season, that we have this assumption that it's always going to be one of our kind of core group of favorites.
  232. And I think OKC was the team that was the team to beat most of the year. Boston was in that conversation.
  233. San Antonio worked their way into that. And we kind of said, hey, we've seen enough of Wimby.
  234. We can put them in that tier. The Knicks were at the, maybe the edge of that tier, certainly before they started going
  235. crazy in the playoffs and dominating everyone. And even by the end, I mean, San Antonio was favored going into the series.
  236. And it looked like when the Knicks seemed like they lost that game three, that the
  237. momentum was shifting away. People were hopping on the Spurs and being like, OK, the Spurs are going to actually
  238. come back and win this. And then New York shut it down. But my point is that we sometimes forget
  239. that there are every so often, maybe like once a decade or so, weird teams that sneak into the mix and through some combination
  240. of luck playing above their heads or just being way better than we expected. They do win the NBA championship, the twenty
  241. nineteen Raptors, the twenty eleven Mavericks come to mind. The Royals won the World Series.
  242. Yeah, the Royals. I mean, baseball is its own like baseball that happens quite frequently.
  243. Basketball is a random game, predictable sport, the sport where the favorite wins more often.
  244. And even there you have examples. The 0 4 Pistons were another team in that conversation.
  245. So probably like once or twice a decade, you get a team that's not one of the absolute cream of the crop teams as
  246. we expect them to be going into the playoffs comes through and still wins regardless.
  247. And sometimes we weirdly just kind of discount that possibility or it's just like, well, you
  248. got to pick a favorite, right? You know, of course, we were going to side with Oklahoma City as having the highest
  249. odds. But maybe we need to back off on the degree to which we favored the top teams, especially now in light of what we
  250. were saying earlier about the fact that, hey, eight different teams have won in the past
  251. eight seasons, which is the first time that's ever happened in NBA history. That's got to mean something.
  252. Yeah. Can I try a hypothesis out on you, Neil, and obviously all of you guys. Brent Tillis was on our show speaking of
  253. Carolina a few, I guess, a couple of months ago. And he made a very interesting statement, which
  254. almost speaks to the area in analytics that we fail to model. He said that we don't appreciate how the
  255. top player is able to make everybody around them better. And it's essentially, mathematically, we would call that
  256. interactions. So when we model in sports, almost always it's additive. We just add up the contributions, even in,
  257. it makes most sense in baseball because your contributions really are, for the most part, independent
  258. of the rest of the contributions of the rest of the team, makes the least sense in basketball, if you think about it.
  259. And I don't know, I'll have to turn to you, Shane, and Neil to tell me where hockey is on that.
  260. But in basketball and football, it really matters how you work together as a team, as
  261. a group. And maybe the Knicks figured that out in a way that we just haven't modeled.
  262. And what they did was get a bunch of players that really work together. And I don't even remember this, but they
  263. were all together in college. And we've heard, Dean Oliver has been on our show and has talked about research that
  264. they've done and how basketball players that play together become better as a pair, dramatically, throughout
  265. the season. And I guess that would be magnified over a course of the career. So maybe we and the analysts are missing
  266. it. We haven't modeled it. And maybe the Knicks are leading the way, or maybe I'm just bullshitting.
  267. What do you think? I mean, no, go ahead, Shane. No, no, go ahead, Neil. Okay. Yeah.
  268. I mean, I think there have, first of all, the things like the adjusted plus minus
  269. type metrics, those try to pick up on at least an amplification of sort of the best player being able to make the team,
  270. make the teammates better by essentially taking away the individual stats from the teammates and attributing
  271. them to the star player who kind of made that more possible. And so there's that credit splitting.
  272. I will say though, I hate to interrupt this, totally additive model. That is not an interaction model.
  273. Oh, it is. Oh, for sure. I'm saying like, in effect, it's forcing into an additive kind of framework, just essentially taking
  274. away credit from, linearly from the teammates and giving it to the player, even if it's
  275. not appearing in their individual stats. So that's one attempt to do that. I also think there's been efforts to try
  276. to model at the lineup level, try to kind of predict what combination of skills and
  277. attributes of players will succeed together. That's kind of a holy grail type of
  278. thing in basketball. And I think every team that has analysts are trying to do that in some way,
  279. shape, or form. The problem with lineup data is it's notoriously kind of low sample, noisy.
