The Astros and Braves: How Many K’s is Just Too Much?

We know we’re going to strike out. That’s just a given with guys who have power. And we have a lot of guys who can hit the ball out of the park. And that kind of goes hand in hand. But you look at some of the studies — and our guys have looked at them — and there’s not a direct correlation with strikeouts and offense.

— Atlanta general manager Frank Wren, interviewed by Jayson Stark on 2/18/13

Through their first two games, Braves hitters have 24 Ks in 75 PAs. But they also hit six HRs, they scored 16 runs in two games, and they’re 2-0. They’re living up to expectations. Unfortunately, so are the Astros. They have 43 Ks in 97 PAs through their first three games, becoming just the fifth team in history to strike out at least 10 times in the first three games of the season. The team is 1-2 with just 8 total runs scored. So how relevant are strikeouts to a team’s success?

The 2010 Arizona Diamondbacks set the all-time record for strikeouts, and were so displeased that they got rid of all of the worst offenders, allowing Adam LaRoche to depart as a free agent at the end of the year, trading away Mark Reynolds for a pair of relievers, trading Kelly Johnson the following summer, and finally trading away Justin Upton and Chris Young in this past offseason.

I took a look at every team from 1998 to 2012, the modern expansion era. It turns out that that was an appropriate endpoint, because pre-1998 teams simply never struck out like they have in recent years. Since 1998, 50 teams have had a strikeout rate above 50% 20%. Before 1998, only one team had ever done so: the 1968 Mets, in the thick of the Year of the Pitcher. They finished 73-89 but won a World Championship the following year.

The 2010 Diamondbacks have the highest team strikeout rate of any team, ever. Their team strikeout rate was 24.7%, and it’s nearly two percentage points higher than the number-two team, the 2001 Brewers, who are at 22.8%. Right now, the Braves are at 32.0%. The Astros are at 44.3%. What, me worry?

So there are 450 teams in the sample. First, here’s a plot of team K% against runs scored, just to give a visual sense of the spread.
ks and runs
(And here’s the regression: the intercept is 1029, the coefficient of K% is -1550, adjusted R2 is .127, and the finding is significant.)

Just visually, it looks like there isn’t much correlation between team strikeouts and runs at a strikeout rate below 20%. But after 20%, the runs peter off considerably, as does the sample, because only 49 of the 450 teams were above that threshold. Still, keeping small sample size in mind, here’s a blowup of that portion of the scatterplot:
ks runs 2

The fact is, we’re in fairly uncharted waters. The 2010 Diamondbacks are the only team to crack 23%. Only six other teams have cracked 22%: the 2001 Brewers, 2012 Astros, 2012 Pirates, 2012 Athletics, 2010 Marlins, and 2008 Marlins. And their records are all over the map: the Brewers finished fourth with a 68-94 record, the Astros finished last at 55-107, the Pirates finished 79-83, the A’s were 94-68 and won the division, and the Marlins finished in third place in both years, 84-77 in 2008 and 80-82 in 2010.

So it isn’t necessarily the case that strikeouts kill you. Arguably, the 2010 Diamondbacks’ biggest problem wasn’t even the offense, it was the bullpen, which had a punishing 5.74 ERA. That fell by two runs in 2011, to a 3.71 ERA, and the Diamondbacks won their division. On the other hand, the Astros had the second-worst wRC+ in baseball last year, as well as the third-highest team K% ever, so it’s not like they’re selling out for power like the Braves are. They’re not selling out for anything. They’re just striking out a lot.

So, in conclusion, just to state the obvious: strikeouts don’t kill you, but they aren’t good. The Braves and Astros won’t continue to strike out at the same rate they are right now, but it’s a reasonable bet that they both wind up around where the Astros did last year, or the Diamondbacks did in 2010. While the Athletics proved that it isn’t impossible to go to the playoffs with that kind of performance, it certainly isn’t easy.

Baseball has changed a great deal over the years, and batters’ willingness to strike out is one of the biggest changes. Power numbers in the 1990s weren’t hugely different from the 1930s or 1950s, and the neo-deadball era of the ’80s shared a lot in common with aspects of the 1940s, 1960s, and 1970s. However, the numbers of strikeouts truly are unprecedented. It’s hard to predict from history exactly how their strikeouts will affect the fates of the Braves and Astros.

Since a homer is more good than a strikeout is bad, the Braves will be willing to live with the whiffs as long as they can make up for it with taters. But the Astros are just in for another long year.



Print This Post



Alex is a writer for FanGraphs and The Hardball Times, and is a product manager for The Washington Post. Follow him on Twitter @alexremington.


Sort by:   newest | oldest | most voted
cs3
Guest
cs3
3 years 3 months ago

“Since 1998, 50 teams have had a strikeout rate above 50%. ”

This cant possibly be correct.
Typo, right?

Entropy
Guest
Entropy
3 years 3 months ago

Based on the rest of the article, I think it was supposed to be above 20% not 50%.

