00:00:01
Welcome, welcome to Wharton Moneyball
00:00:04
here on the Wharton podcast network.
00:00:06
This is Eric Bradlo, professor of
00:00:07
marketing, statistics, and data science
00:00:09
here at the Wharton School. I'm here
00:00:10
today with my 11 and a half year
00:00:12
collaborators on Wharton Moneyball, but
00:00:14
of course a lot longer than that as
00:00:16
friends and colleagues. I'm here today
00:00:17
with Professor Shane Jensen, Professor
00:00:19
Audi Winer, both professors of
00:00:21
statistics and data science. Some
00:00:23
combination of the three of us and Cade
00:00:24
Massie are here every week on Wharton
00:00:27
Moneyball. And of course, this is the
00:00:29
last show of 2025 as we're sitting here
00:00:31
on December the 30th recording the show.
00:00:33
It's been a great year of interesting
00:00:36
things in statistics and data science,
00:00:38
especially applied to sports. Even more
00:00:40
so happening on telecast now, even
00:00:43
broadcasters using them more and more,
00:00:45
including more in writing. Uh something
00:00:47
I talked with Aaron Shatz about last
00:00:49
week on the show, we're starting to see
00:00:51
award winners even more reflective of
00:00:54
advanced statistics. So it's been a and
00:00:57
you know we now have thanks to Audi and
00:00:59
Wasabi we have a Wharton sports business
00:01:02
journal here at the Wharton School which
00:01:04
has supplemented the already great
00:01:05
publications that are happening at the
00:01:07
application of statistics data science
00:01:09
and sports of which I think it's fair to
00:01:11
say the two of you I wouldn't say the
00:01:13
pioneers but the two of you were in the
00:01:15
early boat legitimizing statist the
00:01:18
application and methodological
00:01:20
development of statistics and data
00:01:22
science and sports and so uh it's great
00:01:24
to be on the show with you guys for the
00:01:26
last show of uh 2025.
00:01:28
>> Yeah, I I'll take the acknowledgement.
00:01:30
Shane and I we did remember that years
00:01:32
back when Shane had first started as a
00:01:34
professor and we got money from ESPN um
00:01:37
to do research into baseball statistics.
00:01:40
That's the paper that we wrote I think
00:01:42
is still maybe my top favorite paper of
00:01:45
all time.
00:01:46
The only reason the only reason I didn't
00:01:48
call you guys the pioneers is of course
00:01:49
you know uh Shane remembers I mean we
00:01:51
were both graduate students at Harvard
00:01:53
stat uh Fred Meller had written many
00:01:56
many papers of course the original Efron
00:01:58
and Morris shrinkage estimation paper
00:02:00
uses baseball data um you know our
00:02:04
former colleague and you know friend who
00:02:06
passed away recently Dave Schmidline
00:02:08
wrote a number of papers including you
00:02:10
know can a hot goalie take you win you
00:02:13
the Stanley Cup. So the only reason I
00:02:14
was saying is I was pulling you guys in
00:02:16
the first wave or in that
00:02:19
>> we weren't the first to do sports at
00:02:21
>> No, no, no, no. And of course a mutual
00:02:23
>> your son's very old as well. So that
00:02:25
can't can't be
00:02:27
>> and also we all have a mutual connection
00:02:30
to this person. Of course, one of my
00:02:31
adviserss, one of Shane's adviserss, Hal
00:02:33
Stern, obviously did a lot of work in
00:02:35
there. Obviously, his PhD adviser was
00:02:37
Tom Cover, someone that Audi knew
00:02:39
extremely well. Tom Cover is is uh is
00:02:42
possibly one of the most influential
00:02:43
statisticians in my career. Not only did
00:02:45
I uh go to Stanford because he was a
00:02:48
professor there, but I did research
00:02:50
information theory because that's what
00:02:51
he did. And he wrote one of the classics
00:02:54
and I today we kind of we don't we
00:02:56
discount it, but he wrote one of the
00:02:58
most original and beautiful papers in
00:03:00
baseball analytics um called the
00:03:02
offensive erra. It's a classic paper
00:03:04
from 1975.
00:03:06
Hm. I I it's not a paper I've seen, but
00:03:09
I will look. The stuff of Tom's that I
00:03:10
was familiar with is because I, you
00:03:12
know, I was dabbling in lottery stuff.
00:03:13
He had done a lot of work there. Uh
00:03:16
specifically what h how do you infer
00:03:18
sell entries of a contingency table? You
00:03:20
only observe the margins, which is the
00:03:22
classic thing, you know, for the
00:03:24
lottery. They tell you how much how
00:03:26
frequently each number is used in the
00:03:28
lottery, but they don't tell you the
00:03:30
probability of the n tupils. Like they
00:03:32
will not give how many times, but how do
00:03:34
you infer something from that? So I
00:03:36
became familiar with a lot of his work
00:03:38
uh but not the offensive erra one. I I I
00:03:42
paper it's it's cover and king. We could
00:03:44
devote a um we could we can devote a
00:03:46
session to talking about it some future
00:03:47
time. We could also talk about how how
00:03:50
the oldfashioned method that they used
00:03:52
wouldn't would be um just too simplistic
00:03:55
for today's analyses. But uh listen, he
00:03:58
also, you know, Kofer was right there
00:04:00
with Shannon and Thorp inventing
00:04:03
gambling um techniques to be used in the
00:04:06
casinos. And I remember going to his
00:04:09
house as a grad student. He showed me
00:04:10
the old roulette um device that they had
00:04:14
made made to predict what fraction of
00:04:16
the wheel they expected the ball to land
00:04:18
in with excess probability. And they
00:04:20
would let they would use that to uh
00:04:23
place bets on a roulette wheel.
00:04:25
Incredible. and they just f felt it was
00:04:27
too risky. I mean, because of course it
00:04:29
was monumentally illegal to use that
00:04:31
kind of device, [laughter]
00:04:34
but but they they they were doing it.
00:04:37
And of course, Thorp um who was at MIT
00:04:39
wrote the first book called Beat the
00:04:41
Dealer on on how to beat blackjack.
00:04:43
Impenetrable book. Today, you can you
00:04:46
can ask Chat TBT to make you an expert
00:04:48
in a few hours. Um but back then, he had
00:04:51
that book and it was impossible to
00:04:53
decipher. And I remember spending quite
00:04:55
a bit of time learning how to be a a uh
00:04:58
a a a competent card counter from that
00:05:00
book. Um it kept the the pool of people
00:05:02
able to do it low because the barrier to
00:05:05
entry was so high.
00:05:07
Probably a good thing for all. Well
00:05:09
guys, we always um today we don't have a
00:05:11
guest on the show. Uh it is December the
00:05:13
30th or we could have gotten a guest. I
00:05:15
figured why not just you know you guys
00:05:17
are the guests. So, I figured I'd take
00:05:18
this opportunity to, you know, what
00:05:21
caught your eye in sports. And of
00:05:22
course, we always do it from a
00:05:23
statistics angle because we're a sports
00:05:25
and statistics show here. So, Shane, I
00:05:27
thought why not? Let's start with you
00:05:28
and we'll just alternate between you and
00:05:30
Audi and I might chirp in every now and
00:05:32
then with some stuff, but Shane, let's
00:05:34
just start with you.
00:05:35
>> Well, I mean, obviously, I guess what's
00:05:36
caught my eye. I'm absolutely over the
00:05:38
moon with Drake May and the Patriots
00:05:41
this season. And I mean, the my
00:05:44
expectations were low at the start of
00:05:46
the season. I think most people's word,
00:05:48
but to have somebody who is in the MVP
00:05:50
conversation and to be a team in the
00:05:54
conversation for the number one seed in
00:05:56
the conference, I just I I mean, it's
00:05:58
one of the many I think very unexpected
00:06:01
things uh to happen this season. Not not
00:06:04
the least of which is kind of some of
00:06:05
the sort of top contenders not being
00:06:07
anything uh this year. So, yeah. So, I
00:06:09
guess that's what's kind of caught my
00:06:10
eye. I continue to be amazed by the run
00:06:12
they're on. And maybe you get you can
00:06:14
kind of temper my excitement a little
00:06:16
bit or something. Tell me something that
00:06:18
you know makes me not absolutely gaga
00:06:20
over this Drake May situation for
00:06:21
potentially the next 10 years.
00:06:23
>> I would think something you might be
00:06:24
equally happy about and I'd love your
00:06:26
thoughts on this. Maybe just from a
00:06:28
statistical perspective or even if it's
00:06:30
I I don't know that I'm looking for a
00:06:32
precise answer, but how you think about
00:06:33
it, I think you have to be equally happy
00:06:35
that Mike Frabel's the coach
00:06:38
>> because you know, you put in an average
00:06:40
coach there and you know, it's hard to
00:06:43
know what the counterfactual is, but
00:06:46
it's not obvious that there's a 13 and3
00:06:48
team sitting there. It's not obvious.
00:06:50
You know, Mike Frabel's relatively young
00:06:52
for a coach and he's got a long way to
00:06:54
go. He could be there for all of Drake.
00:06:56
Matter of fact, if he's rational, he
00:06:58
should be there for all of Drake May's
00:07:00
career. How do you think about the role?
00:07:02
>> No. No. I mean, I I think you it speaks
00:07:04
to I think the difficult like, you know,
00:07:06
kind of the good fortune I guess the
00:07:07
Patriots perhaps have had here or just
00:07:09
the difficulty of turning around a
00:07:11
franchise that had like a top three pick
00:07:13
in the last like three or four drafts is
00:07:15
that you need to kind of, you know,
00:07:17
teams that are that bad usually need
00:07:19
both the coach and the quarterback kind
00:07:21
of like refreshed. And getting both
00:07:24
those kind of an elite level version of
00:07:26
both those at the same time is not an
00:07:28
easy feat. And of course they correlate
00:07:30
with you know kind of I I I think they
00:07:31
feed back on each other once you uh have
00:07:34
them together. But again, the I guess
00:07:36
good luck basically that the the
00:07:38
Patriots have had that I think that's
00:07:39
really facilitated their quick
00:07:41
turnaround. Whereas I think the Jets
00:07:43
might have had I think the Jets had a
00:07:45
good coach and uh you know Salah a
00:07:48
couple years ago but not the good
00:07:50
quarterback and they you know it's hard
00:07:51
lining those things up.
00:07:53
>> So let me ask you a question now. Um how
00:07:55
much do you basianly update? In other
00:07:57
words, I think we all know last year I
00:07:59
think they were three and 14 or whatever
00:08:01
they were last year. They weren't a good
00:08:03
team last year.