  280. There's a lot of garbage time that you have to kind of get out of the system. So there's huge challenges to that.
  281. And oftentimes, yeah, it's the matchups as much as you want to try to find like proxies for the upcoming matchup.
  282. Like when you've got Wimby going up against a small guard in Jalen Brunson, you know,
  283. you might try to model that, but it's very difficult to find like seven, five guys
  284. that are rim protectors against like five foot 10 or whatever he is lead guards.
  285. So yeah, the specificity of the past data to your particular problem going forward is always
  286. going to be a challenge. And I think that's why basketball is one of the most complicated and I think interesting
  287. sports. Like somebody was saying, and I think this is funny, you guys will kind of appreciate
  288. this. I saw somebody being like, well, the analytics didn't see this coming. So it means that it's kind of worthless
  289. and it's like, okay, well, here's a quick guide for you guys. So if the analytics predict the winner correctly,
  290. then the analytics are ruining the game because they're making it too predictable.
  291. If the analytics don't predict the winner correctly, then the analytics are worthless because they're wrong.
  292. I hope that that can kind of clear things up either way, the analytics are going
  293. to be to blame for things. But honestly, like, isn't this why we watch sports? We love things like this as fans to
  294. see a team that basically by all logic and reason and past performance and all these
  295. things you would not think would be particularly likely to win the NBA title, much less
  296. have the greatest point differential in postseason history, just goes and does that.
  297. Isn't that, I mean, that's pretty awesome. I think even though it like confounds us
  298. as predictors and analysts, I think it's cool. Neil, I'll just say two things. First, I'm a Knicks fan.
  299. There you go, everyone. But mostly ignoring that because of all those years, the earliest championship that I can remember
  300. growing up outside of New York was 1973. So that's the first one I remember. Now it's the last one that I remember.
  301. So it's a short period of time. Hopefully a New York team will do something again pretty soon.
  302. I wonder which one I'm thinking about. You all know, but not likely. But you know, when you mentioned that, I
  303. mean, we love the game as game, right? It's so exciting, particularly when you're a fan
  304. and when you have a rooting interest in the team that does this. But as analysts, let's just stand back.
  305. We want to crack it open. I mean, and that's kind of the goal. And I'm wondering whether the big gap in
  306. our analysis has been combinations in basketball. We really haven't done a good job figuring
  307. out how to model players that work together, combination skills that work together, simply because it's
  308. hard. But now the date is coming, and maybe we'll move forward. Yeah, and I'll only push back in that
  309. one of the reasons we choose additive models and not interaction models is we typically don't
  310. necessarily have the data to support that kind of that interaction model. It's more a matter of faith than it
  311. is kind of an empirical thing. And so for I mean, New York, we will get some data.
  312. I mean, again, we keep talking about like, what's going to, you know, prospectively, you know,
  313. they had nobody could argue they had a gigantic positive residual by this year. If there's any signal in that, you know,
  314. one piece of evidence that their signal is they keep going next year, if all of a sudden, like it bounces back down to
  315. zero or negative next year, when there's some amount of you know, that's, that's some observation
  316. on whether or not they've kind of figured out sort of a team building matchup kind
  317. of thing that's, you know, kind of sustainable, or whether it was more of a kind
  318. of Yeah, they caught lightning in the bottle, basically. Either. I mean, I don't want to take anything
  319. away from their achievement. You know, they caught lightning in the bottle. That is like an established fact.
  320. Well, I, for one thing, 29 point comebacks are totally sustainable. Right? Yeah, no, that's, that's the new way to
  321. win. All right, go down by 29. On that note, we're gonna talk the remainder of the sports high on the agenda.
  322. We talked the hell out of the Knicks and basketball. We alluded to some hockey stuff.
  323. The hockey stuff is super interesting. I mean, is there any, what was it called?
  324. There's an ESPN, great ESPN article out yesterday about the nerd jock marriage with the Hurricanes,
  325. the Tolsky Brindamar. It's really an extreme version of nerd jock. It's hard to get more jock at a
  326. head coach than Brindamar. It's hard to get nerdier than our friend Eric Tolsky was on the show just a
  327. couple of months ago. But that combination certainly was effective for them, has been effective for them the last couple
  328. years, even as an assistant GM, Tolsky's been there. So maybe a little bit on that.
  329. And then World Cup, of course, I would love to hear y'all's thoughts on the, the draw fest that it has been so
  330. far. 50% of the games coming into the day have been draws, whereas the international average
  331. is something like 22. Now, technical, technical challenges on my end, I'm gonna step away.
