00:00:00
Welcome everyone to this week's edition
00:00:02
of Wharton Moneyball. I'm Eric Bradlow,
00:00:04
professor of marketing statistics and
00:00:05
data science here at the Wharton School.
00:00:08
Some combination of myself, my two
00:00:09
colleagues are here today, Shane Jensen
00:00:11
and Adi Wyner, both professors of
00:00:13
statistics and data science, and Cade
00:00:15
Massey are here every week on Wharton
00:00:17
Moneyball here on the Wharton Podcast
00:00:19
Network.
00:00:21
Well, Adi and Shane,
00:00:22
uh nothing changes when I say the best
00:00:25
part of our, I guess I'll call it jobs,
00:00:27
non-paying jobs, uh doing what doing
00:00:30
Wharton Moneyball is interviewing people
00:00:32
who have made seminal contributions to
00:00:34
the applications of statistics and
00:00:36
sports. Uh today's guest, Ken Pomeroy,
00:00:39
many-time returning guest, is certainly
00:00:42
no exception to that. Um just for our
00:00:45
fans that don't know Ken, although I'm
00:00:46
sure most of our fans on Wharton
00:00:48
Moneyball do, uh Ken is a college
00:00:50
basketball statistician. He's the
00:00:52
founder of the analytics website KenPom.
00:00:55
It publishes both tempo-based efficiency
00:00:57
ratings and predictive rankings for
00:01:00
every D1 team. So, Ken, on behalf of
00:01:02
myself, Adi, and Shane, welcome back to
00:01:04
Wharton Moneyball.
00:01:06
Yeah, thanks, Eric. It's uh great to be
00:01:07
back on.
00:01:08
So, let's just start with the beginning.
00:01:10
I always like to go back in time, and
00:01:12
then we'll catch up, if you'd like to,
00:01:14
2026 in this year's NCAA tournament. Um
00:01:18
for those people that don't know, when
00:01:20
you first built the KenPom ratings and
00:01:22
website, like usually someone does that
00:01:25
because they see either a deficiency in
00:01:27
what's being done out there, they see a
00:01:29
better way to do something out there, or
00:01:32
they just want to be, you know, build a
00:01:33
following and a community. Um why did
00:01:36
you build your KenPom system to begin
00:01:39
with?
00:01:40
Yeah, it was a a little bit of all of
00:01:42
those things. Um
00:01:44
when I, you know, when I first built it,
00:01:47
it was the kind of the advent of uh
00:01:49
uh
00:01:50
you know, kind of
00:01:52
random people logging on the internet.
00:01:54
Uh the the was just starting to kind of
00:01:55
mature, and uh
00:01:57
uh certainly uh the baseball side of
00:01:59
analytics was starting to explode, and
00:02:01
there were just, you know, these people
00:02:02
online who were getting into advanced
00:02:04
baseball stats, and
00:02:06
you know, kind of like outfoxing like
00:02:08
traditional media in terms of analysis,
00:02:10
and
00:02:11
you know, I like baseball, but I really
00:02:13
love basketball, and I really love
00:02:14
college basketball, and I I looked and
00:02:15
looked for the the version of that
00:02:17
online for for months, and
00:02:19
didn't find it. And so, that's kind of
00:02:21
what uh motivated me to to start my
00:02:23
site, and I never, you know, imagined it
00:02:25
would get to to this point, but I I did
00:02:27
think there would be, you know, a little
00:02:28
bit of a following just because there
00:02:29
was, you know, you were seeing that
00:02:30
following develop for baseball. Mhm.
00:02:32
Um what's What do you think, as you
00:02:35
know, it's been uh so, a few seasons
00:02:36
when the internet started to take off,
00:02:37
has Kenpom, as you know, a site, as the
00:02:40
statistics, has it been around Am I
00:02:43
saying 25 years now? Is that about the
00:02:45
right
00:02:46
like it?
00:02:47
Yeah, you know, when I uh first
00:02:50
uh posted offensive and defensive uh
00:02:52
efficiency for all the teams, that was
00:02:54
at the end of the
00:02:55
2004 season.
00:02:57
Uh Kenpom in some form, like the website
00:02:58
existed before that, and I was doing
00:03:00
college basketball ratings, kind of
00:03:01
basic ratings, like you'd see uh you
00:03:03
know, Jeff Sagarin do. He was He was
00:03:05
really an inspiration for me uh
00:03:07
growing up. But uh yeah, the kind of the
00:03:09
the modern version started in 2004, and
00:03:11
I've I've backdated, you know, some
00:03:13
seasons before that. But yeah, it's, you
00:03:15
know, we're we're definitely over 20
00:03:16
years at this point.
00:03:18
So, what would you say is the most,
00:03:20
like, you know, if you had to be for a
00:03:22
minute, you tend to be modest, if you
00:03:23
were not modest, what would be you would
00:03:26
say would be the biggest impact that
00:03:28
your work has had? Like, what's been the
00:03:29
biggest misconception the way that
00:03:31
you've modeled things has either
00:03:33
corrected, or what do you think you've
00:03:35
added to, you know, as we sit here in
00:03:37
2026, what do you think you've added to
00:03:40
the way in which, let's call it
00:03:42
player or teams effectiveness or
00:03:45
efficiency, or, you know, what are the
00:03:47
sufficient statistics one should measure
00:03:49
when one's thinking about team strength?
00:03:51
How do you think about your
00:03:52
contribution?
00:03:54
I think in a in a really general sense,
00:03:56
it's just about trying to account for
00:03:58
the opportunities that teams or players
00:04:00
have to do things. Um
00:04:03
so,
00:04:04
you know, the the biggest impact on the
00:04:06
team level where, you know, we talk
00:04:08
about possessions per game or pace or
00:04:09
tempo or whatever you want to call it.
00:04:11
Uh you know, in college basketball,
00:04:13
there's such a wide range of styles in
00:04:15
that regard that you can't just
00:04:17
Traditionally, you know, when I started
00:04:19
people just looked at good points scored
00:04:20
per game and points allowed per game and
00:04:22
attributed that to offense and defensive
00:04:24
quality, but you really do have to look
00:04:25
at points per possession and points
00:04:27
allowed per possession uh to give you
00:04:30
the the true picture of of which teams
00:04:31
are most effective and
00:04:33
um
00:04:34
you know, people were thinking about
00:04:35
that before I started my site, but I
00:04:37
just to have kind of a an easy reference
00:04:39
to see how all the teams did in those
00:04:41
stats and you know, make it easy to
00:04:42
compare. That's really I think what what
00:04:44
my main contribution has been.
00:04:46
I would think also and Audie I mean just
00:04:47
Audie wants to jump in in just a second.
00:04:48
I would also think that number of
00:04:50
possessions, you know, and you know, big
00:04:52
end teams that get lots of offensive
00:04:54
rebounds, teams that push the ball more,
00:04:56
uh teams that get turnovers. I would
00:04:58
imagine that, you know, it's some, you
00:05:00
know, as I always say, you know, I'll
00:05:02
use my language of business here in the
00:05:03
marketing department. What the hell, I'm
00:05:04
sitting in my office in the marketing
00:05:06
department. What do retailers care
00:05:08
about? They care about making money per
00:05:09
unit and they care about inventory turn.
00:05:11
So, what do I care about in basketball?
00:05:13
I care about getting a lot of
00:05:14
possessions and being efficient on each
00:05:16
of those possessions. Is that too simple
00:05:17
a way to think about it? No, not at all.
00:05:20
I uh No, that's it gets back to kind of
00:05:22
the opportunity thing, right? Like that
00:05:24
you know, that was one of the early
00:05:25
discoveries, too, is you'd see these
00:05:26
teams that
00:05:27
you know, functionally you'd watch them
00:05:29
and they they'd look really ugly
00:05:31
offensively, you know, uh you wouldn't
00:05:34
you wouldn't say they had an effective
00:05:35
offense, but uh the trick was, you know,
00:05:37
they were getting like a lot of
00:05:38
offensive rebounds or maybe they weren't
00:05:40
committing a lot of turnovers, you know,
00:05:42
it's very easy to count things that do
00:05:43
happen, but to notice things that don't
00:05:45
happen, like not committing turnovers is
00:05:46
something that humans don't necessarily
00:05:48
do very well. So,
00:05:50
you know, you you'd see like teams
00:05:51
ranked pretty high in my offensive
00:05:53
ratings and maybe didn't look pretty,
00:05:54
you know, functionally, but it was
00:05:55
because they were
00:05:57
you know, just getting a lot more
00:05:58
opportunities at shots like you're
00:05:59
talking about.
00:06:01
So, you since we're almost talking
00:06:02
historically here. So, Dean Oliver's
00:06:04
spoken to uh he's been on the show many
00:06:06
times. He's come to speak to our our our
00:06:08
programs here at at Wharton and Penn.
00:06:11
And one of the things that he talks
00:06:12
about is that when he first started
00:06:13
doing basketball analytics and he's sort
00:06:16
of the Bill James of basketball
00:06:17
analytics if you will, he said that's
00:06:19
the statistic of possessions didn't even
00:06:21
exist. It wasn't even something that was
00:06:23
ever recorded, right? And he almost had
00:06:25
to had to take in basketball paper he
00:06:27
talks about how he actually had to go
00:06:29
and record and figure out like how you
00:06:31
can figure out what the pace of play is
00:06:32
cuz it wasn't there. So, uh scroll now
00:06:34
to your work in in college basketball.
00:06:37
How did you get to possessions? Is it
00:06:38
now a stat or is it did you have to work
00:06:40
that as as hard as he did to create that
00:06:42
number? And then of course once you have
00:06:44
possessions you can talk about
00:06:45
efficiency. So, how did it come to be
00:06:46
for in your in your perspective?