00:08:03
>> Yeah. Um, let's suppose this year your
00:08:06
prior might have been, well, Drake May
00:08:08
is going to improve. You know, you got
00:08:09
Mike Brael, maybe there would be a 500
00:08:11
team. Well, let's put this way. I think
00:08:12
a reasonable confidence band might have
00:08:15
been between six to 10 wins somewhere in
00:08:18
that band. Um, they've obviously
00:08:20
exceeded that, but now that they're in
00:08:23
the playoffs,
00:08:25
possibly at least the two seed, maybe
00:08:27
the one seed. I think they've locked in
00:08:29
at least the two.
00:08:30
>> Yeah.
00:08:30
>> Yeah. Well, well, I mean, I Yes. I think
00:08:34
Jacksonville can maybe catch them, but I
00:08:36
mean
00:08:36
>> Oh, you're right. Maybe. So, okay. But
00:08:37
they're one of the top.
00:08:38
>> I mean, they're hosting they're
00:08:39
certainly hosting a home their first
00:08:41
they're hosting game in the divisional
00:08:43
round at the minimum.
00:08:44
>> Um and and possibly uh skipping uh the
00:08:48
uh wild card round two. They still could
00:08:51
>> right how much
00:08:54
how much do you believe they have a
00:08:55
legitimate chance to win the Super Bowl?
00:08:59
>> I mean, it seems very wide open. I I
00:09:01
mean I I can talk myself into it. I I
00:09:04
mean and Audi probably wants to jump in.
00:09:05
I I they haven't I would love to see
00:09:08
them beat more good teams. You know,
00:09:09
they haven't did they did not play the
00:09:11
toughest schedule.
00:09:12
>> Just so you know, Aaron Shats just on
00:09:13
before just to interrupt just for one
00:09:15
second statistic. I think I tweeted out
00:09:18
or exed out on W Moneyball last week.
00:09:21
Aaron Shatz had an interesting stat
00:09:24
which was this Patriot team up till last
00:09:27
week. I assume the same might be true
00:09:29
has the third easiest schedule since
00:09:32
1978.
00:09:33
Like it's a fact. Like that's
00:09:35
>> Yeah. I mean it's certainly not going to
00:09:37
look harder after last week.
00:09:38
>> Yeah. Right. So now maybe it's the but
00:09:41
I'm Yeah. Go ahead, Audi.
00:09:43
>> I mean so this is what this is what
00:09:44
confuses I mean, all season long people
00:09:46
thinking that that we're saying that the
00:09:47
Patriots are the best the I mean, the
00:09:50
the the most overrated team g I mean,
00:09:53
given their opponents
00:09:55
>> and they just, you know, and there's
00:09:56
been and this is a short season. I mean,
00:09:59
no matter how you slice it, we've got 16
00:10:00
games, not a lot, right? And the
00:10:03
standard deviation in a football game is
00:10:05
about 14 points. So, what how truly good
00:10:10
are they? And and and that would be the
00:10:12
my question.
00:10:12
>> I think that is the primary question. I
00:10:14
I would just say that that all all the
00:10:16
AFC teams I think it's just particularly
00:10:18
wide open the year that all the AFC
00:10:20
teams kind of have also questions. I
00:10:23
mean, Denver has had a tougher schedule,
00:10:25
but has been barely ekking out their
00:10:27
wins and, you know, I mean, if you had
00:10:29
to talk I mean, they would probably be
00:10:31
my pick to go to the Super Bowl, but
00:10:33
like but you know, I could certainly
00:10:35
argue that the Patriots could beat them
00:10:36
in a game, you know, and and and
00:10:38
similarly, you know, the I mean, the the
00:10:41
Jaguars, I mean, think about all the
00:10:43
top, you know, the teams that are going
00:10:44
to be playoff teams. I think you could
00:10:45
argue certainly um the pay I bet you the
00:10:48
Bills will be something like the sixth
00:10:50
or seventh seed and will probably be
00:10:52
favored in like every
00:10:54
>> Let me just let me just bring let me
00:10:55
just bring that up by the way. It's
00:10:57
actually something very interesting. So
00:11:00
um this is something that's fascinating.
00:11:01
I did I did just use chat GPT for this
00:11:04
but it's not that anybody could do this.
00:11:06
It's not that impressive a thing. So
00:11:09
right now the betting favorite
00:11:11
for the Super Bowl, the betting favorite
00:11:13
to win the Super Bowl are the Rams plus
00:11:17
475. Now they lost yesterday to the
00:11:20
Falcons.
00:11:21
>> They're going to be like the five or six
00:11:23
seed, right?
00:11:24
>> They're likely the six. They could be
00:11:26
the five. So I just want to say this
00:11:28
again to everybody and I want to add a
00:11:30
couple stats from Aaron Shats and then
00:11:31
Audi get your perspective on this. Right
00:11:34
now, they're the betting favorite
00:11:37
as potentially the six seed facing the
00:11:40
world champion Philadelphia Eagles on
00:11:42
the road. Okay.
00:11:46
According to Aaron, according to his
00:11:48
DVOA metric, which he invented, you
00:11:51
know, it's something that pretty well
00:11:52
known in the analytics literature in
00:11:54
football, the Seattle Seahawks
00:11:58
and the um Rams, according to sorry, the
00:12:02
Seahawks and the Rams, according to him,
00:12:05
by the way, are two of the strongest
00:12:07
teams like in the top 10
00:12:10
ever by his metric. So do you find it
00:12:12
odd and I also this is something I did
00:12:14
do by the way I did ask chat GPT took it
00:12:18
only about 15 seconds to do I told it to
00:12:22
uh provide me a plot on the x-axis of
00:12:25
the betting line implied probability so
00:12:28
on the x-axis is you take the betting
00:12:31
line you get an implied probability of
00:12:33
winning the super bowl on the y ais is a
00:12:37
strengthbased probability of winning the
00:12:39
super bowl now we can define what that
00:12:41
is it ended up using some ver some
00:12:44
amalgam of ELO and other based ratings
00:12:46
from profocus NFL.com etc. I asked to do
00:12:51
a bariat scatter plot for me and I also
00:12:54
asked it maybe to Shane's point tell me
00:12:57
the and so then asked to put a 45 degree
00:12:59
line on the plot and then tell me which
00:13:01
team is farthest from the 45 degree
00:13:04
line. It turns out, by the way, it's got
00:13:06
the Bills being the farthest, meaning it
00:13:09
has the Bills to Shane's point
00:13:13
stronger than the implied betting odds
00:13:16
by the largest gap of any team. So, Aie,
00:13:20
any
00:13:22
>> hold on just just sure I'm correct. So,
00:13:24
I understand very clearly what the
00:13:26
implied odds of winning the Super Bowl
00:13:27
are. That's easy. Um you're using now a
00:13:30
probability of winning the Super Bowl
00:13:32
using using essentially a statistical
00:13:34
model. Is that
00:13:36
>> so what it did is I'll tell you what I
00:13:38
asked it
00:13:38
>> and then I'll aren't the probabilities
00:13:41
pretty small.
00:13:41
>> They are small. I'll tell you what it
00:13:43
did.
00:13:44
>> So I just this is literally I always
00:13:46
think it's important when you tell
00:13:47
someone you used a large language model.
00:13:49
What exactly was your prompt?
00:13:51
>> Here's exactly the only thing I asked
00:13:53
it. I'll say verbatim to our listeners
00:13:55
here on Morton Moneyball. I mean,
00:13:57
tomorrow will be a different answer
00:13:58
anyway, even if it is.
00:13:59
>> Yeah, of course. But here's what I said.
00:14:01
I said, provide me a plot of betting
00:14:04
line implied probability on the x-axis
00:14:08
versus strength probability on the y ais
00:14:11
for each team to win the Super Bowl. And
00:14:14
so, and that's what it did. And I know
00:14:18
what it did because it now that chat GPT
00:14:22
5.2 into thinking mode exists. It's
00:14:25
telling me what it's doing at each step.
00:14:28
So, I'm watching it. I'm not just
00:14:29
getting the output of this. I'm watching
00:14:31
it. It's creating a simulation. It
00:14:34
simulated 10,000 playoff scenarios going
00:14:38
forward and simulated those out based on
00:14:41
strength parameters from an amalgam of
00:14:43
NFL.com, Profoot Focus, and ESPN. That's
00:14:47
how it got the strengthbased
00:14:48
probabilities. And the other ones it got
00:14:51
from I think it was BM MGM and some
00:14:53
other and some other betting line.
00:14:55
That's that's what it did.
00:14:58
>> No. So did you adjust for the vig
00:15:00
directly or did you not bother with
00:15:01
that?
00:15:02
>> Good question. It automatically now
00:15:05
adjusts for the vig when it it I didn't
00:15:08
tell it. I just told all our listeners
00:15:10
exactly what I typed in. It in its
00:15:12
thinking adjusts for the vig when
00:15:15
computing those Super Bowl-based
00:15:17
probabilities. In other words, it know
00:15:19
it it adds up to more than one because
00:15:21
of the vig. It renormalizes it because
00:15:23
of the vig. It does that automatically
00:15:26
now when you ask it to give you an
00:15:28
implied probability.
00:15:29
>> I assume the point is that the bills are
00:15:32
are are the bills the one that are the
00:15:34
most diff.
00:15:34
>> Yeah, they're most likely to win.
00:15:36
They're the biggest they're much more
00:15:38
likely to win according to the
00:15:39
statistical model than the implied.
00:15:42
>> That is correct.
00:15:44
They're the most under underbet team if
00:15:48
you want to call it that. They should be
00:15:50
they should have a better betting odds
00:15:52
than given the strength model uh
00:15:55
predicts. Uh they have the largest
00:15:56
deviation from the 45 degree line. Now
00:16:00
now the betting odds have to do with um
00:16:02
how much action there is. So on some
00:16:04
level you can say it's because they're
00:16:06
being underb, right? Not enough people
00:16:08
are are taking them. Um or it could be
00:16:10
that the statistical models have see
00:16:13
things that the public just doesn't
00:16:14
agree with.
00:16:15
>> I'm kind of fascinated on the other side
00:16:17
here that the Rams
00:16:19
um are the you said where the betting
00:16:22
market comes.
00:16:23
>> They are there. This is from from the
00:16:24
from ESPN and FanDuel. The Rams are the
00:16:28
number they're both at they're plus call
00:16:31
it 475.
00:16:33
>> Because you know the number uh the two
00:16:34
other teams in their division are
00:16:36
playing off for the number one seed.
00:16:38
Well, that's that's why I brought up
00:16:40
this upcoming weekend. So, I think
00:16:41
that's why I'm speaking to you guys.
00:16:44
>> I I mean, I I I don't know if again,
00:16:46
this is maybe uh
00:16:48
>> that is interesting. I don't know if
00:16:49
there's some kind of built-in
00:16:51
>> um you know, Stafford effect or
00:16:53
something like that, though. He
00:16:54
certainly looked uh less than impressive
00:16:56
uh last night.
