The common belief that "defense wins championships" is challenged by data, which suggests that while defense is crucial, offense is more stable, predictable, and potentially more influential in winning games and championships.
Mind Map
คลิกเพื่อขยาย
คลิกเพื่อสำรวจ Mind Map แบบอินเตอร์แอคทีฟฉบับเต็ม
Offense wins games.
>> Defense defense
>> defense wins championships.
>> This isn't just cliche. It's gospel.
>> Defense wins championship.
>> Defense wins [music] championships.
>> On its surface, this doesn't make a
whole lot of sense. Not just because
Nico [music] Harrison traded Luca Donuch.
Donuch.
>> You can imagine how surprised I was.
>> Clearly, both sides of the ball here are
important. But I think we know what
people mean when they say this. Offense
is valuable, but it's defense that is
more valuable when you get to the playoffs.
playoffs.
>> The Eagles are going to win the Super
Bowl because of their defense.
>> It's defense that gets you over that
final hump to win a Super Bowl.
>> Eagles fly in Super Bowl.
>> This isn't just folklore. There have
been several research pieces that have
furthered this belief, the importance of
defense being vital to winning a Super
Bowl. But I've always been curious which
side of the ball is more important. So I
wanted to look into it further and it
surprised me.
>> First, let's establish the obvious. A
good defense limits points. A good
offense scores points. You don't need to
venture far to find a good example of this.
this.
>> Touchdown Detroit Lions.
>> The Detroit Lions scored more points
than any other team in the 2024 regular
season. The Eagles allowed the second
fewest points in the 2024 season. The
Lions crashed out in the divisional and
the Eagles won a Super Bowl. Go
>> defense wins championships.
>> It's over.
>> But rather than just settling for last
year's result, let's widen our lens to
include a few more years. Okay, so there
are many different ways to look at this,
but let's start with the unbiased, often
worshiped nerd standard, EPA. 84% of the
top five offensive teams made the
playoffs and 64% of the top five
defensive teams made the playoffs.
Expanding it out to the top 10, similar
pattern. 75% of the time you'd make the
playoffs with a top 10 offense. Elite
and good offensive teams are more likely
to make the postseason. Let's continue
with the regular season, but look at
something slightly different. The point
of this graph is to view the predictive
power of various measurements for a
team. For example, offense. How
predictable is a team's offense based on
past performance? How predictable is a
team's offense in game three based on
the first two games? Not very
predictable. The higher we are
vertically here, the more predictable
the variable. So, when we're just two
games into the season, no one really
knows what will happen in week three.
There just isn't enough sample size.
There are frauds and teams that probably
haven't found their rhythm yet. AND THE
49ERS CLOSING THE DOOR ON THE SEAHAWKS.
>> SO, a team's offensive production
through two weeks [music] tells us
basically nothing, only about 3%. As the
season progresses, this line rises, and
that makes sense. We have more games
under our belt and more offensive data.
So, we can predict what will happen with
more certainty in the following week. If
you don't know what R 2 is here, the
Yaxis, don't worry. The gist is easy to
understand. If R square was 100% we
could say with 100% certainty we can
predict how the offense will perform
based on how the offense performed in
previous games. That obviously is not
the case. Football is chaotic and fairly
unpredictable. Now if it was 0% we would
not be able to predict anything based on
how the offense performed in previous
games. Again that is not the case. A
[snorts] great offense is likely to be
great in the following game. As we get
further into the season, the R squared
rises and that intuitively makes sense.
[music] We have a larger sample of games
played and thus a better idea of how the
offense will perform based on how they
performed in previous weeks. Again, this
is an average. If you lose Patrick
Mahomes in week 15, you can probably
throw any offensive predictive power out
the window.
Now, offense is only one element. What
about defense? First, quick shout out to
today's sponsor, Zach. Finding the right
doctor is so important and I think we
can all be guilty of maybe putting that
off. You maybe move to a new area. You
don't get anything set up. Then you get
sick and before you know it, you're late
night doom scrolling just taking the
next option you can find. Not optimal.
You want a doctor that knows you, that
fits well with you. So maybe for this
new year, you take care of that. And
that's where Zachdoc can help. You can
easily search and find the right
professional for you. Zakdoc
appointments can happen fast, typically
within 24 to 72 hours. and sometimes
same day appointments. So baby, stop
putting out the doctor appointments. Go
build a relationship with one. Click the
link down in the description or use this
QR code you can book with just a few
clicks. Thank you to Zach Do.
Okay, again, how predictable is a
defense? After 2 weeks into the season,
we're around the same predictive power
as offense. We basically have no idea
how a team will perform in week three
based on how their defense performed in
week one and two. What about mid-season?
Oh, uh, far far less than the offense.
In fact, defense becomes less predictive
as the season goes on. All the way to
about week 10. At this point, we have
about 17th the predictive power for our
defense in comparison to our offense. In
other words, how the defense performed
in previous games doesn't really tell us
If you've played fantasy football,
you probably understand this feeling.
Defense and special teams is highly
unpredictable week to week. As the
season progresses from week 10, [music]
we start to see this rise though,
meaning the defense is getting a little
bit more predictable based on how
they've performed in previous games.
Offense is much more predictable though
by season's end about four to five times
as much. Is anything else more or less
predictive than offense? Well, point
differential tracks with it pretty well
all the way until about week 12. How
about the run game? We hear all the time
that establishing the run is something
that is consistent where passing is
maybe more variable. That's not the case
here. Offensive EPA per run is actually
less predictive than defense by the end
of the season. Win percentage also
doesn't really tell us a whole lot. In
fact, it's less predictive than point
differential, meaning you can tell more
about a team [music] probably based on
their point differential than their win
percentage. Lastly, [music] let's look
specifically at passing EPA per
dropback. This is more predictive than
anything besides offensive EPA, meaning
that a team's passing game is more
predictive and stable than their
defense. I think this runs contrary to
what a lot of us maybe believe
intuitively. A defense is a stable part
of a team's game while the passing is
variable. You maybe hit some home run
shots certain games, but a lot of games
you don't. >> Intercepted.
>> Intercepted.
>> Offense is variable week to week, but
defense is something you can always rely
on. As the saying goes,
>> it's loose. The Eagles have it.
>> So, if the offense is more predictable,
does that mean it's more valuable? Not
necessarily. All we're saying here is
the defensive production is more random
week to week. In other words, if you're
convinced your team has an elite defense
around week 10, I wouldn't bank on it.
For example, the Lions and the Bucks
were both in the top five about
midseason this year. OFFICIAL SAYS LIONS
FOOTBALL that changed dramatically.
However, if your offense, especially the
passing attack, has been playing great
through 10 weeks, it's much more likely
to continue. So, our coaches here are
still safe. I mean, the Lions won a lot
of games last year. They were the one
seed in the NFC, but they fell apart in
the playoffs. While the Eagles rolled
with a great defense,
>> offense wins games, the defense wins championships.