00:06:49
Yeah, well first of all I
00:06:50
you know, Dean was a a huge inspiration
00:06:52
for for my work and his book, you know,
00:06:54
his original book I think came out in
00:06:55
2002, which so it's not a coincidence
00:06:58
that like my site started shortly after
00:07:00
that. Like just the ideas he had really
00:07:01
resonated with me. Um yeah, I yeah, so
00:07:04
possessions aren't a published stat. Uh
00:07:06
so, you can either count them
00:07:08
uh
00:07:09
specifically or you can estimate them
00:07:10
from box score data and I still estimate
00:07:12
them from box score data for for most
00:07:14
things I do. Um that gets you really
00:07:16
close enough to the number, but uh
00:07:18
yeah, I mean there's no doubt that uh
00:07:21
there were there were some hurdles, you
00:07:22
know, early on to kind of getting things
00:07:24
started and
00:07:25
that was one of them, but like Dean kind
00:07:27
of blazed the trail on that. So, it
00:07:28
actually, you know, made it really easy
00:07:30
for me to just, you know, follow follow
00:07:32
his work which he helpfully laid out in
00:07:34
his book.
00:07:35
Do you do the four factors in college
00:07:37
basketball? Is that part of your So, can
00:07:39
you explain to our listeners what the
00:07:40
four factors are, please? That'd be
00:07:42
great.
00:07:43
Yeah, so the the four factors are, you
00:07:46
know, just a basic kind of the basic
00:07:49
building blocks of offense or defense.
00:07:52
Uh so, you have shooting, you have
00:07:53
turnovers, you have offensive rebounds,
00:07:55
and you have free throws. And uh the way
00:07:58
those are measured again is by
00:07:59
opportunity. So, you know, field goal
00:08:01
percentage is not straight field goal
00:08:02
percentage. It it's called effective
00:08:04
field goal percentage. It accounts for
00:08:06
the value of a made three-point shot.
00:08:07
So, you know, if I'm shooting
00:08:09
>> That's all these favorite math. Three is
00:08:10
50% more than two. Sure [laughter] is.
00:08:13
In basketball. Yeah.
00:08:15
Yeah. And you find teams that shoot a
00:08:16
lot of three-pointers tend to have lower
00:08:18
field goal percentages than teams that
00:08:20
don't. But, obviously, there's a
00:08:21
trade-off there. When they make their
00:08:22
shots, they're, you know, getting more
00:08:23
points out of it. So, so, there's that.
00:08:25
You know, rebounding is offensive
00:08:26
rebounding percentage. So, it's
00:08:28
basically the percentage of times that
00:08:29
you have a chance to get an offensive
00:08:30
rebound, how often you do so. Turnovers
00:08:33
is turnover percent percentage, so the
00:08:34
percentage of possessions you get a
00:08:36
turnover on. So, yeah, those are the
00:08:38
four factors, and they're they're just
00:08:40
great at
00:08:41
you know, basic explanation of what a
00:08:43
team does or doesn't do well offensively
00:08:45
and defensively.
00:08:46
>> [snorts]
00:08:46
>> So, one of the things that I've noticed,
00:08:47
just to jump in on the four factors, I
00:08:49
actually teach this in my senior
00:08:50
capstone.
00:08:52
And one of the things that's remarkable,
00:08:54
I'm not a basketball person by nature. I
00:08:56
mean, I've certainly played enough
00:08:57
basketball as a kid and in high school.
00:09:00
But, I don't really study it. It's one
00:09:02
of my weakest sports, I think, in terms
00:09:04
of the majors.
00:09:06
But, I so I've I've come to learn about
00:09:08
basketball through statistics, which is
00:09:09
kind of odd, right? But, those four
00:09:11
factors are oddly
00:09:13
uncorrelated, which is shocking to me.
00:09:15
And at least at the professional level.
00:09:18
Is that Is that Can you explain that?
00:09:20
And is it true at collegiate level, too?
00:09:21
I would imagine the collegiate level's
00:09:22
got to be more correlated, cuz good is
00:09:24
good. And when you have tremendous
00:09:26
variation, which you do at in the NCAA,
00:09:29
you probably your best teams are
00:09:30
probably better in some level. But, I
00:09:32
like to
00:09:32
>> And just before you answer that, Ken, I
00:09:33
just want to be clear.
00:09:35
What Audi may be referring to or not,
00:09:37
correct me, correct him.
00:09:39
At the individual player level, this is
00:09:41
a classic aggregation issue I would I
00:09:43
could imagine.
00:09:44
If I'm a good three-point shooter, I may
00:09:46
also be a very good free throw shooter,
00:09:48
but at the aggregate level, at the team
00:09:51
level, they may be uncorrelated when I
00:09:52
look across teams. So, if you could I'm
00:09:54
just modifying Audi's question. I just
00:09:56
want to know if this uncorrelatedness is
00:09:58
at the player level, the team level when
00:10:00
we aggregate, or is that not true at
00:10:01
all? And then then I'll take Curtis
00:10:03
mean.
00:10:04
Uh yeah, that's a lot to ponder. I
00:10:07
I'll
00:10:08
It's It's more I I would say it's more
00:10:11
uncorrelated at the team level than the
00:10:13
player level. Um so, Audi made a good
00:10:14
point that, you know, in college, the
00:10:16
distribution of talent is much greater
00:10:17
than it is in in the NBA, right? It's
00:10:19
the the difference in the best and worst
00:10:20
team is enormous in college. And so,
00:10:23
yeah, so you do tend to see that like
00:10:26
teams that are good in one thing are
00:10:27
going to be good in other things. But,
00:10:29
uh
00:10:30
but there's no like there's no reason
00:10:32
for you to be good at shooting and also
00:10:34
good at offensive rebounding, right?
00:10:35
Like those two things really aren't that
00:10:37
related fundamentally skills. And in
00:10:39
fact, you see like almost no teams are
00:10:42
great at every single four factor. Like
00:10:44
most teams are great at even the great
00:10:47
teams. They might be great at like four
00:10:49
of the eight four factors if you
00:10:50
consider both sides of the ball. Um and
00:10:52
they might be good at two others, and
00:10:54
they might be like below average at two
00:10:57
others. Like nobody's great at
00:10:59
everything. So, teams have all sorts of
00:11:01
different styles, and they specialize in
00:11:03
different things, and uh and yeah, the
00:11:05
fact of the matter is like the way those
00:11:07
four factors are defined, they're
00:11:08
basically orthogonal. Like they're not
00:11:09
the stats don't bleed into to other four
00:11:11
factors. So, you can be great at one
00:11:14
thing and poor at another, and it's not
00:11:15
unusual to see very good teams behave
00:11:17
that way. Yeah, well, I guess you're uh
00:11:18
just to kind of follow up, you're
00:11:19
talking about kind of specialization at
00:11:22
the
00:11:23
team level. I when when kind of I
00:11:25
thought about what might drive a lack of
00:11:27
correlation, uh it would be more
00:11:29
specialization at the individual level.
00:11:30
And I know like cuz I I think about
00:11:32
baseball. Obviously in baseball it's
00:11:34
very there's very much of a you know,
00:11:35
like fielding even like fielding and
00:11:37
hitting are not are often negatively
00:11:39
correlated because there's a real
00:11:40
specialization there.
00:11:42
Basketball obviously there's not as much
00:11:44
positional specialization, but is is
00:11:47
that a large part of what's driving
00:11:48
maybe a lack of correlation? Yeah, I
00:11:50
mean there's there is some
00:11:51
specialization like often times you if
00:11:53
you look at a you know, a team so first
00:11:55
of all like in basketball we really
00:11:56
struggle to measure measure individual
00:11:58
defense, right? We don't have really
00:11:59
many good stats for that. You know,
00:12:01
block shots are good for for tall guys,
00:12:03
but for small guys
00:12:04
you know, they're not going to block
00:12:05
shots and they still can be effective
00:12:06
defenders. Um so often times when you
00:12:08
look at like uh season stats and you see
00:12:10
like a player who doesn't contribute
00:12:12
much offensively like it's almost
00:12:14
certain that they're a good defender.
00:12:15
Like that's why they're on the court. So
00:12:17
you do see that. I think another like
00:12:19
another great example of aggregation is
00:12:20
that when you compare offensive
00:12:22
rebounding percentage and defensive
00:12:23
rebounding percentage for a team and and
00:12:24
look at all of division one, there
00:12:26
really is hardly any correlation between
00:12:28
those two. So if I tell you that a team
00:12:30
is a great offensive rebounding team, I
00:12:31
really haven't told you anything about
00:12:33
whether they're a good defensive
00:12:34
rebounding team. And yet on the player
00:12:36
level there obviously is a high
00:12:37
correlation there, right? Like a good
00:12:39
offensive rebounder individually is
00:12:40
probably a good offensive rebound or a
00:12:42
good defensive rebounder as well. So
00:12:44
that's to me that's the best example of
00:12:45
that effect you guys are talking about.
00:12:47
Let me ask you the following question.
00:12:48
Let's imagine
00:12:50
I took just those four factors
00:12:53
and I use those as a predictive model to
00:12:55
predict the outcome of winning versus
00:12:57
something more sophisticated that I'm
00:12:59
sure you done and built over the last 22
00:13:02
23 plus years. How much do we lose? Like
00:13:06
are those sufficient statistics? I don't
00:13:08
mean literally, but quasi sufficient for
00:13:11
predicting game outcomes? And if the
00:13:13
answer's yes,
00:13:14
great. If not, what else is kind of
00:13:16
missing?
00:13:18
No, there's not there's not much
00:13:19
missing. Like if all you have are are
00:13:20
box score stats,
00:13:22
those four factors explain like you
00:13:23
know, 99% of offense and defense. The
00:13:26
only thing that's missing are like data
00:13:27
errors and
00:13:29
uh you know, the possession estimation
00:13:30
and things like that.
00:13:32
But yeah, I mean for the most part, you
00:13:33
know, yeah, you're going to find you
00:13:35
plug those guys into a regression or
00:13:37
whatever and you know, obviously like
00:13:38
shooting's going to dominate. It's going
00:13:40
to be the most important thing and then
00:13:41
offensive rebounding and turnovers are
00:13:42
somewhere like tied for a decent second
00:13:44
and then like free throws are like
00:13:46
you know, the weakest thing. But yeah,
00:13:48
you you you'd have a really nice model
00:13:50
if you if you use those four things. Can
00:13:52
I uh let me just jump in cuz actually I
00:13:53
have this data for the NBA, so not the
00:13:55
collegiate. So if you try to predict
00:13:57
wins using
00:13:59
point differential, that's about as best
00:14:01
as you can get. I mean the game is down
00:14:02
to points, right? So yeah, you can end
00:14:04
up have a lot of blowouts and your your
00:14:06
point differential might not that be
00:14:08
perfectly accurate with your wins. You
00:14:09
can it's it that's you can't expect
00:14:12
perfection. But four factors works
00:14:14
almost as good.