  332. Audie and Shane, Neil, you guys are doing fantastic. Leave it to you. Wish I could join you.
  333. Y'all have a good time. Aud, safe travels back. Neil, great to see you. Shane, thanks for everything.
  334. All right. So let's bring us back to hockey. I'm gonna let you guys take, take over
  335. this, this wrap up. We talked about the Hurricanes and the, and the World Cup, not the World Cup, the
  336. finales of the, the final of the hockey. But what do we got here? What, what, what, what's, what's the quick summary?
  337. Well, I think one of the interesting things and I'm curious what you think about this,
  338. Shane, as well, is if all you knew was the top line result, Carolina Hurricanes win
  339. the Stanley Cup, certainly, like we said, been one of the best teams for a long time.
  340. It made sense that they would have their breakthrough. If you knew anything else about how that
  341. happened, though, you would be like, this is just the most chaotic possible outcome, both in
  342. terms of there was a goalie switch in the middle of the Stanley Cup final, which
  343. kind of turned the series around for Carolina, was like the single biggest factor.
  344. I know Freddie Anderson was also hurt. They revealed later, but going to Brandon Bussie,
  345. the rookie backup, he wasn't the backup during the season. And that's, I think what just makes hockey
  346. so bizarre is they decided to go with Anderson at the start of the playoffs. It was kind of a toss up, but
  347. they gave it to the veteran because he had been there before. And, and he goes off and has just
  348. a, he looked like the favorite to win the consummate trophy. If it wasn't Mitch Marner of the golden
  349. Knights through the first three rounds of the playoffs. And then all of a sudden he's not
  350. playing so well in the first few games of the, um, the Stanley Cup final. Nobody is really playing that much defense or
  351. goaltending all that well in those there's like a record or near record number of goals
  352. and certainly comebacks just galore early in that series. Brendan Moore makes the switch to Bussie and
  353. net. He comes in and relief. He actually enables them to do a record setting for goal third, uh, third period comeback,
  354. uh, in game three, they end up losing it anyway, but it gave them the confidence
  355. to go with him going forward. He's the goalie of record the rest of the series and they kind of shut them
  356. down. They had a shut out in the deciding game, but also he had like a nine 50 save percentage.
  357. And that was really the hinge point in the series. Um, as far as Vegas went, they stopped
  358. converting as many chances. They had a number of prime chances, especially in the clinching game that they just like
  359. hit a post or they sailed it over the net. Uh, and it was really that difference between
  360. them finishing and not. And also early in the series, Jordan stall ended up winning the, um, the cons my
  361. trophy. And he had really not done that much of note and was not really on many people's radar to win playoff MVP, uh, up
  362. until having a five game goal streak in the first five games of the Stanley cup final, which ended up kind of clenching him
  363. that as the oldest and one of the lowest scoring playoff MVPs. Uh, so I just felt like the series
  364. was completely resistant to narratives aside from if you just zoom out and look at the
  365. big picture. Yeah. Carolina was probably the best team and they ended up winning, but Shane, what are your
  366. thoughts? Well, I think one thing with Jordan stall, I mean, I think specifically, I think he
  367. is an interesting sort of, I guess, Stanley cup finals MVP, because I think obviously I
  368. think he was kind of given, given it based on his, his, his goal scoring, but he's not really not the finals MVP.
  369. It's the playoff MVP. Yeah. I misspoke. He was basically, I think given the playoff
  370. MVP trophy based on his finals performance. And I think that, I think in most people's mind was driven by the goal scoring,
  371. but it came from kind of a surprising place because he's, I think famously like, you
  372. know, an amazing sort of defensively oriented center. And I, I'll kind of just point out
  373. one of the things he, all other things besides goal scoring, he didn't have finals.
  374. He won face offs at a 68% clip, which is, I think a record since that actually has been tracked in hockey over
  375. the last 30 years. So that I think his contributions basically went, went beyond the goal scoring, but I agree
  376. with you in general, it's, it's a, it's a hard kind of series to put a really strong narrative on.
  377. We were talking a little bit in the basketball about like kind of, you know, luck
  378. versus something systemic. I mean, I would like to say that I think in general, one of the reasons
  379. Carolina was able to kind of manage that goaltender shift and everything like that and, and
  380. kind of feed off it is that they are generally a very good defensive team, but Neil, as you pointed out, they weren't playing
  381. very good defensively in those first few games with Anderson and net all of a sudden
  382. they switched to Bussey. And I feel like the defense also got better in front of the goaltender.