Kiss my Go Nats
Guest
Kiss my Go Nats
3 years 3 months ago

Obviously, I think he may mean 20% or 25%.

Fausto Hernandez
Guest
Fausto Hernandez
3 years 3 months ago

I’m guessing it’s meant to read 20%.

Ben
Guest
Ben
3 years 3 months ago

So is the basic contention here that batters are behind the increasing strikeout rate? I think it was Dave in a recent chat that said that at some point baseball might need to look at readjusting the pitcher/hitter balance because the strikeout rate has increased so dramatically. But it may simply be a function of changing hitter approaches?

Anon21
Member
Anon21
3 years 3 months ago

Since 1998, 50 teams have had a strikeout rate above 50%.

but

The 2010 Diamondbacks have the highest team strikeout rate of any team, ever. Their team strikeout rate was 24.7%

So I take it that first threshold is meant to be, what, 20%?

macseries
Guest
macseries
3 years 3 months ago

i think he meant 20%

Jim
Guest
Jim
3 years 3 months ago

Angels fanned 36 times in losing 2 of 3 to Cincy to open the season….but since they have a lot of highly paid stars, I guess that’s ok…..

Shawn Taylor
Guest
Shawn Taylor
3 years 3 months ago

The Braves struck out 16 times last night and still crushed the Phillies. Strikeouts absolutely do not matter. I’m not sure why this is even a discussion point amongst baseball circles. The Astros suck because they have terrible players, not because they strike out a lot.

X
Guest
X
3 years 3 months ago

(semi-serious) Maybe strikeouts are actually good because you avoid grounding into too many double-plays? A team with nothing but walks, strikeouts and homers completely neutralizes the opposing team’s defense.

Hurtlockertwo
Guest
Hurtlockertwo
3 years 3 months ago

As stated, a small sample size for sure. But the fate of the Braves over the long haul may be telling, they have such high hopes for the team. It will also be interesting to see if the Dbacks are better without the strikeout guys.

Ruki Motomiya
Member
Ruki Motomiya
3 years 3 months ago

Something to consider in the sample is that the strikeouts-runs corrolation might be difficult to notice because bad hitters often strike out in addition to other things.

For example, the Astros strike out a lot, but they also just don’t have a superb offense in many other ways either, so it’s difficult to say if the Ks are the issue, a small part of the issue or not related to the issue.

marlins12
Guest
marlins12
3 years 3 months ago

Strikeouts are overrated. The Marlins don’t strike out. This offense is the worst, though.

jac
Guest
jac
3 years 3 months ago

The problem with running single variable regressions is that you are at risk of having a bunch of confounding variables. In this case, the obvious confounding variables are how many outs a team makes (OBP) and what the team does when they don’t get out (SLG), which we know are both highly correlated with scoring runs.

So, if strikeout rate is correlated with either of those (likely it seems) then most of what you could be measuring in this analysis are the impacts of those factors on runs, rather than the impact of strikeouts. That’s technical way of saying “isn’t it possible taht teams that strike out a lot are just worse at hitting in lots of (more important) ways?”

I’d like to see this regression done again with team OBP and SLG included. The coeficient on strikeout rate would then be telling you the impact of Ks on scoring for teams with the same OBP and SLG. My guess would be that the impact would not be nearly as great and possibly non-significant.

Will H.
Guest
Will H.
3 years 3 months ago

Alex, did you read the article on Ks at BTB a couple days back?: http://www.beyondtheboxscore.com/2013/4/1/4165664/how-can-strikeouts-be-great-for-pitchers-but-not-that-bad-for-hitters

Thoughts comparing how each of you went at it?

Baltar
Guest
Baltar
3 years 3 months ago

Great link, Will. Thanks.

Hank
Guest
Hank
3 years 3 months ago

Alex… a crazy thought…

Would it be worth looking at strikeout rate vs run scoring variability in a given year? Maybe the impact on total runs is minor, but I wonder if a higher K rate leads to more game to game variability?

65Kyle08
Member
65Kyle08
3 years 3 months ago

Junk article

Phrozen
Guest
Phrozen
3 years 3 months ago

Arguably, it’s a stuff article.

CB
Guest
CB
3 years 3 months ago

Junk comment

Antonio Bananas
Guest
Antonio Bananas
3 years 3 months ago

Is “junk” the new “clown”?

Bob
Guest
Bob
3 years 3 months ago

A key factor could be productive strikeouts. Obviously a guy who goes up there and whiffs at 3 pitches off the plate is nothing but bad for the team. But someone who sees 6 pitches, waiting for a pitch to drive early in the count and then maybe getting burned on a great 2-2 pitch, is a different story.

I’m thinking it might be worth a try to weight the k% rate by pitches per P/A and see if that yields anything slightly more meaningful?

Jason Roberts
Member
3 years 3 months ago

Nice work Alex.