00:14:15
It's just That's the point in the
00:14:18
>> It's it really is amazing how how
00:14:21
I mean there's leftover residual
00:14:22
variance because of how you distribute
00:14:24
your points. But you're and I'm talking
00:14:26
about season-wide four factors are and I
00:14:29
just use four. I'm but I use the
00:14:31
differential. So the way I use I can
00:14:32
create a delta four factor
00:14:34
for each of them and instead of having
00:14:36
eight, you know, four on each side. So
00:14:38
I'm sure I'd have even better if I did
00:14:40
if I went down to that.
00:14:41
It's just amazing. I'm very excited
00:14:43
about this and I when I teach this I get
00:14:45
excited about my students usually look
00:14:46
at me and I'm like, why do you care? But
00:14:48
it's because they don't have experience
00:14:49
and they don't know how valuable that
00:14:51
is.
00:14:51
>> a dream regression scenario you where
00:14:53
you've got like four different factors
00:14:55
all which are
00:14:56
important but not particularly highly
00:14:58
correlated with each other.
00:15:00
Yeah. Well, I was just going to say I
00:15:02
mean that's you know, that's why Dean
00:15:03
Oliver chose those. That was really
00:15:04
intentional the way he
00:15:06
you know, it's pretty obvious like when
00:15:07
you think about it, it's obvious like
00:15:08
those things should be the four factors.
00:15:10
But just the way they're they're
00:15:11
measured and the fact that they're you
00:15:13
know, again, they're like the
00:15:15
measurements are completely orthogonal,
00:15:16
so they're not you know, directly
00:15:17
related to each other in any way.
00:15:19
So let me ask you a follow-up to that.
00:15:21
So I in the when I introduced you, I I
00:15:24
read and predictive rankings for every
00:15:26
D1 team.
00:15:27
So,
00:15:28
how do you I mean, obviously you have a
00:15:30
ranking vector. Obviously we get
00:15:32
outcomes at the end of the season,
00:15:33
which, you know, get the NCAA
00:15:35
tournament, but how do you score
00:15:37
yourself? Like, anybody that does
00:15:40
predictions always has, let's call it an
00:15:42
error metric of some sort and says, "How
00:15:44
well did I do?" And then possibly that
00:15:47
leads in good models to model
00:15:49
modification if you notice some sort of
00:15:51
systematic errors or stuff. So, I So,
00:15:53
how do you score how well you've done in
00:15:56
ranking teams? And how does that lead,
00:15:59
well, if you notice something to any
00:16:01
type of modification of how you do
00:16:02
things?
00:16:04
Right. So, the the main way I score
00:16:06
myself is by looking at my game
00:16:08
predictions.
00:16:09
Um,
00:16:11
yeah, so I do look at the overall
00:16:13
ranking as well, but primarily I
00:16:15
I use that to like judge my I have a
00:16:19
preseason ranking as well, which is
00:16:20
completely separate calculation, but
00:16:22
they're really like judged on, you know,
00:16:23
compared to the final ranking. But,
00:16:25
overall like the algorithm, I yeah, I
00:16:27
basically focus on game prediction. So,
00:16:29
both uh, you know, predicted uh, point
00:16:31
differential and uh, win probability.
00:16:34
And, you know, just a basic How is your
00:16:36
model How is your model We always like
00:16:37
to ask this question. Is your model
00:16:39
properly calibrated? Which means when
00:16:42
you just I want to make sure our
00:16:43
listeners understand, but they do. Um,
00:16:45
when the KenPom system says team A has a
00:16:47
70% chance, if you bucket all of those
00:16:50
around 70%, is it near 70%? If you
00:16:52
bucket the 20% ones, is it around 20%?
00:16:55
So, what do you find there and do you
00:16:57
have the classic problem Shane talks
00:16:59
about this all the time? Do you have
00:17:00
problems out in the tails, which is
00:17:02
typically where we find, you know,
00:17:04
models don't do as well out in the
00:17:06
extremes?
00:17:08
Yeah, so it is pretty reliable. Um,
00:17:10
yeah, I have like I have a page for
00:17:12
those for those stats that updates in in
00:17:14
real time basically. And I'm probably
00:17:15
the only probably the only one that
00:17:16
looks at it, but you know, I'm always
00:17:19
curious. Uh, Uh yeah, it it does really
00:17:21
well. In the extremes, it probably um
00:17:24
under predicts the favorite slightly.
00:17:26
Like I think this year there were like
00:17:28
260-ish
00:17:30
games where I gave the underdog like
00:17:33
less than a 2% chance of winning.
00:17:35
And they won zero of those games.
00:17:38
And they should have probably won like
00:17:40
three or something, you know, based on
00:17:41
the probabilities. So, there's a little
00:17:43
bit of uh
00:17:45
overconfidence, I guess, with with
00:17:47
extreme underdogs.
00:17:49
But outside of that, like for the most
00:17:50
part, it does pretty well. Certainly
00:17:51
does pretty well in the probabilities.
00:17:52
It has you know, it has another issue
00:17:54
with like scoring margin for extreme
00:17:56
underdogs as well. It's constantly like
00:17:58
overconfident on these just really super
00:18:00
lopsided matchups.
00:18:01
Um but
00:18:03
Overconfident in in favor of the
00:18:05
underdog. Exactly, yeah. So, like, you
00:18:08
know, when the market says a team's
00:18:09
favored by 40, like I have them favored
00:18:11
by like 32 or something. And Well,
00:18:13
Oddie's all for that. Oddie's all for
00:18:15
regressing back toward some average.
00:18:17
Well, I mean, let me let me just point
00:18:18
out a couple things. Um first with
00:18:20
response to that, this is something I
00:18:21
also do with my class.
00:18:23
There's a sort of the big home dog
00:18:25
effect in betting,
00:18:27
um which is that when a team is playing
00:18:29
at home and they're the big underdog,
00:18:32
they have a tendency historically to
00:18:34
cover the spread.
00:18:35
And uh fascinating because that goes
00:18:37
back to the '70s. And I I look at this
00:18:39
historical data in football in
00:18:41
particular where I have that data. And
00:18:43
it just in every 5-year period, it just
00:18:45
sticks out, sticks out, sticks out. And
00:18:47
it goes all the way through like 2015
00:18:49
and then it stops. It's just sort of
00:18:50
gone. As if And that may be because
00:18:52
people are starting to bet it more more
00:18:54
um
00:18:55
uh globally.
00:18:56
Hard to know exactly. But that's that's
00:18:58
potentially what you're seeing in
00:18:59
basketball is that the model, because
00:19:01
you're not really talking about betting,
00:19:02
the model says 40 points, but we know
00:19:04
that home team's going to outplay there
00:19:06
when they're the big dog. Maybe you're
00:19:08
seeing that. But my question is, um Um,
00:19:10
way back in the way In fact, you were
00:19:12
one of our earliest guests in in in in
00:19:14
what in what Moneyball. Um, we were
00:19:16
trying to figure out what some of the
00:19:17
secret sauces are in NCAA tournament
00:19:20
prediction. And I have to thank you
00:19:21
right away because, you know, I I only
00:19:23
play in a small family tournament, but
00:19:25
I've won 40 per 40 four out of five of
00:19:27
them. Um, and I just use your your
00:19:29
inside information. So, I should send
00:19:30
you send you a check, but if only for
00:19:32
you for a couple bucks. Just to get that
00:19:34
out there. So, thank you, Ken. But, the
00:19:35
question, um, this is I remember Nate
00:19:37
Silver pointing this out. Um, he used to
00:19:40
say some of the secret sauce was
00:19:41
shrinkage or regression towards
00:19:43
preseason rankings. And that um, you
00:19:46
tended to over overvalue the people the
00:19:49
teams that had really good seasons.
00:19:52
And um, and one way to to account for
00:19:54
that was just regress them slightly to
00:19:56
their preseason expectations. Is that
00:19:58
still true?
00:19:59
Yeah, it is. You do see that effect. Um,
00:20:02
so, you know, in my system, I you know,
00:20:04
my preseason ratings are pretty much
00:20:06
washed out by now and like
00:20:08
I really wanted maximum accuracy. They
00:20:10
They wouldn't, but there's sort of a
00:20:12
perception issue with, you know,
00:20:14
the fact that my ratings are used on the
00:20:15
NCAA team sheets for selection and
00:20:17
things like that that, you know, you
00:20:18
probably don't want preseason ratings
00:20:19
influencing them, but
00:20:21
um, that is a factor. I mean, one of the
00:20:22
one of my favorite stats, which may or
00:20:24
may not come into play this year, but
00:20:26
uh, teams that are teams that
00:20:29
end up being one or two seeds in the
00:20:30
NCAA tournament, uh, if they were not
00:20:32
ranked in the preseason AP poll,
00:20:35
uh,
00:20:36
there's like 38-ish, I think, of these
00:20:39
cases over the past
00:20:40
40 years, none of them have made the
00:20:42
final four. Wow.
00:20:44
Which you normally expect these are one
00:20:45
or two seeds, right? You normally expect
00:20:46
like
00:20:48
eight or nine of them to make the final
00:20:49
four, but so, it's like definitely a
00:20:50
significant trend and like I said, I
00:20:52
don't think it'll happen this year.
00:20:53
Nebraska's the only possible case if
00:20:55
they end up as a two seed, but uh,
00:20:56
that's, you know, really an example
00:20:57
where like, yeah, clearly like teams
00:20:59
have like overshot their preseason
00:21:01
expectations and it it does, you know,
00:21:04
you do see some regression when it comes
00:21:05
to the NCAA tournament.
00:21:07
With strength of schedule
00:21:09
kind of is is that part of that story as
00:21:11
well or is it more just
00:21:14
um what what do you think is is is kind
00:21:16
of driving that? No, I think it's just
00:21:18
what we're we're talking about, right?
00:21:20
Like you you know if you had
00:21:22
you know, you want to get perfect
00:21:23
information on a team, right? Going into
00:21:25
the tournament. And perfect information
00:21:27
is not solely based on what they did
00:21:28
that season. Like the NCAA basketball
00:21:30
season short, you know, it's only 30 33
00:21:32
games before you get to the tournament.