00:16:57
>> Well, let me ask you guys a question.
00:16:59
How much value do you guys place?
00:17:02
Suppose I say the following. Um, I'm
00:17:05
going to keep the teams the same because
00:17:06
obviously if you take Stafford off the
00:17:08
team and you put some average
00:17:09
quarterback on it, obviously the Rams
00:17:11
are very different. How much value do
00:17:13
you put on, you know, let's look at it.
00:17:16
Rams, Shawn McVey has won the Super
00:17:19
Bowl. Denver Broncos, Shawn Payton has
00:17:23
won the Super Bowl. Mike Frael, I know,
00:17:25
hasn't won the Super Bowl as a coach.
00:17:27
He's won it many times as a player.
00:17:28
>> Tommy Mc has also lost the Super Bowl.
00:17:30
>> He's also lost the Super Bowl. Yeah, but
00:17:32
he's been he's been to the Super Bowl
00:17:34
and he certainly has his teams have
00:17:36
performed well in the Super Bowl. Do you
00:17:38
guys put any value on the coach, let's
00:17:42
call it let's even winning the Super
00:17:44
Bowl cuz maybe that's a that's too
00:17:46
specific a metric. How about playoff
00:17:49
success? Like suppose we did the
00:17:50
followup the po I'll give you another
00:17:52
coach that's won the Super Bowl. How
00:17:54
much would you change the Rams odds or
00:17:57
the Broncos odds if I told you Mike
00:17:59
Tomlin was the coach of their team?
00:18:02
Well, I I mean,
00:18:04
>> he's won the Super Bowl. I
00:18:05
>> I mean, obviously, I think playoff
00:18:07
success does move the needle a bit,
00:18:09
especially coaches that have
00:18:10
demonstrated playoff success. At the
00:18:12
same time, you know, I think the McVey,
00:18:15
it's based much more on what we know
00:18:17
McVeyy's body of work in general than
00:18:19
his particular playoff success. For
00:18:21
example, Nick Serriani, I bet you has a
00:18:23
better playoff record than Shawn McVey
00:18:26
or than probably anybody we're talking
00:18:28
about right now. consider him based on
00:18:31
that like probably the best playoff
00:18:32
record of any playoff coach out there
00:18:36
right now. Would you consider him the
00:18:37
best head coach kind of among the
00:18:39
playoff teams? Obviously not or at least
00:18:41
I wouldn't either. So
00:18:42
>> no I would not.
00:18:43
>> So I I kind of think it's you know
00:18:45
there's a lot of small sample randomness
00:18:46
to kind of play off record. Shawn McVey
00:18:48
has been around for long enough where I
00:18:51
think yeah I think his body of work both
00:18:53
regular season and playoffs clearly
00:18:55
suggest he's elite coach. Let me just
00:18:57
stick with the Patriots for a second.
00:18:58
Then Audi, I want to get to you about
00:18:59
what caught your eye in sports. There's
00:19:02
an going to be an interesting debate
00:19:03
about the MVP of the league this year. I
00:19:06
think it's down to one of two players. I
00:19:07
think everyone agrees it's either going
00:19:08
to be Drake May or Matthew Stafford. I
00:19:11
think that's likely to be true. Okay.
00:19:13
What's interesting, back to the
00:19:15
analytics versus not
00:19:18
passing yards, Matthew Stafford.
00:19:22
Touchdown, Matthew Stafford. touchdown
00:19:24
interception ratio Matthew Stafford. But
00:19:27
if you look at any let's call it more
00:19:29
advanced metric
00:19:31
>> any rate stat
00:19:32
>> Drake may
00:19:34
>> so I think it's going to be fascinating
00:19:36
and by the way not that there's let's
00:19:38
call let's we all look at there's a
00:19:40
recency bias I don't think particularly
00:19:42
Matthew Stafford paid particularly well
00:19:44
in the game last night against Atlanta
00:19:46
and I would say I don't know um I'll
00:19:49
just put what Shane put in the rundown
00:19:52
May was the first player in NFL history
00:19:53
I understand it's you know they p cherry
00:19:56
picked these He threw for five plus
00:19:58
touchdowns, 250 plus yards, and
00:20:00
completed 90% of his passes in the game.
00:20:03
Not bad.
00:20:04
>> Yeah, he was out of the game by the
00:20:06
middle of the third quarter.
00:20:07
>> Right. So, I all I'm saying is I think
00:20:12
this will be another piece of evidence
00:20:15
because based if if I just took the
00:20:17
names off and I just provide you the
00:20:19
advanced statistics, I think Drake May
00:20:22
would win the MVP this year. I do. I
00:20:24
personally, it's not about deserve it. I
00:20:27
actually do not think he will win
00:20:29
because I think Stafford leads in the
00:20:31
traditional metrics, especially
00:20:33
touchdowns by so many. I think it'll
00:20:36
just overwhelm it. But that's just my
00:20:37
opinion.
00:20:38
>> Yeah, you're probably right. Um I I I
00:20:41
think it's kind of similar to, you know,
00:20:42
kind maybe like the Mike Trout uh um
00:20:45
Miguel Cabrera kind of, you know, MVP
00:20:48
race from a few years ago where Miguel
00:20:50
Cabera, you know, Mike Trout kind of had
00:20:52
the sort of rate stat like was was was
00:20:55
dominant but Miguel Cabrera was, you
00:20:57
know, had had these kind of gotty kind
00:20:59
of totals.
00:21:00
>> Absolutely.
00:21:01
>> So Audi, what caught your eye? Why don't
00:21:02
we move on to you? I mean, not that by
00:21:04
the way, this is the nice thing. We've
00:21:05
talked about we talked about the
00:21:06
Patriots in some sense, but we got to
00:21:08
talk a little bit about coaching. We got
00:21:10
to talk about, I'll call it advanced
00:21:12
stats for the MVP versus kind of more
00:21:14
traditional stats. We got to talk about
00:21:15
the Rams and maybe they're the um you
00:21:18
know, they're the third, you just
00:21:20
pointed out the third best team in their
00:21:22
division and they're the favorite to win
00:21:24
the Super Bowl according to the betting
00:21:26
odds. They're the third. Forget again,
00:21:28
you might say, well, strength is they're
00:21:30
the third best in their division. They
00:21:33
play other teams in their division the
00:21:35
same as Seattle and the 49ers do.
00:21:38
They're in the same division.
00:21:41
>> All right. So, a lot of things have
00:21:42
caught my eye. I I I'll start at the
00:21:44
bottom. Um the Jets, that's my my
00:21:48
defunct my def de facto favorite team or
00:21:51
historically the favorite team. Um how
00:21:54
awful they've been and is just sort of
00:21:56
disappointing and also considering I I
00:21:59
been spending time thinking about Sam
00:22:00
Darnold. If you're thinking if Aaron
00:22:02
Shat says the Seahawks are this, you
00:22:04
know, one of a great team, how could how
00:22:08
does Sam Darnold go being so bad with
00:22:10
the Jets to being I mean, I don't I
00:22:12
don't think he's particularly
00:22:13
extraordinary with the with with
00:22:15
Seattle. I mean, he's had a very good
00:22:17
season, but and so it makes me wonder
00:22:20
like how how good are we really are at
00:22:22
assessing the quality of these players?
00:22:24
And I think that reflects back on our
00:22:26
discussion about Drake May. I mean, what
00:22:28
do we really know? It does. It this
00:22:30
season does seem to be like ex with the
00:22:33
exception of some very awful teams
00:22:34
including the Jets, the Giants, the
00:22:36
Raiders. Um, and if you look at it,
00:22:39
there's there's at least three or four
00:22:41
horrible teams. There's a a whole bunch
00:22:44
of teams top 10 that are not necessarily
00:22:47
that distinguishable from each other.
00:22:49
Um, and then I mean I watched the the
00:22:51
Eagles game um on uh Sunday and they're
00:22:54
not a good team. I mean they're they're
00:22:57
not a good team. Um
00:22:59
>> they have a defense. They have a good
00:23:00
defense. Can
00:23:01
>> they have a good defense and by and
00:23:02
defense just just to talk about as
00:23:04
general defense is the least predictable
00:23:07
aspect of a football team. And I mean
00:23:09
that not only from next season
00:23:12
>> to next season that doesn't predict
00:23:14
well. But even next game it doesn't
00:23:16
predict that well.
00:23:18
>> Um so what do we really know? And so I
00:23:21
I'll turn it to you. Um uh is the Jets
00:23:25
are the Jets problems so managerial
00:23:29
coachwise ownership? What do you see the
00:23:31
problem there? I mean I mentioned them
00:23:33
but Jets pay people just I mean they put
00:23:35
their hand on their back and they go h
00:23:38
it's really a weird feeling. I mean what
00:23:40
is it?
00:23:41
>> I so let me first say it's it's I don't
00:23:44
I don't buy particularly any of that
00:23:47
franchise stuff like franchises matter,
00:23:49
ownership matters. um them having first
00:23:52
class facilities matter, but obviously
00:23:54
the Jets have a great market, infinite
00:23:57
wealth, um all of that. I think you have
00:24:00
to I go back to what Shane said earlier.
00:24:02
The Jets sooner than later have to
00:24:05
decide whether Aaron Glenn is the coach
00:24:09
that can build and turn around that
00:24:11
team. Because if he's not, what's going
00:24:14
to happen is you're going to draft
00:24:17
probably, you know, uh, another
00:24:19
quarterback, right?
00:24:21
>> Yep. Yep. You are. And now the question
00:24:24
is, if he can't develop that
00:24:26
quarterback, maybe it's the
00:24:28
quarterback's no good or maybe Aaron
00:24:30
Glenn's no good as a coach or who has
00:24:32
offensive coordinator is no good. Shane,
00:24:34
I know you wanted to jump in, please.
00:24:35
Yeah, I mean I would just sort of I
00:24:37
think what the Jets have exhibited over
00:24:40
the last like like part of their
00:24:42
continual mediocrity and I think it's
00:24:44
kind of partly the age we live in I
00:24:47
suppose is there there's no patience,
00:24:49
you know, like like Aaron Glenn's got
00:24:51
like a year or two before, you know,
00:24:54
he's going to get canned and he doesn't
00:24:56
look I mean I mean, you know, I'm not
00:24:58
really I don't think he necessarily is a
00:25:00
keeper, but you know, part of the
00:25:01
reason, you know, the Darnolds, you
00:25:03
know, and and the Gino Smiths and
00:25:05
everything of the world is that you've
00:25:06
got like a year or two to really hit big
00:25:09
or you've got essentially a year to hit
00:25:10
big in New York in the New York market
00:25:12
or or you know all of a sudden the
00:25:13
pressure is on you starting to be
00:25:15
considered a bus. I kind of like look at
00:25:17
some of some of the like like Pton
00:25:18
Manning's first year. Look at look at
00:25:21
like or or Rogers sitting for a few
00:25:23
years. Like there's like a lot of I I I
00:25:25
think we
00:25:25
>> Tom Brady's first year
00:25:28
>> that's not I I I mean sure if you want
00:25:31
that for the standard, but like I I
00:25:33
think people like need to be more
00:25:35
patient with quarterbacks and recognize
00:25:37
that they'll often struggle because
00:25:39
they're usually on a bad team um coming
00:25:41
out of the draft and and you kind of
00:25:44
have to build a whole organization
00:25:46
around them. And I think, you know, the
00:25:48
Jets keep kind of hoping for a quick fix
00:25:51
or like, you know, that a number one
00:25:52
will come in without support and like,
00:25:54
you know, just be awesome out of the
00:25:55
gate. And sometimes that does happen.
