You may have noticed I
don't talk about bullpens or relief pitchers very much. If you have, then
congratulations - you're the only person who has read multiple posts on this
blog.
But seriously, I looked
back. In 18 career blog posts to date, I have mentioned relief pitchers
exactly two times. In one of those instances I made a comparison between Mariano Rivera* and
Edgar Martinez. And we discovered that, over their careers, Edgar
participated in 78% more Plate Appearances (PA) than Rivera
had Batters Faced (BF). Yes, indeed. Edgar Martinez - the guy who
wasn't a major league regular until age 27 - stepped into the batter's box
almost twice as often as Rivera faced a new hitter.
Relief pitchers - even
ones than play for 20 seasons, like Rivera - just don't end up totaling a lot
of innings. They get a surprisingly small amount of participation over
the course of the season.
In that context, it's
not too surprising to learn the truth... which is this:
Relief pitchers just
aren't THAT important, statistically.
Now, this makes me sad. Every former player (even if you stopped playing in Little League) knows
that RP's are almost always the best personalities of the team:
I love RP's! But I
also understand how the analyst world evaluates the true value of RP's. And the bottom line is, RP's - even a whole bench full of them -
can't really affect a season THAT much.
Let's demonstrate this
more rigorously. Today we sit on June 14th, approximately 67 games through
the 162-game season. We are 41% through the MLB season.
Below I've summarized
the 2014 Relief Pitcher Fangraphs WAR (link) totals for each team in the MLB, as of
today:
In the right column,
I've projected year-end stats in a linear fashion. This is
sabermetrically sacrilegious. By projecting in a linear fashion, I am completely
dodging mean regression. You can't do that. More on that in a
moment.
First, note that median
team in 2014 should only expect about 2.3 WAR from their RP's. Remember, an MLB starting position player tends to average about 2.0 WAR. Above-average starting position players create significantly more than
+2.0 WAR. So, the typical MLB team gets about as much marginal
value from their entire bullpen as they do with just one "ho-hum"
starting position player.
Note also that the very,
very best, elite bullpen (if liberally projected to +8 WAR) should be
approximately as valuable on the margin as an MVP candidate position player. (See here for WAR grid, and see here for 2013's team bullpen WAR
leaders.) That sounds impressive! But... note that I could rewrite
the first sentence of this paragraph as: A single great position
player can be as valuable as an entire bench of arms - even the very best
ones.
Now it doesn't sound
quite as impressive.
And again, remember that
the +8.0 WAR number above is a linear projection, and assumes that the
MLB's best bullpen to date in 2014, stays that way, and does not revert to a
more typical level of performance.
In summary: Relief pitchers can push a very good team over the top to
greatness. But they can't make a bad team average, or an average team
good.
Turning our focus back
to the Mariners: The Mariners
relief pitching has been cited as being a big positive for this 2014 team.
Perhaps it's been a reason they have been modestly successful, over all, this
season.
What does the data say?
It says that the Mariners bullpen has been worth 1.9 WAR so far this
season. That is 9th best out of 30 MLB teams. If we continue at
this pace (linear projection - unrealistic), then the Mariners bullpen will add
about 4 wins to this year's squad, relative to a replacement-caliber squad. Not bad. But probably not hugely impactful, either.
Well, bad news. +4
WAR from the 'pen isn't happening this season. Why? Because the
team's performance will very likely regress to the mean.
I alluded to mean
regression above, but let me explain it in more detail now: We should
expect the Mariners bullpen to be less lucky in the next
portion of this season.
Why? What if they
are truly this talented? After all, +1.9 WAR doesn't
seem like a lot. Why can't they keep up that pace?
They could. But we
definitely should not expect that to happen. Here is
why.
Below are summary stats
for the 2014 Mariners bullpen at the moment. These are MLB ranks:
- #4 in ERA - tells us they
aren't allowing a lot of runs, if we "back-out" the number of
errors
- #9 in FIP - tells us they
aren't allowing a lot of runs, if we adjust their numbers to the
league-average fielding
- #6 in xFIP - tells us they
aren't allowing a lot of runs, if we adjust to league-average
fielding, and correct for HR/FB rate
These are really good rankings. But it's important to note that only one of these metrics
accounts for any degree of "luck" - the xFIP, which corrects for
HR/FB%. (An outlier HR/FB% is thought to be a random statistical anomaly
that can greatly affect performance outcomes - so xFIP corrects for it.)
None of these metrics
correct for another type of luck - the percentage of inherited
baserunners stranded: Left On Base Percentage, or LOB%.
LOB% is thought to be
highly interdependent with sequencing, i.e. the lucky/unlucky timing of batted
ball event sequences. Remember my analogy I made here, about an extreme example where all
doubles happened to occur with the bases loaded. Sequencing can wildly
affect outcomes, all the way up to W-L record.
Well, where do the
Mariners rank on LOB%? They are #1 in the MLB - the highest percentage. The Mariners are stranding 81% of inherited base runners, versus an
MLB-average of 74.3%.
Let's play devil's
advocate. In order to believe that the Mariners will maintain an LOB% of
81%, you need to believe either (a) the Mariners are more "clutch"
than the rest of the league, (b) the Mariners are lucky or (c) some combination
of the two.
**takes Analyst hat off
for one minute**
My personal opinion is
that there is indeed some innate "clutch factor" in
existence. This is where I fundamentally disagree with a number of
sabermetricians, including this one.
But, the fact that the
serious analysis on this topic has yielded mixed results, makes me believe
that randomness and chance play a much, much larger role than
"clutchness", especially in the long run.
So, with MLB's #1 LOB%,
the Mariners RP's have most likely been quite lucky so far this season.
We should expect them to perform worse, based on the "luck"
factor... all else equal.
Subtle, but important
point - I am NOT saying that the Mariners bullpen is bad. Since I can't
compare the magnitude of the "clutch" factor versus the "luck
factor," I can't tell you exactly how lucky they've
been or exactly how talented they are. It's certainly
conceivable that, stripping out all the luck, the Mariners bullpen is still far above-average.
But keep in mind, even
at +4 WAR pace (unlikely to be sustained), the Mariners bullpen simply doesn't
get on the field enough to matter, that much. Relief pitchers just aren't
THAT important, relative to other players.
*Side note:
Mariano Rivera had an incredible career, but his book throwing former
teammate Robinson Cano under the bus is an embarrassment. I was very excited Cano
played it off. He's a chill dude... which is probably what bothers people
when they realize that, statistically, Cano is elite. He compares very
favorably to Derek Jeter: link
No comments:
Post a Comment