00:21:34
A lot of those games are mismatches, you
00:21:36
know.
00:21:36
Yeah, I guess that's what I was going
00:21:38
with with kind of very uneven schedule,
00:21:39
I guess, yeah. Yeah, so you know, you
00:21:41
maybe have like 20 to 25 games of
00:21:43
meaningful information and uh
00:21:46
there's still meaningful information in
00:21:47
the preseason, Paul. So yeah, you
00:21:49
definitely uh you know, want to include
00:21:50
that when you're making projections. So
00:21:52
can I haven't been following it as
00:21:53
closely as I normally do. I don't know
00:21:55
why cuz I love college basketball. There
00:21:57
is an undefeated team, is there not?
00:22:00
Yeah, it's really one of the best
00:22:02
stories of of all time. Tell her this So
00:22:05
for those of us that don't follow, like
00:22:06
what the what conference are they in?
00:22:09
How likely are they to win their
00:22:10
conference? And how far do you project
00:22:12
them to go in the NCAA tournament? I
00:22:14
would imagine not very far, but what do
00:22:16
you see for Miami of Ohio? Yeah, so
00:22:19
they're in the Mid-American Conference
00:22:20
and uh you know, notoriously didn't play
00:22:23
a very difficult non-conference
00:22:24
schedule. Like most teams from weaker
00:22:25
conferences, they end up playing you
00:22:28
know, one of the top teams on the road
00:22:29
just because they actually get, you
00:22:31
know, a a nice check from that team for
00:22:32
doing so. Um but Miami of Ohio did not
00:22:36
play any of those teams.
00:22:38
Debatable whether that was by choice or
00:22:39
not. Like they'll tell you they were
00:22:40
they were not able to schedule those
00:22:41
teams, you know, under the terms they
00:22:43
wanted. But uh anyway, they ended up,
00:22:45
you know, running through their entire
00:22:45
schedule undefeated, which uh
00:22:48
you you know, people are saying it's
00:22:49
entirely schedule-based, but still
00:22:50
really hard to do that even playing a
00:22:52
weak schedule.
00:22:53
They're ranked 90th in my predictive
00:22:54
rating, so uh
00:22:58
Yeah, so you know, if they played in a
00:22:59
power league, they probably would have
00:23:00
taken quite a few beatings over the
00:23:02
course of the season playing the way
00:23:03
they did. Like you look at teams at the
00:23:04
bottom of the Big East, right? And
00:23:06
they're ranked in the 90s. Like, you
00:23:07
know, that's about what what they'd
00:23:09
equate to. They're oddly like playing
00:23:11
getting ready to play their conference
00:23:12
tournament. They're not even favored to
00:23:14
win their conference tournament as an
00:23:15
undefeated team. Akron lost one game in
00:23:18
conference play to Miami at Miami, but
00:23:21
overall has, you know, better point
00:23:22
differential metrics and that's kind of
00:23:24
carrying the day in in these
00:23:25
predictions. But So, if they go
00:23:27
undefeated going into the tournament,
00:23:29
are they like
00:23:30
>> a
00:23:31
I I don't see how they they have to be a
00:23:33
top eight seed, right? If they go
00:23:35
undefeated. I don't even know if they're
00:23:36
guaranteed to make the tournament.
00:23:38
>> if they win their conference tournament,
00:23:40
by definition, they have to cuz there's
00:23:41
an automatic bid. But what Ken's saying
00:23:44
is if they lose, let's say in the
00:23:45
semi-finals or something of their they
00:23:47
could be whatever 33 and one and be left
00:23:49
out. Yeah, so if they they it's very
00:23:53
unlikely they'd be left out based on
00:23:54
what we know about how teams are
00:23:55
selected and the metrics that are being
00:23:57
used now.
00:23:58
Um so they're going to get in. But yeah,
00:24:00
seeding-wise, even if they win their
00:24:02
conference tournament, I
00:24:03
eight would be pretty optimistic, I feel
00:24:05
like. They could be an eight, but yeah,
00:24:07
probably more in that 9 10 11 range.
00:24:08
Just because their schedule I mean, they
00:24:09
have one of the five weakest schedules
00:24:11
in the country, which just put it this
00:24:12
way, like no team that's earned an
00:24:13
at-large selection has ever had a
00:24:15
schedule remotely that weak. So, they
00:24:17
are they're just like such a wild team.
00:24:20
I mean, it's first of all, amazing they
00:24:21
went undefeated. They had so many close
00:24:22
calls and every time, you know, they
00:24:24
pull it out. Um but they've also like,
00:24:27
you know, have this kind of
00:24:28
Frankenstein-like profile that nobody
00:24:30
has ever seen before. It's just a it's
00:24:32
just a wild wild story.
00:24:34
So, let me ask you another question
00:24:35
about the tournament. One of the like if
00:24:36
you think about the whether it's the
00:24:37
four factors and that each of the teams
00:24:39
having four factors or maybe it's eight
00:24:41
factors, whatever it is. Um one of the
00:24:43
things you hear all the time at the time
00:24:45
of the tournament is matchups.
00:24:48
Do matchups matter that much? Like I
00:24:51
tend to think like
00:24:53
they're talked they make great stories.
00:24:55
And they're over they're talked about a
00:24:57
lot. But I you know, whether you want to
00:24:59
call them interaction effects, if you
00:25:01
got like the you know, the loyal and
00:25:02
maramounts against the slow down teams,
00:25:05
someone's got to win and pose their will
00:25:07
on the style of the game. How much do
00:25:09
matchups matter or when you're making
00:25:11
predictions, are there matchup variables
00:25:15
in there that move the predictions what
00:25:17
we would call a significant effect size?
00:25:21
Right. So, there's Yeah, in my
00:25:23
predictions there's no matchup effects.
00:25:25
Like historically when I've looked at
00:25:26
this or other people have looked at
00:25:28
this, it's been extremely challenging to
00:25:30
find any like specific matchup effects
00:25:33
where for instance like you know, a good
00:25:35
offensive rebounding team is playing a
00:25:36
bad defensive rebounding team. Like
00:25:38
might there be some sort of extra
00:25:39
advantage there like that's an example.
00:25:41
But the trying to find those like
00:25:42
combinations has been
00:25:44
pretty much impossible. Like you watch
00:25:47
if you watch the tournament and watch
00:25:48
the commentary, like almost everybody
00:25:50
will talk about matchups. They'll talk
00:25:51
about that pace effect that you're
00:25:52
talking about when you have two
00:25:53
contrasting paces and well, it's going
00:25:55
to be really important that one team
00:25:56
controls the pace in this game. But man,
00:25:58
when you look at it historically,
00:25:59
there's almost nothing to that and
00:26:01
sometimes I just feel like
00:26:03
you know, analysts feel like they sound
00:26:04
smart talking about matchups and stuff.
00:26:06
But in the reality is like
00:26:08
it's just it's I'm not saying that those
00:26:10
effects don't exist, but it's been a
00:26:13
very challenging to tease that out of
00:26:15
the data. By the way, I we as at Wharton
00:26:18
Moneyball, myself, Audi, Shane, Tate is
00:26:21
not here today, but he we all appreciate
00:26:23
this. I we all like the way you phrased
00:26:25
it. You didn't say these effects don't
00:26:26
exist. You just said it's hard to detect
00:26:30
them. It either means a couple things.
00:26:32
One is there's not tons and tons of data
00:26:34
to do so or the effects are small and
00:26:37
therefore
00:26:38
it's hard to detect small effects. And
00:26:40
so if they exist, they might exist on a
00:26:43
you know, I always call it a second or
00:26:44
third effect size level as opposed to a
00:26:47
primary one. If I can interject, it's
00:26:49
like it could be also that these effects
00:26:51
are are kind of real at that season that
00:26:53
time, but they're ephemeral. They're not
00:26:55
like consistent over year to year that
00:26:57
like they'd show up kind of
00:26:59
in a historical analysis. This
00:27:01
discussion's kind of making me think
00:27:03
about how we retrospectively every year
00:27:04
in baseball we're like, "Oh, well,
00:27:06
relief pitching is what wins in
00:27:07
playoffs." Because, you know, that's
00:27:09
what
00:27:10
you know, won the previous year and
00:27:12
stuff like that. I think there's a lot
00:27:13
of narrative forming that after
00:27:15
retrospective narrative forming. Because
00:27:17
again, as again, as your point yeah,
00:27:19
that you know, the people we're
00:27:20
analyzing it, especially analyzing in
00:27:21
real time have to talk about something.
00:27:24
Uh and I I kind of wonder how much of
00:27:26
these are sort of us we we kind of wrap
00:27:28
up particular analysis or narrative into
00:27:30
what's currently happening, how
00:27:32
consistent that is
00:27:34
you know, year to year, next year, 10
00:27:35
years from now.
00:27:36
>> Good point.
00:27:36
>> I don't I don't think there's much to
00:27:37
that.
00:27:39
Yeah, I mean, you like the one thing for
00:27:41
me is, you know, I'll I'll talk to like
00:27:43
coaches for like pretty successful
00:27:44
coaches and they'll they'll mention, you
00:27:46
know, "Oh, this game's a favorable
00:27:47
match-up for us." And they'll explain
00:27:49
why and so, you know, you you have to
00:27:51
respect that knowledge. And I think
00:27:53
possibly like to the extent match-ups
00:27:55
exist, you know, it's it's probably one
00:27:57
of those things that swings like your
00:27:58
expected outcome by a point or two or
00:28:00
something like that. Like so, that's you
00:28:03
know, that's going to be something
00:28:03
that's really hard to detect in in
00:28:05
specific situations. But uh
00:28:08
I guess the the key point is you know,
00:28:09
match-ups don't turn a 10-point underdog
00:28:11
into a favorite. Like I think we can
00:28:12
safely say that. Right. Can you tell us
00:28:14
so, before I ask uh a last few questions
00:28:17
about the future and what you're working
00:28:19
on now, um what do you see happening in
00:28:21
this year's NCAA tournament? Like are
00:28:23
there a couple teams you really like
00:28:26
based on your model? Who are those
00:28:28
teams? Who is doing better than let's
00:28:30
say either the betting markets or you
00:28:33
know, either coaches rankings? Uh how do
00:28:35
you see things going this year?