00:25:57
You know, there there have been, you
00:25:59
know, teams that have kind of hit very
00:26:01
quickly like that. But I I think
00:26:03
>> I think there doesn't seem to be a lot
00:26:05
of long-term vision there.
00:26:06
>> Well, I'm going to say I'm gonna let me
00:26:08
say one thing about Sam Darnold, then I
00:26:09
want to ask a question. So, let me start
00:26:11
with Sam Darnold. So, I just saw
00:26:12
something a stat the other day. There
00:26:14
are only six quarterbacks in the history
00:26:17
of the NFL
00:26:19
to win 13 games two seasons in a row.
00:26:23
Okay, Jim Darnold is now one of them.
00:26:26
Now,
00:26:28
great. Maybe it's teams around him.
00:26:30
Maybe they're be he's being coached
00:26:32
well. Now, of course, Brady's on that
00:26:33
list five times, but let's I'm just I'm
00:26:36
counting Brady just once. I'm not
00:26:37
counting Brady five times. But
00:26:40
Sam Darnold
00:26:42
is one of six quarterbacks on two
00:26:45
different teams now. By the way, he may
00:26:48
be the only to do it on two different
00:26:50
teams
00:26:51
to win 13 games in a row twice. I mean,
00:26:54
it's incredible. I mean, at some point,
00:26:56
you have to give something to Sam
00:26:58
Darnold. Great. Maybe he's a system
00:27:00
quarterback. Great. He's a system
00:27:02
quarterback that maybe win 14 games and
00:27:04
be the number one seed in the NFC.
00:27:06
Just make sure you didn't say he's
00:27:07
hasn't won 13 in a row two seasons in a
00:27:10
row twice.
00:27:11
>> He's won 13 in a row.
00:27:13
>> But I'm saying we're two different
00:27:14
teams. Yeah, he's done it twice last
00:27:16
year and this year he's still got a game
00:27:18
to go. But I'm saying he was the Vikings
00:27:21
quarterback last year and now he's
00:27:22
>> Yeah.
00:27:23
>> And that was my point. I mean, he's not
00:27:25
just good. He's very good yet we all
00:27:27
thought of him as terrible with the
00:27:29
Jets.
00:27:29
>> That's my point. Really?
00:27:30
>> Maybe I'll give you another name you
00:27:32
speak to. Tell us again the whiner
00:27:36
rating system of Trevor Lawrence. Maybe
00:27:39
it takes a little bit of time and how's
00:27:41
Trevor Lawrence playing? How would you
00:27:42
feel? By the way, if you're the Jaguars
00:27:45
right now, I know there's there's
00:27:47
probably half a dozen teams there if I
00:27:49
know the Patriots would be one of them.
00:27:51
If you're the Jaguars, you wouldn't
00:27:53
trade Trevor Lawrence for anybody right
00:27:54
now.
00:27:55
>> No, like not remotely. But my question
00:27:57
is this. Let's maybe But there's a
00:28:01
handful of teams that
00:28:03
Yeah. No, no, no. I'm saying there's a
00:28:05
handful of teams that would make that
00:28:07
same statement. The Chargers might make
00:28:09
that statement. The Patriots would
00:28:10
definitely make that statement, but the
00:28:12
Jaguars are now one of those teams that
00:28:14
would make that
00:28:15
>> like top 10 quarterback. Finally.
00:28:17
>> Oh, yeah. Let me let me ask you guys a
00:28:19
statistical question. Yeah.
00:28:20
>> Know, one of the things that we brought
00:28:22
often in baseball, we have these ways of
00:28:24
measuring the quality of a player
00:28:26
outside of their quality of opponent and
00:28:29
quality of the team around them. for
00:28:30
baseball, it's easy because the their
00:28:32
their players around them don't matter
00:28:33
that much. Opponents matter um and we
00:28:36
and we and we can adjust for that and we
00:28:38
can generally measure the quality of a
00:28:39
player in the abs in the in the outside
00:28:42
of their context.
00:28:44
>> Have we gotten any qu ability to do that
00:28:46
for quarterbacks? Is there any way to
00:28:49
really question
00:28:51
measures?
00:28:52
>> Well, the problem with EPA, and I'm I'll
00:28:54
bring this up. problem with EPA is
00:28:57
you're if you do you really care about
00:29:00
EPA or do you really care about
00:29:02
>> I mean are you ask
00:29:05
yes I mean as as a peripheral as a
00:29:08
ratebased peripheral it's kind of like
00:29:10
do I really as a pitcher do I really
00:29:11
care about strikeout rate I mean yeah
00:29:15
>> but that is kind of like pretty what I'm
00:29:18
trying to do
00:29:19
>> let me try to answer your question
00:29:20
because I looked at this so if you also
00:29:23
look in the rundown down. Although I had
00:29:25
a mistake. I I I don't know. I was
00:29:26
dreaming the Bucks were 6-1 at one
00:29:28
point. They were 6-2. Whatever this Let
00:29:30
me type in again what I asked chat GPT.
00:29:35
How has Baker Mayfield's quality of play
00:29:38
declined from when the Buccaneers were I
00:29:40
should have said 6-2. 6-2 until now?
00:29:44
Please be specific on what metrics you
00:29:46
are scoring him on. Is there any way to
00:29:49
know whether it is in quotes his fault
00:29:52
or whether it is due to poor coaching or
00:29:54
a poor offensive line? So, I only copied
00:29:57
the first half of what it showed, but
00:29:59
Audi, it gave me a bunch of metrics
00:30:03
that it thinks are predictive of
00:30:05
quality. His yards per attempt has gone
00:30:07
down. His passer rating's gone down. His
00:30:09
completion percentage has gone down. His
00:30:11
interception rate has quadrupled. His
00:30:14
touchdown rate has been flat. his sack
00:30:16
rate has gone up. And so it also had a
00:30:19
bunch of other more advanced stats that
00:30:22
I I probably should have put there. But
00:30:25
it
00:30:25
>> it probably doesn't have it probably
00:30:27
doesn't even have access to PFF, but his
00:30:29
PFF kind of grades per game have been
00:30:31
going down.
00:30:32
>> It had it had PFF grades, but it did. By
00:30:34
the way, this is beauty. By the way,
00:30:36
look, just because I'm Morton's vice
00:30:38
dean of AI doesn't mean I'm all in. But
00:30:39
let me just say I'm sort of all in.
00:30:40
>> No. Yeah. Yeah. You understand? For
00:30:43
those of you that have access to it,
00:30:45
>> use chat GPT 5.2 thinking mode. And and
00:30:49
let me just say why. It's I'm not saying
00:30:51
its answers are perfect, but I'm saying
00:30:54
you get to watch in real time its
00:30:57
thought process. It's literally telling
00:30:59
you step by step. So even said Shane, it
00:31:02
it's almost said in like you and I were
00:31:03
speaking. It said, "I'd love to have PFF
00:31:07
data, but that's behind a firewall."
00:31:10
>> It said that. Okay. Yeah. So I I did
00:31:13
some play. Let's get back to the So the
00:31:16
metrics. The reason I brought it up is
00:31:17
those are metrics.
00:31:18
>> Yeah. No, the thing is so but let's go
00:31:20
back to my question. Those metrics would
00:31:22
say that Sam Darnold was bad when he was
00:31:25
with the Jets and now he's good with the
00:31:28
two teams that he's been very good with
00:31:29
the Vikings and Seattle. What I'm asking
00:31:32
for is it because Sam Darnold is now
00:31:34
better
00:31:36
or is it because we just weren't
00:31:37
evaluating him properly? Well, I mean,
00:31:39
clearly for now,
00:31:41
>> I think you can conclusively say and and
00:31:44
he also was not good with Carolina, if I
00:31:46
remember correctly. I mean, he's been
00:31:47
around.
00:31:47
>> He was not
00:31:48
>> um
00:31:49
>> the Jets did not put him in his in the
00:31:51
best position to win. That that's a
00:31:54
pretty safe statement. I mean, the
00:31:55
ceiling on Sam Darnold has been
00:31:57
demonstrated to be much higher than the
00:31:59
Jets could even, you know, muster. So,
00:32:02
that I I mean, there's obviously an
00:32:03
organizational effect here, I think.
00:32:06
>> Well, let me ask you a question. We've
00:32:07
seen it happen not just to Sam Darnold,
00:32:09
Gino Smith, you know, I mean,
00:32:12
>> let me ask you guys both a question.
00:32:14
When evaluating Sam Darnold or any such
00:32:17
quarterback, will you allow me to
00:32:19
compute a set of metrics, but I'm in a
00:32:23
condition on various things? For
00:32:26
example, I might condition on a bin of
00:32:29
win loss records, or I might condition
00:32:32
on the team making the playoffs. And so
00:32:35
how does his metric how do his metrics
00:32:39
look compared to other quarterbacks that
00:32:43
have played on similarly successful
00:32:45
teams? And I won't even use postseason I
00:32:48
won't use postseason outcomes. I'll use
00:32:50
let's take all teams that have won 10
00:32:52
plus games. Look at the metrics of their
00:32:54
quarterback. look at in a multivariate
00:32:57
way where Sam Darnold lies on those
00:32:59
multivariat distributions and try to say
00:33:03
is Sam Darnold good compared to
00:33:06
quarterbacks that have been on
00:33:08
successful regular season teams. Is that
00:33:10
a fair to condition on that and do it
00:33:11
that way?
00:33:12
>> I mean yes and one one piece of
00:33:13
information we that have already
00:33:15
conditioned is you know that previous
00:33:17
MVP discussion we had even though he is
00:33:19
on a 13- win team did Sam Darnold come
00:33:22
up at all?
00:33:24
So we kind of implicitly know his
00:33:27
metrics are perhaps, you know, not what
00:33:31
we at least see as driving the success
00:33:33
of his team.