This is really a complicated problem to unravel. First question is can a hitter be successful w/ a high K rate? The answer is yes, a low K rate is not a necessary condition for providing positive offensive value. The more complicated question is how does having a large number of high strikeout guys affect your overall run output? One theory is that lots of K’s interrupts rallies so that your total runs scored is less than the sum of its parts so to speak. You could think of this as a 9-way interaction term, but that would give you a headache. Seems like simulations might be the best way to really get at this.

J
Guest
J
3 years 3 months ago

In addition to raw runs, should we also look at runs relative to league average? Those 2010 Diamondbacks scored 713 runs, and the league average was 710.

tahititaco
Member
tahititaco
3 years 3 months ago

I hadn’t noticed that the D-back had scrapped the whole team from a few years ago…. which I thought was a good one

Youthful Enthusiast
Guest
Youthful Enthusiast
3 years 3 months ago

This isn’t scientific, but the article implies that there are two causes for strikeouts for batters.

1. You suck
2. You swing hard and don’t care about consistent contact

Strikeouts from the first reason have no redeeming value and should be avoided. I’m guessing the Astros are racking up a lot of these. Strikeouts from the second reason are a trade-off with power and there is some redemption. Someone here at FG wrote an article on how much power you have to showcase in order to sustain a high K-rate. Obviously there’s a limit to offensive production, so there has to be some limit to K-rate. At some point, you just can’t hit enough HRs to justify an astronomically (pun totally intended) K rate.

In short: It’s OK to strike out if it leads to harder hit balls. If it doesn’t, you’re striking out for no reason.

Youthful Enthusiast
Guest
Youthful Enthusiast
3 years 3 months ago

To make a sort of analogy, strikeouts are like debt. Debt/credit, in reasonable amounts, is incredibly useful. It’s essential for any business. However, with debt comes an obligation to produce enough value/earnings to repay the debt.

Strikeouts work the same way. If you’re going to take some big hacks and not shorten up on 2-strike counts, the Ks will pile up. In order to offset this, you have to balance that out with power production. If not, your career is heading for foreclosure. The burden of the high K guy is power production. When that dries up, all you’re left with are a bunch of negative value Ks.

So you can look at power guys as good debt, since they are likely to pay bake the strikeout-debt with offensive production. Low production guys are bad debt, since there is no chance they can produce the value to offset their strikeouts. If teams are the lenders of playing time, they want to lend it to good debt, rather than bad debt.

Continuing the analogy, several other commenters noted the feast vs. famine analogy. This could be extended to teams being over-leverage with Ks and eventually that bubble collapses from time to time. I remember a previous piece here at FG saying that consistent offensive production correlated with more wins than a erratic one of the same total production. This lead me to believe that High-K High power guys can be a good investiment, you should diversify with some other player types to prevent a bubble bursting. Since its easier to K than HR, the risk of a collapse is significant.

Baltar
Guest
Baltar
3 years 3 months ago

Why do you have to make up for K’s with power production? Why not with any kind of production? Admittedly, the latter might be harder to do.
I’m not trying to say this one example proves anything, but the Rays have a high K% and a good offense with park factors considered, but not HR’s.

jac
Guest
jac
3 years 3 months ago

This is a great point, and sort of what the other article linked above gets at. This is why I think it’s important to control in someway for the offensive output of the teams if what you’re interested in seeing what the actual efect of Ks are.

Slugger
Guest
Slugger
3 years 3 months ago

This is what happens when you have decreased ‘uppers.’

It’s harder to concentrate late in the game…

I bet if you looked at K ration (per inning) over the first 4 innings of the game versus the last 5 innings of a game, you’d find that there’s little to no difference between the K rates of yester-year/today within the first 5 innings of the game.

After the first 5 innings, I bet you see a huge drop off in today’s 5 – 9 inning average k/rate an inning versus yester-year.

OldDogScount
Guest
OldDogScount
3 years 2 months ago

Slice any way you want, bottom line: contact skills are still a necessary component of a *consistent* offense. Alex is dead on right about the embarrassment factor associated in the past with high whiff totals. The past 15 – 20 years that subsided in favor of robust HR totals.

I’d encourage this community to consider just maybe you guys are now in a fundamentally new era & that some of your long held beliefs as well as measurement techniques could be out of phase with the new phenomena.

Keep in mind, too, not every relationship is linear. One of the commenters challenged Alex on a suspected interaction effect among the predictor variables which is definitely a step in the right direction, but please keep in mind there are a great many important variables which are missing from your world simply because they are *not* captured into your data source: box scores.

I’m not talking about “intangibles”, but rather quantifiable elements of the inner game which for this article’s premise, do hold a tangible impact on aggregate run scoring.

For example, just as all runs generated don’t hold the same value, the same holds true for whiffs. The timing & situation of when the whiff takes place can cause a counter reaction by the defense and/or the following batter that do have an impact on team level run scoring.

I encourage this community to go deeper in it’s pre-analysis thinking before defaulting to simple single variable & linear correlation by default.

Much of baseball is non-linear and more closely resembles models of physics.

Cheers

wpDiscuz