00:28:38
Well, uh the story is really going to be
00:28:39
about the the top teams. Like last year
00:28:41
we had a notoriously
00:28:43
uh
00:28:44
upset-free almost tournament. Uh the top
00:28:48
teams were rated exceptionally strong
00:28:50
heading into the tournament and it
00:28:51
turned out that
00:28:53
it played out that way. Like the top
00:28:54
eight teams in my rating heading into
00:28:56
the tournament ended up in the Elite
00:28:57
Eight and the top four teams ended up in
00:28:59
the Final Four. So, it was kind of a
00:29:00
predictable tournament from that
00:29:02
standpoint and
00:29:04
it's a similar structure this year. Like
00:29:05
the top four teams are are really
00:29:07
strong. You know, Duke, Arizona,
00:29:09
Michigan, and Florida
00:29:11
are separating themselves from the rest
00:29:13
of the country. I doubt all four of
00:29:15
those teams get to the Final Four. Like
00:29:17
I feel like last year was an anomaly on
00:29:18
some level, but we are seeing the
00:29:19
structure of college basketball change a
00:29:21
bit where
00:29:22
there is more separation now between the
00:29:25
teams at the
00:29:26
you know, at the top of the rankings and
00:29:30
some of the teams that we
00:29:31
>> Why do you think that is? Why do you
00:29:32
think there's a more separation?
00:29:35
So, really in recent years uh
00:29:37
player rules about player movement have
00:29:39
been relaxed uh tremendously and you
00:29:42
know, it used to be it was difficult for
00:29:45
players to to change programs. Uh often
00:29:47
times they'd have to sit out a year
00:29:49
which would discourage them from doing
00:29:50
that, but those restrictions have been
00:29:51
removed and so now uh teams can
00:29:54
aggregate talent really quickly. You
00:29:56
know, if they get a star player on their
00:29:58
roster in the off season, well like that
00:29:59
might attract three or four other really
00:30:01
good players from other schools to join
00:30:02
them. And so, I think the aggregation of
00:30:04
talent is just more efficient at these
00:30:05
top programs and it uh just makes them
00:30:07
uh you know, able to produce more
00:30:08
dominant teams than than we used to see.
00:30:11
So, a lot of this a lot of obviously the
00:30:12
loosening of the transfer uh rules is a
00:30:15
big thing, but also the NIL money. I
00:30:16
mean, it's uh the top college players, I
00:30:19
mean,
00:30:20
they
00:30:21
only a few of them go to the to the NBA
00:30:24
and because because there's just aren't
00:30:26
that many openings in the NBA.
00:30:28
It it seems that with collegiate money
00:30:31
you can make almost a career or at least
00:30:33
you want to stay in the NBA in a way
00:30:34
that you never used to. And it just that
00:30:37
have any um direct or important impact
00:30:40
on the fact that now the top four,
00:30:41
eight, even really are substantially
00:30:44
better than the rest of the field
00:30:46
because of that. Is Is there Are you
00:30:47
detecting that?
00:30:49
Yeah, I think you can say that. Like
00:30:51
certainly still for the top players it
00:30:53
makes sense to to go pro. I mean
00:30:54
ultimately when you get your second NBA
00:30:56
contract you're going to make more money
00:30:57
than you can make in college. So you
00:30:59
want to get that
00:31:00
clock started.
00:31:02
Yeah, but certainly for like the top,
00:31:04
you know, five to 10 players like
00:31:06
they're
00:31:07
almost certainly
00:31:08
Um but beyond that, you know, like
00:31:10
Michigan has Yaxel Landeborg, right? He
00:31:12
could have been a first round draft pick
00:31:13
last year but he chose to stay in
00:31:14
college for another year and presumably
00:31:17
got, you know, compensated pretty well
00:31:19
by Michigan. Moussa Johnson, another
00:31:21
player on Michigan who um
00:31:23
you know, kind of falls into that boat.
00:31:25
Uh so yeah, so I'd say I'd say for like
00:31:27
players outside the top 15 or so, yeah,
00:31:29
they're more likely to to stick around
00:31:31
than they used to be and uh
00:31:33
there's no doubt college basketball more
00:31:35
talented than it's ever been. Throw in
00:31:36
the fact by the way that international
00:31:38
players now who have like professional
00:31:40
experience are allowed to play college
00:31:42
basketball. G League players who are not
00:31:44
who have not signed a contract with an
00:31:45
NBA team are allowed to play college
00:31:47
basketball. So uh everybody's looking
00:31:49
for a competitive advantage in finding
00:31:51
these guys and it's it's uh you know,
00:31:53
made made the college talent pool, you
00:31:55
know, better than it's ever been.
00:31:57
So let me just ask you a couple uh
00:32:00
complete conclusion questions if you'd
00:32:01
like. Um what has changed in your mind?
00:32:04
Like what do you think you've learned
00:32:06
about college basketball analytics over
00:32:08
the last 20 years? Like what has been
00:32:10
the biggest learning for you?
00:32:12
Yeah, that's a good question. I There's
00:32:14
obviously a lot of things and uh I'm
00:32:16
blanking on uh on all of them but uh
00:32:20
Yeah, I mean, you know, the main thing
00:32:21
is like,
00:32:23
you know, there was a time
00:32:25
when I started where it was kind of
00:32:27
fun to, you know,
00:32:30
mock conventional wisdom and you know,
00:32:33
you learn over time that like
00:32:35
conventional wisdom is
00:32:37
you know, right more often than not.
00:32:39
Like it's not always right. And there
00:32:41
are certainly things that I feel like
00:32:43
biases that coaches still have. But over
00:32:44
time like as you know, a new generation
00:32:46
of coaches have have come along and kind
00:32:48
of grown up with this stuff. You see
00:32:49
some of these
00:32:51
you know, old school fallacies go away.
00:32:53
But you know, there's no question like
00:32:55
there are things that that coaches do
00:32:58
and and that you know, are on gut
00:33:00
instinct or whatever that you know, when
00:33:02
you start to analyze them, you might
00:33:04
think they're incorrect. But
00:33:06
you know, obviously there's there's some
00:33:08
wisdom there that that coaches have
00:33:09
acquired over the years even if it's
00:33:10
just ignoring data and and and using
00:33:12
their gut instinct.
00:33:14
So, let me ask you um one last question.
00:33:16
So, I always talk about the following uh
00:33:20
uh two things.
00:33:21
Um it's a kind of a two-part question.
00:33:23
Um and Shane and Audie are going to love
00:33:25
this cuz I'm going to use my favorite
00:33:27
word is momentum.
00:33:28
How much do you think momentum
00:33:31
matters going into the NCAA tournament?
00:33:34
Like let's imagine Duke for example,
00:33:36
I'll make it up, loses in the second
00:33:38
round of the ACC tournament or Michigan
00:33:41
loses in the first or second round of
00:33:43
the Big Ten tournament. Does that change
00:33:46
anything for you? Now, of course, we
00:33:47
have to It has to do something. In other
00:33:49
words, it is recent more than past. We
00:33:51
all Anybody's model counts the current
00:33:53
power more than the past. But are you a
00:33:56
big believer in momentum or that doesn't
00:33:58
mean that much to you?
00:34:00
Oh, I think uh once you uh account for
00:34:03
what you're talking about, like
00:34:04
obviously there is more value in more
00:34:06
recent data points. Um
00:34:09
once you account for that, like
00:34:12
beyond that, I I don't have any
00:34:13
particular uh use for momentum. I mean,
00:34:16
that's one thing
00:34:19
You've got You're You're joining the
00:34:20
Shane and Audie camp. They don't believe
00:34:22
in momentum, too. Once you account for
00:34:24
all these other factors, of course. But
00:34:25
yes.
00:34:26
Yeah, so I mean, there's obviously you
00:34:28
Yeah. Um I mean,
00:34:30
there's all sorts of examples, you know,
00:34:32
of
00:34:33
momentum fooling you. Like momentum
00:34:35
exists until it doesn't, basically. You
00:34:36
know, it's it's not very predictable on
00:34:38
a a player or team level. Uh you know, I
00:34:41
always try to make the point that like
00:34:44
all data matters, okay? Like the games
00:34:46
at the beginning of the season matter.
00:34:47
Like there's it's kind of becoming vogue
00:34:48
now for people to do analysis and really
00:34:51
like
00:34:52
split the season into like small samples
00:34:54
and make teams look favorable, you know,
00:34:55
they'll find that that a team played a
00:34:56
terrible game like on January 12th and
00:34:58
it's like, well, since January 12th,
00:35:00
this has been the best team in the
00:35:01
country, you know, and it's like, well,
00:35:02
that game on January 12th actually
00:35:04
mattered and you do need to consider
00:35:05
>> [laughter]
00:35:06
>> you need to consider it. And uh so
00:35:08
so I would be like obviously teams that
00:35:10
are playing better now than they played
00:35:11
in November, that matters. Like they're
00:35:13
they're a different team in that sense,
00:35:14
but still those games in November
00:35:16
uh
00:35:17
are still data points that uh should be
00:35:19
considered and uh you know, they they
00:35:21
are considered in my model.
00:35:23
Yeah, Shane and I and probably 99% of
00:35:26
the rest of the planet would rather call
00:35:27
it non-stationarity. Maybe they're just
00:35:28
a better team or playing better. Has
00:35:30
nothing to do with momentum. This should
00:35:32
These are still college kids. I mean,
00:35:34
maybe they're not the college kids that
00:35:36
we used to think of in the back in the
00:35:38
day because they're so professionalized,
00:35:40
which means that the learning curve has
00:35:42
got to be still fairly steep for young
00:35:44
folks. So, there's nothing to prevent
00:35:46
that the idea that momentum is really
00:35:49
learning and some teams are getting
00:35:51
better because they're working to work
00:35:53
together better together better. And I
00:35:54
think in basketball in particular,
00:35:56
there's a strong sense of of um I don't
00:36:00
want to call it esprit de corps, but I
00:36:01
don't know. When two players work
00:36:03
together, their team works together,
00:36:05
they get better at anticipating each
00:36:06
other. And I can imagine that momentum
00:36:08
really appears that way um in so far as
00:36:11
it is non-stationary. Maybe it's
00:36:13
definitional, yeah, cuz I like
00:36:15
everything you describe, I would just
00:36:16
describe classify as non-stationarity.