00:33:36
>> You have more do you have another way of
00:33:37
so I mean my my here's what I would love
00:33:39
if I if I had if I were the, you know,
00:33:42
the the uh the oracle or the you know
00:33:45
the deciser who can decide what I can
00:33:47
observe. I would love to observe to know
00:33:49
how much pressure there is on a
00:33:50
quarterback
00:33:52
>> on every play and and I want it. Well,
00:33:54
can you um
00:33:55
>> You can you can get it publicly on on
00:33:57
every play. How much?
00:33:59
>> Yeah, you can get some good proxy
00:34:02
pocket last. I Yeah, I should have
00:34:04
printed this out when I asked the
00:34:06
question about Baker Mayfield. The
00:34:08
advanced stat it had was pocket pressure
00:34:11
rate, how he did against pocket
00:34:13
pressure. Those were stats that it was
00:34:15
found. No. So those are aggregate
00:34:17
information. What I would like to do is
00:34:18
on every play, every drop back, every I
00:34:21
want to know how much pressure and I
00:34:22
want to know how open the receivers were
00:34:24
and and and so what I'm trying to do is
00:34:26
get something that's like like a what we
00:34:28
would call in baseball peripheral. Like
00:34:30
how good at you are doing your your
00:34:32
basic job, which is open. Exactly. PF,
00:34:37
right?
00:34:37
>> No, PFFs does that. Yes, they do that
00:34:39
with with ratings by by raiders. They
00:34:42
like look at it and they Yeah, they they
00:34:44
do that. You can get a B you can't the
00:34:46
peripheral you can get is a B kind of a
00:34:49
a binary count variable but playbyplay
00:34:52
did they do a good job or not as graded.
00:34:54
Yeah.
00:34:54
>> But you would agree that we're not far
00:34:56
away from a
00:35:00
video enabled large language model
00:35:03
trained by humans to be able to ingest
00:35:07
film data. Determine some definition of
00:35:10
pocket pressure. determine some
00:35:13
definition of how open the receivers
00:35:14
are. Determine some definition of
00:35:17
expected yards gained given this
00:35:19
situation. I mean that data will become
00:35:22
like even if it's not PFF doing it. Um
00:35:26
somebody could
00:35:28
that AI that is what AI is
00:35:32
extraordinary at. It's trained to do
00:35:33
that some blackbox version of that soon.
00:35:37
>> Yeah. Doing that's what they're doing in
00:35:38
soccer. are doing it in and continuous.
00:35:41
>> Nobody will be telling it to anybody.
00:35:45
>> Yep.
00:35:46
>> Well, guys,
00:35:47
>> all right. But before I let you go, guy,
00:35:49
before you go on to your your I just
00:35:50
want to say just
00:35:52
>> I know this is football football, but we
00:35:54
are at the baseball solstice today at
00:35:57
this very moment. We're exactly halfway
00:35:59
between the last pitch of the World
00:36:01
Series and the first pitch of spring
00:36:02
training. It just needed to be noted.
00:36:04
>> Like it's been a particularly cold
00:36:06
stove. No,
00:36:07
>> it has been the news.
00:36:10
>> Well, guys, since we are I just one
00:36:12
other quick thing. This is not this is
00:36:14
not genius math. I just wanted to see if
00:36:16
it could do what we teach in our basic
00:36:18
stat classes. We obviously have might as
00:36:20
well we should spend a few minutes on
00:36:21
it. We have four NCA games coming up. Uh
00:36:25
NCA football games coming up. Um Oregon
00:36:28
a 2 and a half point favorite over Texas
00:36:30
Tech even though Texas Tech was the four
00:36:32
seed. That's fine. Indiana by seven over
00:36:35
Alabama. We could question, you know, if
00:36:37
we look historically, there's no reason.
00:36:39
I mean, this is another thing like how
00:36:41
much weight do you put on Alabama's
00:36:43
historical success? Should they be less
00:36:45
than a seven-point underdog? We have
00:36:48
Georgia 6 and a half point favorite on
00:36:50
Miss. We have Ohio State 9 and a half
00:36:52
point favorite on Miami. I just asked
00:36:54
Chad GPT a very simple question, which
00:36:57
is what's the probability that at least
00:36:59
one of the underdogs advances? And of
00:37:02
course it can do it can do what I call
00:37:05
you know high school level math stat.
00:37:07
The probability of least at least one is
00:37:09
one minus the probability of none. So I
00:37:12
I put this in the rundown. It actually
00:37:14
computed that. It assumed independence.
00:37:16
It took the betting odds. It implied it
00:37:18
implied win probability. It can I mean
00:37:21
it's not shocking that it can do that
00:37:23
calculation. But I'm just saying you
00:37:26
know it's not like I gave it any real I
00:37:29
didn't tell it where to get the data
00:37:30
from. I didn't tell it to do, you know,
00:37:32
it could have done it in a more brute
00:37:33
force way. All I'm commenting on is that
00:37:37
apparently there's an 81% chance, which
00:37:40
again, if you asked most pe This is why
00:37:42
we need a show like ours in my view. If
00:37:43
you asked most people on the street,
00:37:46
what's the probability that
00:37:49
Oregon, Indiana, Georgia, and OSU are
00:37:52
all going to win? They're going to say a
00:37:56
lot more than 19%.
00:37:59
A lot more
00:38:01
a lot more than 90.
00:38:03
>> You think so? I mean, aren't aren't
00:38:04
don't most people think of the games is
00:38:06
pretty tight?
00:38:07
>> No.
00:38:08
>> No.
00:38:08
>> See, they're one tight game, which is
00:38:11
Oregon, Texas, Tech, and and maybe, by
00:38:13
the way, a lot of people have Oregon as
00:38:15
their prediction to win the whole thing.
00:38:17
Um, and then I, you know, the betting
00:38:19
odds don't have Indiana, Georgia, or
00:38:21
Ohio State as particularly close games.
00:38:24
I don't think anybody would be
00:38:25
surprised. As a matter of fact, if you
00:38:27
remove the Oregon game out of there and
00:38:29
I said, "What's the probability one of
00:38:30
Alabama, Old Miss, or Miami advances or
00:38:34
at least one?" I think that probability
00:38:36
would probably be no more than
00:38:41
one minus.8, so I don't know, uh, maybe
00:38:44
10%.
00:38:46
Yeah, I just felt like most people
00:38:48
thought there of there not being an
00:38:50
overwhelmingly dominant team in uh,
00:38:53
college football this year. I think I
00:38:55
think I remember Kate saying like OSU
00:38:56
and Indiana are significantly above
00:38:59
everybody else this year. We could
00:39:01
debate whether Georgia, you know,
00:39:03
Georgia will be there or not at the end,
00:39:05
but either way, I just thought it was
00:39:06
interesting. I I just thought that it's
00:39:08
not that the calculation was very
00:39:09
sophisticated or getting the data was
00:39:11
that hard or sophisticated. I just
00:39:13
thought it surprising that there's an
00:39:14
81% probability assuming independence,
00:39:17
which by the way, I'm not an
00:39:19
independence guy. I look there's no
00:39:21
reason the game should be dependent on
00:39:23
each other. No reason whatsoever except
00:39:28
>> there's none. Except I can tell a story
00:39:31
where if the first game happens and
00:39:34
let's say you know well it only takes
00:39:37
one of these events to happen for my
00:39:39
event to happen that I asked for. But
00:39:41
let's say Miami plays a really close
00:39:43
game against OSU. Does that give any
00:39:47
increased belief for whatever that's
00:39:49
worth to any of the other underdog
00:39:52
teams? Any rational
00:39:53
>> answer that can I go first? Yeah,
00:39:55
>> I would say yes. It means to suggest you
00:39:57
have model miscalibration. Um, and
00:40:00
remember the same model is applied to
00:40:02
always. Will it
00:40:03
>> exogenous latent variable like it's you
00:40:06
know what
00:40:07
>> crazy monsoon rains throughout of the
00:40:10
United States that don't that basically
00:40:13
you know knowing one game 63 tells you a
00:40:15
lot about the Nets game being 6-3 or
00:40:18
something like that there I mean that
00:40:19
that's an example of something that
00:40:20
could cause
00:40:21
>> No but I think more dependent model if
00:40:24
Miami I mean
00:40:25
>> Ohio's pretty heavily st favored over
00:40:27
Miami. No, that's that's the first game
00:40:29
by the way. That's actually there's
00:40:31
>> if Miami wins.
00:40:32
>> Yeah.
00:40:33
>> Um there there can be one simple
00:40:34
explanation for that is they just they
00:40:36
just you know underdog wins, right? Big
00:40:38
deal. And that's probably the the more
00:40:40
likely explanation. But this an
00:40:41
alternative explanation is your model
00:40:43
was miscalibrated to start with and my
00:40:45
posteriors on on each of those
00:40:48
possibilities will in will change after
00:40:50
Miami victory. And that so if Miami
00:40:52
wins, I would suggest that there's
00:40:55
probably a a greater probability that
00:40:58
I've done something wrong and the the
00:40:59
model is wrong across all of them. And
00:41:01
so I would therefore imagine that the
00:41:04
numbers are that it's more likely.
00:41:06
>> I hadn't thought about this connection,
00:41:07
but let me tell you about a paper I'm
00:41:09
literally just about to submit with a
00:41:11
doctoral student maybe today, tomorrow,
00:41:13
as soon as we finish up a couple things.
00:41:15
So, I'm sure you guys know this in the
00:41:18
this a good statistics point for our
00:41:20
listeners out here on Morton Moneyball.
00:41:22
And again, this is Eric Brado. I'm here
00:41:23
with Shane Jensen and Audi Winer, some
00:41:25
combination of the three of us and Kate
00:41:26
Massie here every week on the Wharton
00:41:28
podcast network. So, you guys know,
00:41:31
let's assume you're being basian for a
00:41:33
second and you assume that your data
00:41:35
generative process is Gaussian. So, a
00:41:37
normally distributed, right? And let's
00:41:39
say you put a prior distribution on the
00:41:42
mean of that distribution which is also
00:41:44
Gaussian. So you have what's called the
00:41:45
standard normal normal model. The nice
00:41:47
thing is they're what's called conjugate
00:41:49
which all our listeners probably know
00:41:50
what I mean by that. The posterior is
00:41:52
normal. But one thing for sure is true
00:41:54
is that the posterior variance has to be
00:41:56
lower than the prior variance. That's
00:41:58
just a mathematical result of the normal
00:42:00
normal model. Now Audi so it turns out
00:42:04
that the minute you move away from these
00:42:06
families that of course isn't true.