  383. And I don't know if that speaks to like some kind of, you know, Audie, you were talking about interaction kind of effects.
  384. I think defense, goaltender, it's subtle. And maybe we're over, over reading this one
  385. situation, but I think that that's an example of a hockey interaction that can kind of,
  386. can kind of happen at least locally. Let me ask you guys a question. Our resident hockey expert, Shane and Neil in,
  387. in soccer, we're going to get to that in a moment. You have XG, which is really important because
  388. it's very low scoring basketball. We never bother with it because it's so high scoring.
  389. We don't really think about points above expected. How does the, is there's, there is an
  390. XG for hockey is how'd that play out? I mean, did the hurricanes win on the substantially on the XG score, but just squeaked
  391. by more on the G score or vice versa? Or what, what's the story there? Well, the hurricanes per, well, the hurricanes perennially
  392. dominate XG because they just have such possession of the puck. And they're sort of constantly creating these sort
  393. of positive XG events at all times. So maybe the, the difference there is that like the finishing quality was better than it
  394. has been in those playoff losses that they've had in the past. It's like, they're always going to have opportunities.
  395. It's just a question of whether they eventually cash in on them at sort of a higher than expected rate.
  396. And they finally got that, but certainly they did have, I believe the, the lion's share
  397. of the events, you know, whether it's like scoring chances, expected goals, all those things during
  398. the course of the series, they had let's see, there's a great site called natural stat
  399. trick that I want to shout out. That is my kind of go-to source for this metric.
  400. And so I think that that's one interestingly enough. So real-time correction on air while Carolina
  401. had the 56% of the course events, those are like shot attempts basically. And they also had 52% of the
  402. scoring chances. They actually only had 48% of the expected goals in the series.
  403. Now I don't have broken out by game. So I'd be curious as to see like that typically speaking XG per 60 minutes on
  404. D like allowed is a good measure of defense because it basically says like, is the
  405. other team getting shots in an area of the ice? That's like particularly dangerous.
  406. So I would expect, and again, I'll have to look at it because I was wrong. I was expecting them to have higher XG
  407. percentage overall that they tighten that up as the series went on, but maybe it's also
  408. like, really high XG on that one shot that he missed right in front that would have, I believe tied the game in, in,
  409. in game six. Yeah, no, no, I do think Carolina in general are kind of masters of that because
  410. of the, they're typically pretty good defensively are masters actually. Like I think that Montreal Carolina series, the
  411. Sierra, the semifinal was ridiculous in terms of kind of the, the, the differential and sort
  412. of expected goals and that, you know, kind of eventually realized in an outcome as well.
  413. But I do kind of feel like, again, it maybe comes back to a little bit what we were talking about before of the
  414. kind of feel like Carolina somehow got out of their defensive mode or maybe, you know,
  415. the, you know, the, the, the Knights were just kind of getting, uh, getting some, uh,
  416. lucky opportunities in those first few games. I do kind of think somehow the goalie
  417. switch also led to this kind of defensive or, or at least them kind of getting their usual defensive group back, I guess, is
  418. what the way I would maybe describe it. So I'm just going to pipe in there and say, I'm always looking for projects, um,
  419. for our students and Neil, you're always looking for things to write about, but I'm kind
  420. of curious, maybe you know the answer. What is the rough approximate relationship between XG
  421. differential and winning on the game level and XG differential and winning at the, at the
  422. seven game series level. And I would guess that the latter one is probably higher.
  423. Uh, usually at the, at a game level, there's much weaker relationship between the XG differential
  424. and winning, but cause it actually went the other way. I think it's pretty surprising.
  425. I mean, and, and obviously one of the things that we just finished a, uh, a, a hockey data competition and the inside baseball
  426. that we put into our little nugget into the model was the importance of the goalie.
  427. And because goalies can, can really make a, a different, a real, real, uh, important contribution
  428. to the, the dampening of the XG cause they can stop shots and, and, and that can change it.
  429. So what do you think about that as a project? And, and maybe that it's really just a
  430. goalie performance that, that saved them. Well, we know that sort of, um, the, the, if you had to pick whether you
  431. would win the save percentage battle or sort of, I guess, save percentage above or expected
  432. based on XG. Yeah. Or the XG battle, you absolutely would rather win the save percentage battle because goaltending, uh,
  433. performance backward looking is like the biggest driver of who wins any hockey game and any
  434. series. How predictable it is, is an open question. It's traditionally been quite noisy, uh, trying to
  435. predict who would be the best goalie and which goalies would play the best going forward.