00:36:19
Yes.
00:36:20
And momentum is more like the
00:36:23
psychological
00:36:23
>> like like like non-stationarity is like
00:36:25
a a a in some latent process that's, you
00:36:29
know, of of how they're playing the
00:36:31
game.
00:36:32
Momentum is like some extra juice
00:36:35
because of the outcome at the outcome
00:36:37
level. They keep winning and therefore I
00:36:39
don't know
00:36:40
I don't know if that's a way of
00:36:41
deconfounding these two ideas, but
00:36:44
>> one more I agree longer term and
00:36:45
state-like latent state-like and I
00:36:47
consider the other one localized and
00:36:50
based on outcomes. I think that's fair.
00:36:53
>> correlation in the recent recent
00:36:54
residual is how you define momentum or
00:36:56
something like that.
00:36:57
>> I would agree. Well, Ken
00:36:59
on behalf of myself, Adi Wyner, Shane
00:37:01
Jensen, as always, we'd like to thank
00:37:03
you for joining us here on Wharton
00:37:04
Moneyball. You can catch Ken on his
00:37:07
website kenpalm.com. He's again college
00:37:09
basketball statistician and the founder
00:37:12
of Analytics website kenpalm.com. Ken,
00:37:14
thank you again for joining us this week
00:37:16
on Wharton Moneyball.
00:37:17
All right, thanks everybody. Appreciate
00:37:18
it.
00:37:19
Welcome back to Wharton Moneyball here
00:37:21
on the Wharton Podcast Network. This is
00:37:23
Eric Bradlow, Professor of Marketing,
00:37:25
Statistics and Data Science and I'm
00:37:26
joined today by my two co-hosts, Adi
00:37:28
Wyner and Shane Jensen, both professors
00:37:30
of Statistics and Data Science. Some
00:37:32
combination of the three of us and Cade
00:37:34
Massey are here every week on Wharton
00:37:36
Moneyball. So guys, we just finished
00:37:38
with Ken Pomeroy talking about college
00:37:40
basketball. Obviously, there's a lot of
00:37:42
else lot of else going on in the
00:37:44
statistics and sports world. So we'll do
00:37:46
our normal second half of the show where
00:37:48
we say what caught your eye. Shane, I'll
00:37:50
start with you. Anything in particular?
00:37:52
Prob- It's probably going to be about
00:37:53
the World Baseball Classic, but either
00:37:54
way
00:37:55
>> It is. It is. Yeah, I woke up I woke up
00:37:57
a little bit early this morning and
00:37:59
tuned tuned into the end of the
00:38:01
Chechnya-Japan match going on
00:38:05
this morning. The starting pitcher for
00:38:07
Chechnya, this guy Andrei Satoria,
00:38:11
held Japan scoreless for 4.2 innings
00:38:14
pitched. This guy was already I I I knew
00:38:17
I'd remember this guy. This guy in 2023
00:38:19
the World Baseball Classic struck out
00:38:21
like Shohei and a bunch of Japanese
00:38:23
pitchers as well. I just think it's kind
00:38:25
of notable cuz he tops out at 80 mph.
00:38:28
So,
00:38:30
he struck out like he went through the
00:38:32
Team Japan lineup for almost like a five
00:38:35
innings.
00:38:36
>> he change speeds or does he just have
00:38:37
good location?
00:38:40
Changeup and I you know, I yeah, he's
00:38:42
got a changeup. He's got trickery. He's
00:38:45
got trickery. He's like I don't know.
00:38:48
I don't know what what the analog
00:38:50
would be, but it's pretty no I mean
00:38:52
really what's driving this is
00:38:53
unfamiliarity, right? This guy pops out
00:38:55
of nowhere for every four three years,
00:38:56
but
00:38:57
but I think it's just notable. It is
00:39:00
pretty fantastic little story that this
00:39:02
guy can kind of has dominated Japan
00:39:04
twice now. Um
00:39:06
you know, and obviously it's not not not
00:39:09
something where he could go in and pitch
00:39:10
in the I don't think he's really a major
00:39:12
league prospect despite uh Well, as
00:39:13
you're pointing out and then I want to
00:39:15
say one other thing about the baseball
00:39:16
classic before turning it over to Audi.
00:39:18
Um
00:39:19
you pointed out something. Are there
00:39:21
some pitchers that could be effective
00:39:25
once, maybe two starts in the major
00:39:27
leagues? Yeah, cuz they're not used to
00:39:29
you. They're not familiar with you. But
00:39:32
over a 162 game season, over 30 starts,
00:39:37
everything, especially now in the world
00:39:39
of AI, everyone's going to track
00:39:41
everything, everything's going to be
00:39:42
motion, they're going to know all the
00:39:44
pitches you throw, they're going to know
00:39:45
the locations and the zones you throw.
00:39:48
You would think that might be done in
00:39:49
the world of baseball classic also, but
00:39:51
maybe not as much as it would be over a
00:39:53
long season with
00:39:54
>> No, yeah, I mean I I mean even before
00:39:56
AI, I think the hitters had RI, real
00:39:58
intelligence, where they would just kind
00:39:59
of keep like like experience the same
00:40:01
pitcher many many times and kind of I
00:40:04
think there's a familiarity that even
00:40:05
like pre-analytics
00:40:07
meant that I think every pitcher coming
00:40:09
into the major leagues would have that
00:40:11
kind of curve of their stuff probably
00:40:13
seeming unfamiliar at first and then
00:40:14
hitters adjusting to it and the kind of
00:40:16
back and forth. I just When you only
00:40:18
When you top out at 80 miles per hour,
00:40:19
the sub straight to kind of have that
00:40:21
back and forth and keep the hitters
00:40:22
guessing, I think is probably limited
00:40:25
over like a 162 game season. That
00:40:27
probably is something you can only pull
00:40:29
off in a in a in a short tournament. But
00:40:31
it's still an amazing story. The USA
00:40:33
right now has, according to the betting
00:40:35
odds, a 50% chance of winning. I'm
00:40:37
ignoring the vig a little bit, but
00:40:39
they're about a minus 105. Isn't that
00:40:41
nuts?
00:40:42
>> That's nuts. Absolutely nuts. I mean,
00:40:44
it's sort of like I mean, you posted
00:40:45
some odds too of like kind of like the
00:40:49
um
00:40:50
kind of the odds for like a Major League
00:40:52
for the World Series The World Series
00:40:54
odds going are also nuts. I think it's
00:40:55
like some kind of weird like
00:40:56
unfamiliarity bias or something like
00:40:58
that. Like the fact that
00:41:01
you know, Japan I I mean, you know, the
00:41:03
US has only won like like one out of
00:41:05
four of these things that's even been
00:41:06
held. And I don't even know if they're,
00:41:09
you know,
00:41:10
Japan may be a bit more
00:41:12
kind of well-put-together team. I I'm
00:41:14
not even sure like they they they
00:41:15
probably should be favored.
00:41:17
But, you know, to kind of give them half
00:41:19
the probability is
00:41:20
>> Minus 105. Not that that
00:41:21
>> To give them half the probability is
00:41:23
disrespectful.
00:41:24
It's disrespectful to Japan and to the
00:41:26
Dominican Republic and all the other
00:41:28
teams that could kind of in a one-game
00:41:30
playoff
00:41:30
>> A one-game? Isn't it one game?
00:41:32
Isn't it one Well, yeah, each game each
00:41:35
round is one game. Yeah. Oh, come on.
00:41:37
Then Right. Right. I mean, you know,
00:41:38
it's like I might make an argument to
00:41:40
the following. The Jets could go on a
00:41:42
run. I mean, they're out now, but
00:41:43
exaggerate this, but to say
00:41:45
I might put the 2025-24
00:41:48
Los Angeles Dodgers in there, and I'm
00:41:51
not sure you should take them against
00:41:53
the rest of the field in a one-game
00:41:55
playoff at this round. Maybe. Maybe that
00:41:57
would be fine to do so. But just think
00:41:59
about if there's, you know, if there's a
00:42:00
quarterfinal round, semifinal round,
00:42:02
final round, they have to be at over 80%
00:42:05
probability to win each of those games,
00:42:06
which maybe they are cuz of, you know,
00:42:09
But I don't know. Are the Well, that
00:42:10
makes a the
00:42:11
I would have the Japan team and half of
00:42:13
the American team, you put them together
00:42:14
and you do have the Dodgers.
00:42:15
>> is that's what I was going to say. All
00:42:17
right, so never mind. That was my
00:42:18
[laughter] question.
00:42:20
I mean, one of the things that makes it
00:42:21
tricky is how much of the pitchers are
00:42:22
going to go, right? I mean, schemes
00:42:24
>> Yeah. and uh schemes
00:42:26
>> pitched like one game against Great
00:42:27
Britain and he's out. Yeah, oh well,
00:42:29
that and he's now rethinking it. I mean,
00:42:31
the thing is this is all preseason for
00:42:33
the American team and and the Dominican
00:42:35
I mean, and
00:42:36
it's so it's hard to really know. I'm I
00:42:38
think one of Mine is one of our Maybe
00:42:40
the betting odds, but they cannot be the
00:42:42
probability. No way. That is sounds like
00:42:45
a betting opportunity in my in my
00:42:47
estimate. And I don't usually
00:42:49
you know, jump towards identifying
00:42:51
those, but I it really sounds like a
00:42:53
betting Let's think about We always talk
00:42:55
about some batters, pitchers get off to
00:42:57
slow starts.
00:42:59
If we track these players
00:43:03
through the first whatever 40, 50 games
00:43:05
of the Major League of season this
00:43:07
season. What expectation do you have? Do
00:43:10
these people get off to a faster start
00:43:12
than normal?
00:43:12
>> That's what I would guess, but do you
00:43:14
think that's actually going to be true?