00:42:09
Matter of fact, if you just make a
00:42:10
simple assumption of a mixture
00:42:12
distribution as your prior, then that
00:42:14
phenomenon goes away. The reason I bring
00:42:16
this up is I'll use your words. I'll
00:42:18
just repeat back to our listeners what
00:42:20
you just said. Let's suppose Miami beats
00:42:23
Ohio State. So, one possibility is now
00:42:27
I'm less certain about my model. I have
00:42:30
to raise the posterior variance compared
00:42:32
to the prior variance. Now, I've got an
00:42:34
observation that suggests my model might
00:42:36
need a wider variance. Now that means
00:42:38
even if I don't change the mean
00:42:40
strengths of the teams of the other six
00:42:42
teams playing the probability of the
00:42:44
lower team winning would go up just by
00:42:46
the widening of the distribution. And
00:42:49
no, no, I know I'm saying of course
00:42:50
because you're a statistician, but I'm
00:42:52
just pointing out to people that
00:42:54
>> wait and I'm not even sure that your
00:42:56
first assertion is true that the
00:42:57
posterior variance has to be less than
00:42:59
the
00:42:59
>> It is 100% true.
00:43:01
>> Even if even if there's a mean bias
00:43:03
because there's that, you know, the
00:43:04
discrepancy part of it where there's
00:43:05
there's the extra variance blow up if
00:43:07
the prior mean and the data mean are
00:43:09
super misaligned.
00:43:10
>> So there's two ways by the way in a
00:43:12
mixture you can actually get increased
00:43:15
variance. One is if you get a larger
00:43:18
separation of the means which we call a
00:43:20
change in polarity and the second way
00:43:23
you can do it is of course if the uh
00:43:26
within group variances change but under
00:43:28
the standard normal normal model if you
00:43:30
look at the equation for the
00:43:32
posformation information adds so you can
00:43:35
say 1 / sigma^ 2 + 1 / to^ 2 inverse
00:43:40
that's the posterior variance it always
00:43:43
has to be smaller than 1 sigma squar So
00:43:46
the normal normal model always implies
00:43:49
always implies a lowering of variance no
00:43:52
matter what information signals received
00:43:54
and I'm pointing out that that doesn't
00:43:57
recognize the fact that there could be a
00:44:00
model misspecification or other things.
00:44:03
So I'll be happy to send you the paper
00:44:04
Shane. It's also we're going to post it
00:44:06
on SSRN but I find it and the example we
00:44:09
use is I don't know Shane what's your
00:44:11
you live down in center city. You live
00:44:12
in Philadelphia. What's What's your
00:44:14
favorite restaurant in Philadelphia?
00:44:16
>> I like Barcelona down in East Pass.
00:44:19
>> Barcelona. Great. And so you have a
00:44:21
strong belief that it's a good
00:44:22
restaurant with a pretty narrow
00:44:24
posterior. Okay. Because you've been
00:44:25
there a bunch probably and you like it a
00:44:27
lot. And now all of a sudden imagine you
00:44:28
go to Barcelona and you have a bad
00:44:30
experience. You could easily imagine
00:44:33
your posterior variance being a little
00:44:34
bit wider than it was before. But the
00:44:37
normal normal model will never allow for
00:44:39
that. it will only allow for a smaller
00:44:41
posterior variance no matter what that
00:44:44
information signal is. And that's just
00:44:46
an it's a mathematical restriction of
00:44:48
the normal normal model. And the reason
00:44:50
I know this, by the way, I found this
00:44:52
out. I I'll use Audi as an example.
00:44:54
Audi's a brilliant man. I give him an
00:44:56
I'll use the words of educational
00:44:58
testing because I ETSs for a number of
00:45:00
years. I'm giving Audi an SAT. Audi's
00:45:03
getting questions right
00:45:05
right. I think Audi's really smart with
00:45:07
a low posterior variance. I give him an
00:45:09
easy question. He gets it wrong. Well, a
00:45:12
normal normal model will lower the
00:45:14
variance of Audi's ability, but of
00:45:16
course any rational person would
00:45:18
probably raise the variance of now
00:45:20
because now he's gotten now I'm less
00:45:22
sure that he's as smart as I thought he
00:45:24
was. And so I published a paper of this.
00:45:27
This is my first ever academic
00:45:29
publication. I literally in homage to
00:45:31
Fiser and it is related to that. I
00:45:33
called it negative information that you
00:45:35
can actually receive signals that raise
00:45:37
posterior variance but not under a
00:45:39
simple normal normal model.
00:45:40
>> That would also but you would also think
00:45:42
I'm stupider too.
00:45:44
>> Yeah, it would take the mean.
00:45:45
>> Yes. Yes. Yeah. I'm talking purely about
00:45:49
variance stories. Of course I can
00:45:51
compute the posterior mean. I mean of
00:45:53
course that's going to go downward. That
00:45:54
wouldn't surprise anybody. But the and
00:45:56
the normal normal model will allow for
00:45:58
that. It just won't allow for the
00:45:59
variance to go up. It will not allow for
00:46:02
the variance to go up. Either way,
00:46:03
that's a total aside. It was just based
00:46:05
on your point about if Miami were to
00:46:08
somehow beat OSU, we have to put more
00:46:11
variance now in our model beliefs, which
00:46:13
would by de facto raise the
00:46:15
probabilities of the underdog teams.
00:46:17
>> You got it. That's a good that's a
00:46:19
technical explanation for what I was
00:46:21
predicting. Um,
00:46:23
>> I want to say what caught my eye, and
00:46:24
I'd love to get Shane's thought on this.
00:46:25
So, now this is the third week since I
00:46:28
pointed this out.
00:46:30
that the Colorado Avalanche
00:46:33
still only have two
00:46:36
regulation losses. They've played 38
00:46:39
games. They've won 29,
00:46:42
lost two, and lost seven in overtime.
00:46:45
Now, by the way, you might say it's an
00:46:46
equally statistical anomaly that they've
00:46:48
lost seven in overtime given how good
00:46:50
they are. They're on pace for 140
00:46:53
points, which would break the record of
00:46:54
135. But, you know, I say this every
00:46:56
year about halfway and the teams peter
00:46:58
out and they might end up in the high
00:47:00
120s. I mean, there's no way you would
00:47:01
predict that they're going to win 140
00:47:04
points. Shane, just give I mean to me,
00:47:08
they've played almost half the season
00:47:10
and they've only lost two regular season
00:47:14
games. Isn't that nuts?
00:47:17
>> Yeah. I mean, it's it's it's it's it's a
00:47:20
good record. Like you said, I I I feel
00:47:21
like we've done this to ourselves
00:47:23
several seasons now. um in the last like
00:47:25
decade or so where uh I mean first of
00:47:28
all even if they kind of maintain that
00:47:29
pace you know you know like they may not
00:47:32
even want to be a historical pace going
00:47:35
into the postseason because the last
00:47:37
couple teams that were on that
00:47:38
historical pace flamed out in the first
00:47:40
round of the postseason
00:47:41
>> like the Lightning I remember at least
00:47:42
>> Lightning and the Bruins the Bruins who
00:47:44
have the third point record that team
00:47:47
that year they flamed out quick
00:47:48
>> yeah they won they lost in the first
00:47:50
round same as the and they have the
00:47:52
record at 135 points
00:47:54
Yes. Yeah. Yeah. I mean, again, these
00:47:56
are all sort of the 82 game kind of
00:47:59
records or whatever. You know, this is
00:48:01
uh you know, the Montreal Canadians had
00:48:03
like uh I think won like
00:48:06
>> 131 or 132 or something for years for
00:48:08
like 40 years theirs was considered the
00:48:10
real record. But uh but yeah, I mean I I
00:48:13
do think they are on a historical pace.
00:48:15
The specifically the two losses does
00:48:18
yeah stand out. I mean, if they had if
00:48:20
they finished the season with like, you
00:48:22
know, say less than, you know, eight
00:48:24
losses or something like that, that'll
00:48:26
probably be like the loss record
00:48:28
certainly. Um, and I know, I mean, they
00:48:30
they've just been absolutely excellent
00:48:32
uh um looking unstoppable. It just it
00:48:35
does seem I I think it is a curiosity
00:48:37
that the last few years we've seen these
00:48:39
kind of seemingly unstoppable teams roll
00:48:41
through the regular season and then get
00:48:44
immediately stopped. I don't think it
00:48:46
>> is there any reason you see that um any
00:48:48
reason I maybe we've talked about this I
00:48:50
just don't remember why you know I think
00:48:53
it was the the Canadians record was
00:48:55
something like 1977 I may have or 81 or
00:48:58
somewhere whatever '7s
00:49:00
>> in the 70s okay and then that record
00:49:02
held for 40some years and then all of a
00:49:06
sudden now the Bruins broke it and now
00:49:08
we're seeing pot and the lightning had a
00:49:09
great season and then the Bru any
00:49:11
explanation for that I mean just there
00:49:13
are a bunch of really crappy teams
00:49:14
teams. We're seeing the variation in
00:49:16
team strengths go. Is that what's
00:49:18
happening?
00:49:18
>> Yeah, it does seem like that. But I
00:49:21
don't actually know. I mean, we should
00:49:22
need we we would need to get a real
00:49:25
hockey expert on that kind of is looking
00:49:26
at that distribution, I guess, season to
00:49:28
season a little bit more. It could just
00:49:30
be kind of these sort of one, you know,
00:49:33
I mean, c certainly comparatively
00:49:37
to like the when the Canadians were
00:49:40
doing, I feel like the lead is less
00:49:42
dynastic. I mean the Canadians back I
00:49:44
mean this is pre free agency and all
00:49:46
this type of stuff. I I feel like you
00:49:48
you'd kind of think that most of the
00:49:49
sort of forces that are around now
00:49:51
influencing hockey would go towards more
00:49:54
parody not less. Uh but again the tail
00:49:58
behavior of that we can we've sort of
00:49:59
seen in baseball that like you know
00:50:02
there's been obviously a lot of parody
00:50:04
kind of in general I think over the last
00:50:06
couple decades in baseball in terms of
00:50:07
the teams that contend versus not and
00:50:09
some of the forces in play but you know
00:50:12
there still is like occasionally you'll
00:50:14
have a season where there's like three
00:50:16
100 win teams and there'll be like none
00:50:18
the next season. I don't I don't know
00:50:19
the kind of tail behavior even if you
00:50:21
don't necessarily have a even if you
00:50:23
have a relatively stationary system. I I
00:50:25
I don't know what I don't have a good
00:50:27
intuition about that.
00:50:28
>> Yeah. I think as I remember maybe just
00:50:30
go back to baseball for a second. Am I
00:50:31
I'm right. Right. There was no 100 win
00:50:32
team in baseball last year. Right.
00:50:34
>> There was none.
00:50:35
>> And we were shocked. Shocking. Right.
00:50:38
>> Dodgers started off with like 13 wins in
00:50:40
a row or something and they were their
00:50:41
preseason prediction was nearly 100 to
00:50:43
start with and after winning so many we
00:50:45
would have thought for sure they would
00:50:46
have won. We we thought it we made
00:50:48
those.