  436. And in a weird way, we saw this, I mean, the, the hurricanes, in addition to being the masters of like, Hey, over a
  437. large enough sample, if you control enough possession and have enough of the XG, uh, over,
  438. over many years, even years worth of sample, you will eventually win the cup. They are also an example of how mercurial
  439. goaltending is because they put in the, one of the worst goalies in the league from
  440. the regular season, Freddie Anderson. He was the best goalie in the playoffs through three rounds and then was not good
  441. early in the series. And they put a rookie backup in who then played the way he had been playing
  442. earlier in the playoffs and they won the cup. So it's like this weird duality of goaltending
  443. luck over layered over the long-term, Hey, let's just control all of the possession over
  444. the course of a decade. All right. So that's great. Wrap up for hockey. Why don't we spend the remaining five or
  445. so minutes giving us a quick recap of the world cup. Um, and my, um, distant attention as an
  446. American, you know, I have to confess it. Um, I did follow the Americans. They won, they won surprisingly big.
  447. They look great. And then the other data point that I paid attention to was Spain playing to a
  448. tie against a terrible team. And, and, and there was a lot of ties. And we heard that from Cade before he
  449. slipped out. Um, so what is your assessment of, at least from a statistical angle of the first
  450. rounds of games that we've seen? It seems very chaotic. It's in keeping with what we saw.
  451. It's just like the, the NBA and NHL kind of never left. And we're still, you know, continuing it with
  452. the world cup because yeah, these draws, I think what was notable about them is that
  453. they were also most of them. If not all of them were lopsided like pregame odds, where you would expect the favorite
  454. to, to get the points out of the situation. And in a lot of cases, like a team like Cabo Verde, uh, playing against Spain,
  455. a draw is a win for them. I mean, it's, they, they were not expected to win that game.
  456. It was one of the biggest upsets, even though we think of upsets typically as being
  457. like handing a loss to the favorite, this was handing them a tie and that was a huge, massive upset.
  458. And we're in the middle of one right now, as well, as we record, we don't know how it's going to, um, once you
  459. listen to this, you want, you'll know what we don't know, but France is tied with
  460. Senegal into the 65th minute of this game. And that's kind of been a continuation of
  461. yesterday, uh, as we record this. So it would have been on Monday, four draws. That was the first time since 1958, that
  462. there were four draws on a single day in the world cup. And only the second time that's ever happened.
  463. And those draws, uh, like substantially changed the odds for what were already, I think this
  464. was sort of the favorites were less favored than usual going into the tournament as well.
  465. Like, uh, Spain and France were the co -favorites at about 16% each. If you kind of normalize, uh, everything to
  466. add up to a hundred percent, normally the favorites closer to 20%, uh, rather than around
  467. that 15, 16% mark. And we just saw like Spain's odds of winning the tournament went down by two percentage
  468. points with that one draw in their opening game. Uh, Belgium, uh, was handed a draw.
  469. They dropped in terms of their odds to kind of make a deep run into the knockouts.
  470. Um, uh, Iran even lost, you know, 17 points off of their round of 32 odds because they had that draw, uh, as well.
  471. So it really shows you like for the top teams, the expectation is to get three points, not one point.
  472. But if you were conceived, if you're only getting one point against like the Cabo Verde
  473. of your group, because everybody else is going to play those teams in your group, by
  474. the way, you have to remember, even though it's much easier now to get out of the groups than it used to be, uh,
  475. sometimes in the top eight, uh, third place teams can make it out. In addition to the top two teams in
  476. each group, you still have to get the points that you're expected to get against the
  477. weaker teams in your group to be able to kind of stay on pace with everybody else.
  478. Cause the assumption is those draws are not going to continue and you're going to be
  479. kind of left out in the cold because you only got one point against Cabo Verde and everybody else got three points in their
  480. matches against them. So it does matter a lot. And we're already seeing some pretty big shifts
  481. in the odds early, like just less than a week, just a few days into the tournament.