00:43:16
Yeah, but I
00:43:17
on the other side of it like you'd also
00:43:19
want to track like longevity cuz they're
00:43:20
playing like more intense like, you
00:43:22
know, like you know, I mean, Aaron Judge
00:43:24
is probably going to have an amazing
00:43:25
season again, but he's he's getting in
00:43:27
like very like kind of regular season
00:43:28
at-bats right now and that probably that
00:43:30
to me says somebody like him is even
00:43:33
more likely to hit the ground running
00:43:34
have like an incredible sort of sprint
00:43:37
like May, June. But come September,
00:43:40
October like does that, you know, cuz
00:43:42
already wear and tear we can see
00:43:44
affecting players when they play up the
00:43:46
regular season. Does this have any kind
00:43:49
of influence on that?
00:43:50
>> It's always exciting to see all these
00:43:51
players come play for different teams,
00:43:53
but I just want to point out It's nice
00:43:55
to be able to Aaron Judge cheer for
00:43:56
Aaron Judge Yeah, well, Aaron Aaron
00:43:58
Judge of course is is the world's
00:44:00
greatest hitter right now, but Bobby
00:44:02
Witt Jr. That guy can field. Yeah, and
00:44:05
all that and defense. It's [laughter]
00:44:07
incredible. Yeah, yeah. It's like like
00:44:09
Judge, you're like he's the best like
00:44:10
you have to I mean, he's the best in the
00:44:13
game, but you know, But Bobby Witt, not
00:44:15
all that and defense.
00:44:17
>> bad for the guy to have to have to be
00:44:18
playing in the same league
00:44:20
and the same prime as Aaron Judge and
00:44:22
comes from MVP voting, but it just to
00:44:25
you just really see it. I mean, I have
00:44:27
to say some of our earliest work Shane
00:44:28
my probably my first paper published
00:44:30
paper in baseball was a paper we wrote
00:44:32
years ago. The field and one that's
00:44:33
still my favorite paper. Yeah. And and
00:44:36
you just sometimes the eye test I mean,
00:44:39
I wish we we we knew like the the
00:44:40
probability that ball being and he did
00:44:42
two of them being caught, but it's just
00:44:45
so it's remarkable how he how he got to
00:44:47
the ball then managed to sort of leap up
00:44:49
in a sort of acrobatic motion. He's on
00:44:52
his feet and he's throwing 90 mile an
00:44:53
hour first base. I mean, WHAT?
00:44:56
>> [laughter]
00:44:57
>> JUST JUST it's just it's it's such a the
00:44:59
the beauty of the baseball is so much in
00:45:03
the in those kinds of moments. And then
00:45:05
of course, the walk-off home runs. When
00:45:07
you see these sort of these national
00:45:08
teams have a walk-off and they celebrate
00:45:11
the way
00:45:12
the way they do is just it's just a pure
00:45:14
joy and it's wonderful to participate
00:45:16
in. Just to be clear by the way to what
00:45:18
Shane said earlier, I also posted in our
00:45:20
rundown the World Series odds and
00:45:24
I agree it's equally shocking in the
00:45:26
sense that the next closest We all agree
00:45:28
the Dodgers should be the favorite, of
00:45:30
course. Of course.
00:45:30
>> But they're five times the odds of every
00:45:35
other team. Like that's the gap. That
00:45:37
just can't be
00:45:39
it's it's unreasonable. And again, you
00:45:41
you see that who who they faced off
00:45:43
against not even in that top seven or
00:45:45
eight or six or seven. It's ridiculous.
00:45:47
Those odds.
00:45:47
>> Yeah, so that I I just think you know,
00:45:49
just some of our fans they're plus 210
00:45:51
at the moment. The Yankees are plus
00:45:53
1,000, the Mariners plus 1,200, the Mets
00:45:55
plus 1,300. We're not saying they
00:45:57
shouldn't be the favorite. Maybe even
00:45:58
two to one. All right, maybe you could
00:46:00
stretch it and say three to one. Well,
00:46:01
that's the thing. I I guess it's almost
00:46:03
like a
00:46:03
a meta this season question of like what
00:46:07
is the most dominant a baseball team can
00:46:09
be? Like, you know, like, you know, I I
00:46:11
I like in in in in in in this day and
00:46:12
age. I mean, obviously what the Yankees
00:46:14
did back in the '30s
00:46:16
was as dominant as any baseball team
00:46:18
could be, but you can no longer be that
00:46:20
dominant. Like like in our modern game,
00:46:22
like what's what's the most you'd ever
00:46:24
kind of advantage
00:46:26
>> look at this, but I should have. If I
00:46:27
went back to 2000 and the Yankees were
00:46:30
going for a three-peat back then, they
00:46:32
had already won three of four. What were
00:46:34
their preseason betting odds? And it
00:46:37
might have been similar. I And let me
00:46:39
just be clear. In terms of their
00:46:42
uh gap between them and the second most
00:46:45
favorite team. It'll be It would be
00:46:46
interesting to see that. I will take a
00:46:49
look and I will post something at W
00:46:50
Money Ball. Um so, let me just say what
00:46:52
caught my eye. So, obviously guys, you
00:46:54
know I'm a huge tennis fan.
00:46:56
And
00:46:58
you know, there's definitely the big
00:46:59
two. No doubt about it.
00:47:01
I'm starting to think Alcaraz is now one
00:47:03
and Sinner is 1A.
00:47:06
This Alcaraz is just incredible. I want
00:47:09
to say it a couple of things again.
00:47:12
He's 22 years old.
00:47:14
Okay? He's got the same number of majors
00:47:17
as John McEnroe won.
00:47:19
That's seven.
00:47:21
He's got a career Grand Slam.
00:47:23
The youngest to do that.
00:47:25
He's currently started the season with
00:47:27
14 straight wins.
00:47:29
By the way, the record, if you want to
00:47:30
know what the record is, this is the
00:47:32
most impressive thing I've ever heard.
00:47:34
It's Djokovic. Take a guess how many
00:47:36
matches to to start the season Djokovic
00:47:39
won.
00:47:40
In 2011, by the way.
00:47:42
30?
00:47:43
41.
00:47:46
He won 41 consecutive matches until
00:47:49
Federer actually beat him at the French.
00:47:51
So, he went the entire season until June
00:47:55
without losing a match to start the
00:47:57
season.
00:47:58
He's won 31 straight matches on outdoor
00:48:02
hard courts. So, the hard court season.
00:48:04
Outdoor. Now, Bublik sinner does have an
00:48:06
advantage on inner indoor hard courts.
00:48:08
It's unclear why. He's won five straight
00:48:11
titles on outdoor hard courts. He has to
00:48:14
be the favorite now of every tournament
00:48:16
he goes into.
00:48:17
And I just think, you know, this is
00:48:20
something I'm going to talk about as um
00:48:22
Audi knows we're doing a if you'd like a
00:48:24
webinar on Friday with a bunch I'm not
00:48:26
going to ruin it for our webinar
00:48:28
listeners who might be listening to this
00:48:29
as well.
00:48:30
But, I've asked Chat GPT for a
00:48:32
prediction interval for Carlos Alcaraz
00:48:35
and how many majors he's going to end up
00:48:37
with in a 95% interval and I'll tell you
00:48:39
how it does it on Friday. And for those
00:48:40
listeners, we're going to post that on
00:48:42
the Wharton podcast network as well.
00:48:44
Um let me just say it's hard to imagine
00:48:48
it's hard to come up with any reasonable
00:48:49
prediction where he's not at least
00:48:51
greater than Pete Sampras. In other
00:48:54
words, the I would say, you know,
00:48:55
assuming he stays healthy enough the
00:48:57
lower bound I have with Sampras won 14
00:48:59
majors. He's 22 with seven majors. It's
00:49:03
hard to imagine a scenario where he
00:49:05
doesn't I'm not saying he's going to get
00:49:06
to Djokovic 24. I'm not going to say
00:49:09
he's going to break his record. That's
00:49:10
still 18 more majors. That's a lot of
00:49:13
majors. That's a lot of majors. But,
00:49:15
it's hard to imagine a scenario where he
00:49:17
doesn't get to 15 which makes him then
00:49:20
the winningest player of all time except
00:49:22
for the big three. And I think that's
00:49:24
his lower
00:49:25
And that Yeah, well, I mean, it's not
00:49:27
really cuz you have to have put cuz you
00:49:30
you say it's hard to imagine he could
00:49:32
have a catastrophic injury. And again,
00:49:33
I'm not wish
00:49:35
I would never wish that on anybody. But,
00:49:37
that's got to be like a small
00:49:39
probability P multiplying all this. And
00:49:42
if he, you know, I mean, if he basically
00:49:45
he could never win again type of thing.
00:49:46
And so,
00:49:47
I think I think that type of thing I
00:49:49
mean, we can't we tend not to kind of
00:49:51
think about those sort of rare events
00:49:54
type things and I I don't I don't think
00:49:55
that will happen obviously. I think it's
00:49:57
a very low probability but that's the
00:49:58
type of thing worth it. That's the
00:49:59
scenario you where you would imagine him
00:50:02
not breaking even So so let me let me
00:50:05
respond to that. I think that the the
00:50:07
You have to consider it'll fall off
00:50:09
quickly. Not all tennis players have a
00:50:11
long careers.
00:50:12
Um some of I mean back in the day they
00:50:14
fell off really quickly all the time.
00:50:15
>> Audi, he doesn't need one. He doesn't
00:50:17
need a long career. He's won all he's
00:50:18
won a half of the last major of the last
00:50:21
year. I understand that. eight more
00:50:23
is he like age-wise
00:50:26
if you compared him historically would
00:50:27
he even be at like kind of the peak? He
00:50:30
wouldn't be at the peak.
00:50:31
>> No, 25 is the peak. 25 like 27's the
00:50:34
peak for men's tennis. Yeah, but guys
00:50:35
like Borg and McEnroe they were kind of
00:50:37
done by 27. They were done by Actually,
00:50:39
Borg actually retired at 26. McEnroe
00:50:42
never won a major past the age of 25.