00:50:49
>> We've also had a couple teams that have
00:50:50
kind of hit historical lows over the
00:50:51
last couple years in baseball. Is that
00:50:53
the new rule or is that just kind of a a
00:50:55
two-off kind of scenario or something?
00:50:57
>> Yeah, I lost that bet. Remember this
00:50:59
year I I predicted I forget who it was.
00:51:01
Was it the
00:51:01
>> Colorado? You thought they would be
00:51:03
>> Yeah, I thought and I don't know how
00:51:04
many did they end up at 50.
00:51:06
>> They they did not set any records. No,
00:51:08
they they
00:51:08
>> was the record, but I I know that I had
00:51:10
predicted I think I predicted like 35
00:51:12
and you guys said they're going to start
00:51:14
winning.
00:51:14
>> Yeah, I just followed Gton. You know, I
00:51:16
regression to the mean in the end of the
00:51:18
day. Hard to top. Well, guys, in the
00:51:20
last few minutes, let me throw out some
00:51:21
topics just to get your reaction to. I
00:51:24
put this in the rundown before last
00:51:26
night's game where he got injured and
00:51:27
now he's out for four weeks. I'm talking
00:51:30
about Nicole Joic, uh, the three-time
00:51:32
MVP of the NBA. Let me just tell you
00:51:35
guys his stats.
00:51:38
29.9 points a game, 11.1 assists, and
00:51:43
12.4 rebounds. He leads the league in
00:51:47
assists and rebounds and he's averaging
00:51:50
30 points a game and he's a center. So
00:51:54
at some point
00:51:56
we're going to have to start Am I wrong
00:51:58
that we're going to have to start
00:51:59
listing this man? I if he hadn't injured
00:52:01
his knee he's literally injured it last
00:52:03
or yeah injured his knee. He's out for
00:52:05
four weeks now. He might have won the
00:52:06
MVP again which puts him at four MVPs
00:52:10
but he's got three. Not bad.
00:52:12
You're an NBA guy. Is this one of the
00:52:14
greatest players in NBA history?
00:52:17
>> I mean, starting to look that way. I I I
00:52:19
think it's kind of um
00:52:22
you know, I mean, I I think with these
00:52:24
er era kind of comparisons, I actually
00:52:26
am not the student of his that NBA kind
00:52:28
of history. Um so I I I guess you know,
00:52:33
if we can kind of try and talk at least
00:52:35
anecdotally about the kind of dominance
00:52:36
he's displaying, what the what what kind
00:52:38
of a historical analog would be. Would
00:52:40
it be kind of Shaquille O'Neal? Is that
00:52:42
kind of, you know, I'm trying to think
00:52:45
of sort of center points and reboundated
00:52:48
in in this gamechanging way. I know
00:52:50
Shaquille probably doesn't have the
00:52:52
point totals, but would that be kind of
00:52:54
the closest analog that we've been
00:52:55
looking at?
00:52:56
>> I think so, except he's a much better
00:52:58
passer. I'm pretty sure I don't know
00:52:59
that this is true. I'm my I'm going to
00:53:02
say a statement and our our listeners at
00:53:04
W Moneyball can say Eric doesn't
00:53:06
remember when this happened. I'm pretty
00:53:08
sure Wilt Chamberlain was the last
00:53:10
center to lead the league in assists.
00:53:13
>> Wow. Yeah. So, that's pretty amazing.
00:53:17
>> I mean, to me, the 11.1 assists are more
00:53:20
impressive for a center than the 30
00:53:22
points and 12 rebounds. I mean, that's
00:53:25
impressive, too. But you I'm sure you
00:53:27
could look it up while I'm speaking
00:53:28
here. I'm I'm I know I know for a fact
00:53:30
Chamber led the league in assists one
00:53:32
year, but it might be the last time a
00:53:33
center led the league in assists,
00:53:36
>> right? And and would you sort of say
00:53:38
like you know physically like is he even
00:53:42
the kind of most unique sort of center
00:53:44
in the league right now because you've
00:53:45
got somebody like Wemmen Yama. So like
00:53:48
you know
00:53:48
>> oh Joe gets can't jump. He can't run
00:53:52
fast. He he's got no lift. Um he's not
00:53:56
particularly muscled. Um might have you
00:53:58
know no I know.
00:54:00
>> Is he he is he the most uninterestingly
00:54:02
dominant player ever? [laughter]
00:54:04
>> Yes. Yeah. I would say that's I I'd say
00:54:07
that's I would say that's fair to say.
00:54:09
You know, it's also, you know, since I'm
00:54:11
a big Larry Bird fan, I was always a fan
00:54:13
of his that and partially because of
00:54:14
that they always said, you know, Larry
00:54:16
Bird couldn't jump, he couldn't run
00:54:18
fast, you know, um, etc. But his
00:54:21
knowledge of the game, you know, and his
00:54:23
ability to lead teams, I think when I
00:54:26
see Jokic play, I'm just, you know, guy
00:54:28
almost never makes the wrong play.
00:54:30
Either way, I just think we have to
00:54:31
start talking to him about the best
00:54:32
ever. Um, I you mentioned golf. We have
00:54:35
to, you know, in homage to him today.
00:54:38
Um, today's Tiger Wood's 50th birthday.
00:54:42
>> Wow. Congrats.
00:54:43
>> Young man. Young man. You know, you have
00:54:45
to remember for me, Tiger Woods, we were
00:54:47
both at Stanford at the same time. Um,
00:54:49
so
00:54:51
>> his undergraduate and your graduate
00:54:53
school.
00:54:54
>> Yeah. I I was finishing my graduate
00:54:55
degree. But here's my question.
00:54:58
>> Yeah. something magically happens in the
00:55:00
world of golf which could help Tiger
00:55:02
Woods when you turn 50 is there's this
00:55:04
thing called the senior tour
00:55:06
big advantage you get to use a cart and
00:55:08
we know part of the challenge is he says
00:55:10
I can swing a club I just can't walk
00:55:12
well you can use a cart third they only
00:55:14
second they only play three rounds
00:55:17
instead of four so
00:55:21
this is one of those interesting
00:55:22
questions like I mean he can still play
00:55:24
the Masters by the way and he can still
00:55:26
play all the majors I think I'm going to
00:55:29
make a forecast here. I think he moves
00:55:31
to the senior tour just because his body
00:55:33
he just I mean he has played one
00:55:34
tournament in two and a half years. He
00:55:36
could play as many senior tournaments as
00:55:37
he wants and then plays the four majors
00:55:40
like if he comes back why Shane any
00:55:42
thoughts?
00:55:43
>> Yeah. No, I mean the only thing I would
00:55:45
sort of say is you know we don't know
00:55:47
kind of like motivation for actually
00:55:48
coming back you know I mean he he could
00:55:51
just ride off into retirement. He
00:55:53
certainly got the accolades and
00:55:54
everything like that. I think it will be
00:55:56
kind of whether his accumulated injury
00:55:57
history if he can kind of have that more
00:56:00
great you can kind of put it together
00:56:02
physically for the for the senior tour.
00:56:06
I mean it would be wide open. I mean
00:56:08
he'd instantly be favor, right? I think
00:56:11
coming into that uh situation. So um so
00:56:14
yeah. No, I mean I would I kind of I
00:56:16
would love to see it. I just you know um
00:56:19
don't know necessarily
00:56:21
it'll be kind of I think a motivation
00:56:22
factor.
00:56:24
So, I have one more topic I want to ask
00:56:26
you guys about. So, I I wasted time on I
00:56:28
guess it was Sunday morning and I
00:56:30
watched this awful made for TV tennis
00:56:33
event
00:56:35
where Nick Curios do you know who Nick
00:56:37
Curios is, guys?
00:56:38
>> Yes, he's a pretty bad tennis player.
00:56:41
>> Well, he's he used to be okay. He was a
00:56:43
top 10 player at one point. He made it
00:56:44
to the Wimbleton final. He lost in the
00:56:46
Wimbleton final. He's currently number
00:56:49
672 in the world
00:56:51
>> because he hasn't played much. Um, he
00:56:53
played the number one woman in the
00:56:54
world, Arena Sabalanka, in a best of
00:56:58
three match. Okay. And but the reason it
00:57:02
was awful was because you only allowed
00:57:05
one serve, both players.
00:57:07
>> His court was actually wider than her
00:57:10
court. So literally, Shane, they made a
00:57:12
court that looked like this on one side
00:57:14
of the net and then like this on the
00:57:16
other side of the net.
00:57:17
>> For our listeners who weren't seeing, it
00:57:20
was substantially bigger. his his he was
00:57:22
he was very limited.
00:57:23
>> He was Yeah, it was limited. He won
00:57:25
6363.
00:57:27
>> And by the way, he was also toying.
00:57:30
>> He was No. No. So, look. So, I asked
00:57:34
Here's what I asked GPT.
00:57:38
Can you rate the strength of men's and
00:57:42
women's tennis players in one ranking?
00:57:46
Now, here's what it did. It did it. But
00:57:51
I'll give it credit. It says tennis
00:57:54
abstracts ELO is calculated separately
00:57:56
within each tour's match pool because
00:57:58
there's no cross tour match data. The
00:58:01
absolute ENO numbers aren't calibrated
00:58:03
for men's versus women comparison. So
00:58:05
this combined list is best read as top
00:58:08
by ELO within their respective tours. So
00:58:10
I give it credit for recognizing that.
00:58:13
Interestingly,
00:58:14
Center and Alcarz are at the
00:58:16
>> I'm glad TPT is trying now pushing back
00:58:18
on some of your more uh more elaborative
00:58:20
plans here.
00:58:21
>> I agree with that. Um it has S and
00:58:23
Alcarez on top and then it has the next
00:58:26
four players are all women. Sabalanka
00:58:29
third, Switite fourth, Rabbakana fifth
00:58:32
and Koko Goff sixth. So my only point is
00:58:36
is that um it is true it was this funny
00:58:40
match but look even in their primes
00:58:43
remember this isn't Billy Jean King
00:58:45
against a 55year-old Bobby Riggs by the
00:58:47
way you know which was the original
00:58:49
Battle of the Sex's back in 73 I mean
00:58:52
Curios is a 30-year-old former top 10
00:58:55
player and they're playing best of three
00:58:58
um you know even in her prime Naverova
00:59:01
they always wanted her to play the and
00:59:03
her comment was she couldn't even beat
00:59:05
the top hundth man in the world and said
00:59:07
it wouldn't even be close. She her
00:59:09
comment was she would have lost six love
00:59:10
six love to number 100 man in the world.