  482. Speaker 3 I think it speaks to how just kind of stochastic this sport is and that, you know, even a top team can
  483. struggle. I mean, the other match that we haven't talked about, cause it actually had scoring in
  484. it was Australia upset Turkey to nothing. And I kind of like, I went to kind of quote, unquote, the box score for
  485. this match and Turkey had possession 63% of the time and had 28 shots to Australia's a lot of shots in 28 shots
  486. to Australia's eight and Australia one to nothing. So it's kind of like Australia have good
  487. shots. I mean, what was the XG? I mean, that's really the issue. Yeah. I, I guess, I guess that's right.
  488. But I think it just speaks to kind of like, you can kind of dominate, you know, like I watched, I watched both the
  489. Spain one against Cape Verde and, you know, they were just basically in the Cape Verde
  490. zone the entire time and couldn't convert. And I watched, you know, a day earlier,
  491. two days earlier, Germany, Curacao, basically the same game, except the, the Germany goals went in
  492. and it ended up being seven one. And so like, you know, I don't know how much it's just sort of like, you
  493. know, like if there's something systematic here, or we're just sort of seeing kind of puts
  494. a little less puts on realizations than we should for our puts on rates here. And I'll just throw in that.
  495. I think the, the public tends to overreact to this, these early pieces of information, but
  496. that's something that we can hash out later. Can you remark on the Americans? You have anything interesting to say about their
  497. performance or it's just a relevancy at this point? Well, they're, they were relevant in the sense
  498. that, amidst a tournament where all of the other favorites seem to be, except Germany, as
  499. you mentioned, who loves to win matches seven to one in their history. But against a lot of the other favorites
  500. are kind of top tier teams. Not that we even thought the Americans might be like borderline that going into the tournament,
  501. certainly not among the absolute favorites. They were the ones that probably were, were
  502. them and Germany showed out the most in terms of just being like most impressive in
  503. their win. So, you know, the hype train is going to pick up for sure. And I'm sure that there will be some,
  504. some moments of maybe coming back down to earth at different times, but they couldn't have
  505. asked for a better start. I think that was sort of also, and the way that some of the goals, they
  506. were kind of heroic goals. They were goals where one guy was kind of taking on multiple different defenders and sort
  507. of, you know, doing some pretty precision strikes to make it happen. And they didn't let their foot off the
  508. gas for one second. They literally scored in extra time in the last like 30 seconds of the match to
  509. get that fourth goal, which normally you would say is unsportsmanlike and running up the score.
  510. But I think in a sport like soccer, where we've just been talking about how hard
  511. goals are to come by, you take every goal that you can get and you make no apologies for it.
  512. And so I mean, they matter for tiebreaker and they're tie breakers. Yeah. It is consequential kind of beyond our, our
  513. little like prediction exercises. Right. Well, that was a, that's a great point to end on.
  514. I want to thank Neil for Neil Payne for joining us this hour. He'll be coming back to Philly.
  515. So we expect to see you more in person. And that's great for the show. I'll thank Shane and Cade.
  516. Eric is away doing, as Shane likes to say, Eric Bradlow things. I want to thank our producers, Jacob Grodnick,
  517. Deep Patel, Marissa Renna for doing the things that they do to help make this show
  518. run every single week. We are off next week. So in a couple of weeks, there'll be
  519. a lot more information, lots to talk about. Maybe we'll even mention the word baseball in
  520. two weeks, which would be exciting. We have heading to, we're actually in the doldrums of baseball.
  521. So maybe it was fitting that we not talk about it at all this week. But hope you guys, our listeners come back
  522. and join us in two weeks. We'll be running a rerun next week. Thanks, everyone. And enjoy your sports, enjoy your statistics, and
  523. have a good week.

Brief Summary

Neil Payne discusses the Knicks' surprising playoff success and the unpredictability of sports analytics in basketball and soccer.

Key points

  • Knicks' Historic Playoff Run. The Knicks achieved a remarkable playoff run, setting new records and surprising everyone.
  • Unexpected Championship Dynamics. Neil Payne discusses the unpredictability of NBA championships and roster construction.
  • Knicks' Surprising Championship Run. The Knicks defy expectations, raising questions about luck and analytics in sports.
  • Carolina Hurricanes' Chaotic Victory. The Hurricanes win the Stanley Cup amidst unexpected goalie changes and chaotic games.
  • Hurricanes' Goalie Switch. The switch from Anderson to Bussey improved the Hurricanes' defense significantly.
  • World Cup Draws. Four draws in a single day at the World Cup, a rare occurrence.

Tension

45/ 100Some tension

Related episodes