00:50:44
Right, so those those career
00:50:46
trajectories are possible. Um injury is
00:50:49
possible although I don't know how
00:50:50
common that is in in tennis career
00:50:52
>> Very
00:50:53
injuries are are not that rare but like
00:50:56
career-ending injuries are very rare. I
00:50:59
guess
00:51:00
probably injury slash mileage is really
00:51:02
what I'm kind of talking about. I think
00:51:03
really what's more more of a the real
00:51:05
wild card is who develops
00:51:07
to offer him incredible That's a good
00:51:09
question. That's really I mean so the
00:51:11
the big three had the big three. They
00:51:13
had to go against each other. They were
00:51:15
slightly shifted but Number one, you
00:51:17
pointed out Audi why a lot of people
00:51:18
right now right now By the way, you guys
00:51:21
remember this maybe don't remember cuz
00:51:22
I've been following it. There was the
00:51:24
big two. As a matter of fact, the record
00:51:26
for the most consecutive majors between
00:51:29
two players
00:51:31
is Federer and Nadal, 11 straight.
00:51:33
Sinner and Alcaraz are in now nine
00:51:35
straight major finals just the two of
00:51:37
them. That was until Djokovic came
00:51:39
along. So look, if it stays the big two
00:51:43
he's got a very strong chance of winning
00:51:46
the most majors.
00:51:47
>> Yes. Cuz if it comes to big three, that
00:51:50
changes or four, that changes the math
00:51:53
entirely. Mhm.
00:51:55
That's all. That's all I was going to
00:51:56
say, you know.
00:51:57
>> think that's the biggest wild card in
00:51:58
terms of prediction in the future. Among
00:52:01
them all the wild cards we've talked
00:52:02
about, that's the biggest uncertainty,
00:52:03
whether he'll have to deal with two,
00:52:06
one, or no can can challenge.
00:52:08
>> fascinating sport because I mean, Eric,
00:52:10
for what how old were you
00:52:14
like what what
00:52:15
how how much of your life have you not
00:52:17
convinced yourselves that you're
00:52:18
watching the greatest tennis player of
00:52:20
all time? Like like for almost the
00:52:21
entire adult life
00:52:24
you probably were watching somebody and
00:52:26
you're like, "That's the greatest tennis
00:52:27
player of all time."
00:52:28
>> Federer to like
00:52:30
Well, I'll even go before
00:52:31
>> Nadal to Djokovic, maybe somebody even
00:52:33
before Federer. Yeah, it was never in
00:52:35
the Borg, McEnroe, Connors era, Lendl
00:52:38
because they were all I mean, Borg was
00:52:40
better
00:52:41
but they all beat each other enough that
00:52:44
it it's clear Borg was slightly the
00:52:47
best, but
00:52:48
they each won seven, nine, 10, 11
00:52:50
majors. So, there was a there was a
00:52:52
great strength there. Yeah, so I guess
00:52:54
it's since Federer basically, right?
00:52:56
Well, then Sampras came along and
00:52:58
Sampras was just better. I mean, Agassi
00:53:01
was a close second and they they you
00:53:02
know, Agassi and and Sampras didn't have
00:53:05
that unbalanced a record against each
00:53:08
other, but Sampras was just the best. He
00:53:12
was the best of his generation. And then
00:53:14
of course
00:53:16
there was the big three and that's been
00:53:17
the rest of my adult life.
00:53:19
>> know, it's just fascinating that it's
00:53:20
been like, you know, that's like a
00:53:21
30-year period or something like that
00:53:23
where
00:53:24
at the time you thought you were
00:53:27
watching the greatest
00:53:28
tennis player of all time.
00:53:30
Yeah.
00:53:31
No, I agree. No, no, I'm I agree with
00:53:33
Audi though. I think if he doesn't get
00:53:35
to, let's say, better than Sampras, 15
00:53:38
plus, the most likely reason is two or
00:53:41
three other players come up and then he
00:53:43
wins his share but he wins one of four,
00:53:45
one of five. So maybe of the next 30
00:53:47
majors he wins five or six which is no
00:53:49
slouch. That's nothing to be ashamed of
00:53:51
but he's 30 and he wakes up and he's got
00:53:53
12 majors not 18 to 20 majors and that
00:53:56
could absolutely happen.
00:53:59
So guys maybe one last thing I wanted to
00:54:00
talk about um
00:54:02
something's happened in golf the last
00:54:04
couple weeks and I just want to put it
00:54:06
into comparison.
00:54:07
So Shane Lowry
00:54:10
and I know his last name is Berger maybe
00:54:12
Justin Berger you know they've lost the
00:54:15
last two tournaments in golf
00:54:17
leading three strokes with three to
00:54:19
play.
00:54:21
Now
00:54:22
I the reason I like to point this out
00:54:25
besides that's hard to do as a pro.
00:54:28
Tiger Woods is 51 and two
00:54:32
leading not with three to play
00:54:34
the whole fourth round to play.
00:54:38
And so these seem like extraordinarily
00:54:41
rare events that we've seen the last two
00:54:44
weeks where essentially it's been a
00:54:46
collapse by the leader
00:54:48
going into the last couple holes and
00:54:50
literally it wasn't that the other guys
00:54:51
eagled and birdied. Both these guys made
00:54:54
double bogeys, bogey like they literally
00:54:56
played the last three holes in plus
00:54:57
three and the other person played it in
00:54:59
minus one where you know it's the old
00:55:00
expression you know Ben Hogan said yeah
00:55:03
all I got to do is win the Masters is
00:55:04
par the last 11 holes. Let's see you par
00:55:06
the last 11 holes on the Masters. It's
00:55:09
not that simple but this wasn't that
00:55:11
hard. So I would just want to point out
00:55:14
Yeah. that I think we're seeing
00:55:16
something in golf where some players
00:55:19
just you know let's not take for granted
00:55:22
players that just win at this incredible
00:55:25
pace leading into the final round. It's
00:55:28
not that simple.
00:55:29
>> Oh no I I feel like that's one of the
00:55:30
most unusual ways and I mean Tiger was
00:55:32
unusual in a lot of different ways but I
00:55:34
think that was one of his most unusual
00:55:35
things is that kind of like lack of
00:55:38
like, you know, not even not even a
00:55:39
random amount of collapses or whatever
00:55:41
in the last round. I think that's really
00:55:42
kind of what you're talking about. I
00:55:43
don't know if they do any kind of kind
00:55:45
of collapse metric where it's like cuz
00:55:47
you got you know
00:55:50
you can imagine like some stati- like
00:55:52
kind of like, you know, like how often
00:55:53
is how often do how often does it happen
00:55:55
that a a golfer drops three shots in
00:55:58
three holes straight? I mean, that
00:56:00
probably the you know, all happens kind
00:56:02
of
00:56:03
throughout the week like
00:56:05
>> think I think what what what Shane is
00:56:06
asking for is a baseline. Yeah, yeah,
00:56:08
like like what you know, how
00:56:10
Like obviously we treat ones at the end
00:56:12
of the last round kind of specially
00:56:14
because they usually have more
00:56:15
consequence or at least we notice if
00:56:16
they have more consequence. But like I
00:56:18
kind of wonder what the baseline rate of
00:56:20
that kind And I guess the the other
00:56:21
question is not only the baseline, but
00:56:23
what you point out, Eric, is it's
00:56:24
happening because the the the uh the
00:56:27
player in the lead blows it. Uh I wonder
00:56:29
whether how often it it happens because
00:56:31
of or some combination thereof. They
00:56:33
blow it and they do really well. Those
00:56:35
are great questions. Um but that just
00:56:36
does lead to my question, my I guess my
00:56:39
final thought here, um
00:56:41
how um in which sports does psychology
00:56:44
really play an important role? Um
00:56:47
cracking under pressure. We know we see
00:56:49
it in penalty kicks in soccer in soccer.
00:56:52
And we I've seen it at second base in
00:56:53
Yankee Stadium enough times.
00:56:55
>> [laughter]
00:56:56
>> You know, where they get the yips,
00:56:58
right? And they can't throw anymore. Uh
00:57:00
we've seen that happen. But how and Not
00:57:02
to be up there. Golf golf
00:57:05
I think what I'm thinking about sports
00:57:06
that have a lot of it, it's kind of ones
00:57:08
that almost do have that pause and
00:57:10
action where you can kind of have you
00:57:12
know, like baseball I think is
00:57:13
you know, the ideal substrate for cuz
00:57:15
you have very high leverage leverage
00:57:16
events and there's also like the
00:57:19
you like you really get in your own head
00:57:21
because you know, there's like this long
00:57:22
stop and action between kind of Yeah,
00:57:24
the only thing opportunities. I agree,
00:57:26
Shane. The only thing I would add to
00:57:27
what Audi said is that let's remember
00:57:29
though the two players I'm talking about
00:57:33
were It's Daniel Berger, by the way. The
00:57:36
The two players I'm talking about were
00:57:38
the best players for 69 holes. So,
00:57:41
that's the other thing you have to
00:57:42
condition on. You can't just look at any
00:57:44
three-hole stretch and say, "Well, the
00:57:46
guy's plus three." That happens all the
00:57:47
time in golf. Yeah, but these were the
00:57:49
best players for 69 holes. And then all
00:57:52
of a sudden, the last three? So, that's
00:57:54
my point.
00:57:55
>> No, and I I
00:57:56
I would I would the base rate would be
00:57:57
to sort of see if there's sort of like,
00:57:59
you know, if you kind of had the base
00:58:01
rate of like any three-hole window, and
00:58:03
then you start looking at, "Oh, well,
00:58:05
what happens now if we condition on Is
00:58:07
there a different kind of rate for that
00:58:10
last day or that last afternoon?" That I
00:58:12
think That would be the kind of
00:58:13
interesting comparison. Well, guys, it's
00:58:16
been college basketball to in World
00:58:17
Baseball Classic. We talked some tennis.
00:58:19
We talked some golf. But, this is what
00:58:21
we do on Wharton Moneyball. So, on
00:58:22
behalf of myself, my colleague and
00:58:24
friend Adi Wyner, my colleague and
00:58:25
friend Shane Jensen, some combination of
00:58:27
the three of us and Tae Massie here
00:58:29
every week on Wharton Moneyball. On
00:58:31
behalf of our sound engineer and
00:58:32
producer today, Aaron Tran, on behalf of
00:58:35
Dee Patel and Marissa Reno, we'd like to
00:58:37
thank you for joining us here on the
00:58:38
Wharton Podcast Network. Between now and
00:58:40
next week, enjoy your sports, enjoy your
00:58:42
statistics. We'll see you next week here
00:58:44
on Wharton Moneyball.