00:59:13
All I'm saying is this to me was not an
00:59:16
event to show the relative. That's what
00:59:19
we I mean that's kind of what we're
00:59:20
getting at is if it if it was really a
00:59:23
fair sort of you know fair playing
00:59:26
surface literally um
00:59:28
>> yeah play the match
00:59:30
>> the usual what we call a tennis match
00:59:32
but ranking with the top women like what
00:59:35
would be the kind of fair ranking of
00:59:37
like say number one in the uh women's
00:59:39
side versus like the men's side and you
00:59:42
know Naverova's estimate is like for
00:59:44
herself when it was herself is like at
00:59:46
the 100 level. No, no. She said she
00:59:48
would have gotten beaten easily by
00:59:50
theund. She might put herself at 500.
00:59:55
>> So either way, it was just I I just
00:59:57
found it interesting. I I thought put
01:00:00
this way. Um
01:00:02
it was an interesting match to watch,
01:00:04
but it didn't tell me anything about
01:00:05
like I do want I wouldn't mind answering
01:00:08
the question. I mean, look,
01:00:09
>> why don't every tournament we just kind
01:00:11
of have like a round like some play like
01:00:13
like you know the people that like lose
01:00:15
out in the early rounds? Couldn't we do
01:00:16
a little bit of extra like kind of like
01:00:18
games to like like across the across the
01:00:21
g uh genders to sort of uh try and sus
01:00:23
this out slowly?
01:00:24
>> I love even today I love women's tennis.
01:00:27
I might love women's tennis more than I
01:00:29
love men's tennis. I love women's I love
01:00:31
I have so much uh respect for the
01:00:34
women's game. I love everything about
01:00:36
the women's game. But you know just as
01:00:38
an academic I must admit you know I I
01:00:41
have to admit I'm thinking to myself is
01:00:43
there some way because look we did this
01:00:45
at ETS2 like if there's no overlap
01:00:49
between two like can I like let's take
01:00:52
an a simp that's not a simple example
01:00:54
but it's a simple enough example suppose
01:00:56
I want to know who's smarter someone
01:00:59
that gets a 780 on the physics AP exam
01:01:04
and someone that gets a 780 on the
01:01:06
French seven AP exam. And of course, if
01:01:09
there's overlap, then you could
01:01:11
potentially try to statistically answer
01:01:13
that question. But I'm trying to the the
01:01:15
bigger picture I'm talking about isn't
01:01:16
men's versus women's tennis, cuz I I
01:01:18
love them both. Um what do you do when
01:01:21
you don't have overlap between
01:01:24
distributions? Do you guys remember the
01:01:26
bridging eras in sports paper? That's
01:01:28
the way they did it is that you overlap
01:01:30
and you know Mickey Man Ruth played with
01:01:32
Dagi or Garrick played with Dagio and
01:01:34
Deagio played with Mantle and then
01:01:36
Mantle played with this so you have
01:01:38
overlapping designs but here we
01:01:40
essentially have a zero overlap design a
01:01:43
can be done well and some overlap you
01:01:46
got the mixed doubles right
01:01:48
>> you know that's right
01:01:50
>> that's not I mean a is there anything
01:01:52
that can be done
01:01:53
>> no in fairness I mean until you have
01:01:56
unless you have un there is some
01:01:59
overlap. It's just not at the
01:02:00
professional level. You have to go down
01:02:01
to the amateur level and from there you
01:02:04
can make some you can do something. But
01:02:06
listen, it's it's it's monumental. It's
01:02:08
just absolutely I mean it was not only
01:02:10
Martino I think it was also did wasn't
01:02:12
it um Serena Williams didn't she say
01:02:15
she'd be beaten six love six love
01:02:17
against
01:02:18
>> Serena Williams said the same thing
01:02:19
>> and uh and and basically she says it's a
01:02:22
different game. I mean they they they're
01:02:23
much faster. They hit much harder. Um,
01:02:25
and they're not they're not I mean the
01:02:28
it's just to speak back of that about
01:02:30
that overlapping errors paper. You can't
01:02:32
even do that without without modeling.
01:02:34
You still need
01:02:35
>> I mean you the way you can achieve
01:02:37
overlap is through model assumptions
01:02:39
through model assumptions.
01:02:41
>> I mean it's not like timing. I mean if
01:02:43
with swimming and track it's just a time
01:02:46
you can you can compare that. Um that's
01:02:48
about the only sport that has an
01:02:49
absolute metric. And even then people
01:02:51
talk about Jesse Owens he didn't have he
01:02:53
didn't have the track. He didn't have
01:02:55
the shoes. He didn't have, you know, and
01:02:56
then you you might want to wonder what
01:02:58
would how fast would Jesse Owens have
01:03:00
been if he had been trained and and
01:03:02
geared up in today's world. And people
01:03:05
try to make forecasts of that.
01:03:06
>> I think what we can all agree on here
01:03:08
is, you know, since time happens,
01:03:12
weather happens,
01:03:14
different equipment happens, there's no
01:03:17
really such thing as an exact applesto
01:03:20
apples replication. Like you you you
01:03:22
always need to like there's like and we
01:03:24
can decide those variables are
01:03:26
>> under the assumption that time is a
01:03:27
straight arrow and not a flat circle.
01:03:29
>> Yeah.
01:03:30
>> If time is a flat circle we will come
01:03:31
back to a replicatable. I mean we we do
01:03:34
like I mean a great question for us as
01:03:35
statisticians to ask is across all
01:03:37
sports sporting events but baseball
01:03:39
football not only team but also
01:03:41
individual which what is the oldest what
01:03:45
is the performance of yester year that
01:03:47
would still be
01:03:50
considered dominant by today's standards
01:03:52
considering there's so much improvement
01:03:54
in athleticism gear etc and what
01:03:58
obviously comes to my mind immediately
01:03:59
is secretariat who still has all the
01:04:01
records and that's about 50 years ago
01:04:03
know um
01:04:04
>> 52 years still holds the record all
01:04:06
three tracks
01:04:07
>> and [clears throat] it's not even close
01:04:08
I don't think um and so beyond that I
01:04:10
mean you can't you can you can go back
01:04:12
to you know Hack Wilson 180 plus RBI's
01:04:15
190 that's a different time you wouldn't
01:04:17
even think about it Chad Williams
01:04:19
hitting 406 again it's a different time
01:04:22
you can't it's hard to compare them no
01:04:24
tennis player um no none of the athletes
01:04:28
from the basketball or baseball or
01:04:30
football era would you imagine um If you
01:04:33
transform them to today, unless you this
01:04:37
sport has changed too fundamentally in
01:04:38
each of those.
01:04:39
>> Maybe the the one that came to my mind I
01:04:41
I like I like Hack Wilson. I like all of
01:04:44
those things
01:04:45
>> to me. I I forget if he won 10 11
01:04:48
Michael Phelps at the Olympics
01:04:51
that won 11.
01:04:51
>> Yeah, but but listen, Michael Phelps
01:04:53
records are all gone. His last record
01:04:55
>> I just No, no. I'm not saying his
01:04:57
records are still there. I'm just saying
01:04:59
the his dominance dominance
01:05:02
>> every event he competed in
01:05:04
>> eight gold medals in the Olympics. Yeah.
01:05:06
Yeah.
01:05:06
>> Oh no. I mean that even the fact that he
01:05:09
beat out that Spitz guy who was dominant
01:05:11
and like the fact that there seems to be
01:05:13
like once a generation somebody who
01:05:15
comes and completely sweeps across
01:05:19
>> it. It says to me that maybe this maybe
01:05:21
it's less impre I mean Phelps was the
01:05:23
one that did it. So I'm not taking
01:05:25
anything away from him. But right it's a
01:05:26
once in a generation
01:05:27
>> sport where you can sweep across 12
01:05:29
medals because they're all basic.
01:05:32
>> I mean what I mean here question be what
01:05:34
would Chamberlain have looked like if he
01:05:35
had grown up in today?
01:05:37
>> He probably would have been I mean he
01:05:39
wouldn't have been trained the way he
01:05:40
was back then
01:05:42
>> or Bill had a completely different
01:05:43
competition, a different style. would
01:05:45
have been I mean you so you do have to
01:05:48
recognize that that people wouldn't
01:05:50
would would go into the you know one of
01:05:51
the things I remember my favorite was
01:05:53
when I when when my kids were on college
01:05:55
tours I'm not not sure you did this Eric
01:05:56
you probably you're all your boys went
01:05:59
are went to Penn but my kids were
01:06:01
looking at different schools and I
01:06:02
remember once um I was at at Yale for a
01:06:05
reunion and and I had a senior or a
01:06:07
junior in high school and I went to the
01:06:09
admissions committee and they said none
01:06:10
of you would have gotten in today based
01:06:12
on the standards and most of us are
01:06:14
looking each other is and we wouldn't
01:06:16
have looked on paper the way we looked
01:06:19
back then today because we would have
01:06:20
been in the same environment as the kids
01:06:22
today we would have done all the things
01:06:23
that they would have done to make
01:06:25
themselves look differently and
01:06:26
similarly with athletes I mean if you
01:06:28
put Will Chamberlain in today's world he
01:06:30
would look look like he looked back then
01:06:31
if you put Babe Ruth in the world today
01:06:33
he wouldn't be eating hot dogs like that
01:06:35
it wouldn't happen right that world is
01:06:37
gone he would have been forced to be as
01:06:39
as as uh as athletically and
01:06:41
nutritionally minded as any athlete is
01:06:44
>> why we like the secretariat example
01:06:45
because we're not Oh, Secretariat would
01:06:48
would have slacked off or something like
01:06:49
that. [laughter]
01:06:51
>> Oh, Secretariat had heart though, right?
01:06:53
>> Take that out of the equation.
01:06:54
>> Well, Secretary had the biggest heart.
01:06:56
We have to remember
01:06:57
>> biggest heart we've always heard. Well,
01:06:59
guys,
01:07:00
>> if we keep talking, we're going to be
01:07:01
talking right into the new year.
01:07:03
>> All right. Thank you. This has been a
01:07:05
great edition here of Wharton Moneyball
01:07:06
and the Wharton podcast network. Uh on
01:07:08
behalf of all of us, myself, Audi Winer,
01:07:10
Shane Jensen, uh we'd like to thank you
01:07:12
for joining us today. uh and abstensia
01:07:14
Cade Massie, we like to thank you for
01:07:17
being a big part of our show for all of
01:07:19
2025. And um we always like to think I
01:07:22
hope all of my co-hosts agree with me,
01:07:25
the best is yet to come, 11 and a half
01:07:27
years, but uh we've got a lot more in
01:07:29
us. And so please join us for our next
01:07:32
show. Thanks to our producers, uh
01:07:33
Marissa Ren and De Patel. Thanks to our
01:07:35
associate producer and sound engineer,
01:07:37
Dion Simpkins between now and next week.
01:07:40
There's a lot of it. Enjoy your sports.
01:07:42
Enjoy your statistics. and we'll see you
01:07:43
next week here on the Wharton Podcast
01:07:45
Network.
