Are the Orioles Going to Strike Out Too Much?

Yesterday, the Orioles agreed to terms with Pedro Alvarez, potentially bringing him in — though we’ve learned to not count our chickens with Baltimore signings — to add some additional left-handed power to their line-up. As August noted this morning, Alvarez is a weird fit for the Orioles, because the Orioles needed an outfielder, and Alvarez is a DH. Signing Alvarez forces Mark Trumbo to right field, where he’s terrible, the effect of weakening the team’s defense probably will cancel out most of the offensive gain Alvarez might bring at the plate, making this a non-upgrade, or at least an inconsequential one.

But there’s also another potential story with the Alvarez signing. Pedro Alvarez strikes out a lot. In that way, the Orioles are a natural fit for Alvarez, because the Orioles clearly don’t mind strikeouts. They have Chris Davis, after all, and they traded for Mark Trumbo, and most of their role players don’t make a lot of contact either. Last year, the Orioles ranked third in the majors in strikeout rate (22.2%), and with Alvarez and Trumbo now in the fold, that number is probably going up in 2016.

Of course, guys like Davis, Alvarez, and Trumbo strike out because they’re also swinging for the fences, and the lack of contact is a trade-off in order to get the power they possess. A lot of the best hitters in baseball strikeout a good amount, because the value of hitting the ball a long way more than makes up for swinging and missing a few times. The Orioles are accepting strikeouts as the cost of having power, and historically, teams that have made this trade-off have fared just fine. Jeff Sullivan wrote about the correlation between strikeouts and runs scored a few years, back when the Braves traded for Justin Upton, and confirmed the sabermetric convention that team strikeout rate doesn’t really have a negative impact on the number of runs a team scores relative to expectations.

But as I noted yesterday, I’m working on some research trying to identify what traits teams that have regularly beaten their BaseRuns records have in common. In the process of looking for these commonalities — and yes, I’ll still write up the full findings when I’m done, but it’s more work than I had anticipated — I noticed something that might tie in to the Orioles adding Alvarez to their line-up, and raises some questions about the concept that a strikeout is not much worse than any other kind of out.

As part of the research, I created index numbers for the primary rate stats we use to identify types of offensive players: BB%, K%, ISO, and BABIP. By taking a team’s performance in each of those categories and comparing it to the league average rate in that season, we can come up with something that acts like wRC+ (without the park effects) for the components. I then bucketed the 390 individual team-seasons from 2002-2014 into groups of 30, and looked at how those 13 different tiers performed in terms of wins relative to BaseRuns.

By and large, I found mostly what I expected; there were only small differences in the outcomes for teams looking just at one component of their offense. For instance, here’s the spread in wins relative to BaseRuns for the 13 tiers of teams sorted by ISO+.

ISO+ and Extra Wins
Bucket BaseRunsDiff
1 to 30 -1.4
31 to 60 -0.2
61 to 90 -0.8
91 to 120 -1.2
121 to 150 0.3
151 to 180 -0.3
181 to 210 0.8
211 to 240 0.2
241 to 270 1.6
271 to 300 -0.2
301 to 330 -0.7
331 to 360 1.0
361 to 390 0.9

The teams that hit for the most power underperformed their BaseRuns totals by a little bit, but there’s no clear trend there, and the teams that outperformed the most (+1.6 wins, on average) were in the middle of the bottom set of tiers, which doesn’t really lead to any obvious conclusions. And the tiers around that tier show no similar effect, so this looks like this could easily just be noise. And that’s what most of the data looks like, as I expected; isolating one variable of a team’s offense isn’t going to explain a huge chunk of a team’s overall record versus BaseRuns, especially since we’re only looking at the run-scoring half of the game, and wins are about runs scored and allowed.

But in the sea of small variations, there was one exception, and indeed, it did come in strikeout rate. Here’s the performance of the 13 tiers of teams, broken up by their strikeout rate relative to league average.

K+ and Extra Wins
Bucket BaseRunsDiff
1 to 30 1.6
31 to 60 -0.3
61 to 90 1.5
91 to 120 0.7
121 to 150 1.3
151 to 180 -0.7
181 to 210 1.8
211 to 240 -1.9
241 to 270 0.0
271 to 300 -0.7
301 to 330 -0.1
331 to 360 -0.6
361 to 390 -2.8

If you look at the top 12 buckets, there’s a small trend there; teams that strike out less often have won more often than BaseRuns expects, but it’s not a clear linear increase or anything, and the effect isn’t huge; you’re looking at something like an extra win per year. Strikeout rate does seem to matter a little bit, but it’s not as impactful at the high end as a lot of people have speculated, especially with the Royals receiving a lot of credit for their clutch hitting being related to avoiding strikeouts.

But look at that last line, the tier of teams that were the most egregious strikeout teams in baseball over that 2002-2014 stretch. Those 30 teams didn’t just underperform; they underperformed by a win more than the next worst bucket, averaging almost three wins fewer than expected per team. And given the gap between K+ rates between tiers, it looks like there may be some significant diminishing returns to loading up on high-K hitters.

K+ Tiers
Bucket BaseRunsDiff K+
1 to 30 1.6 0.83
31 to 60 -0.3 0.88
61 to 90 1.5 0.91
91 to 120 0.7 0.93
121 to 150 1.3 0.96
151 to 180 -0.7 0.98
181 to 210 1.8 1.00
211 to 240 -1.9 1.02
241 to 270 0.0 1.04
271 to 300 -0.7 1.06
301 to 330 -0.1 1.09
331 to 360 -0.6 1.11
361 to 390 -2.8 1.19

The next-worst tier only struck out 11 percent more than the average, but the 30 worst teams struck out 19 percent more than the average, putting them on something of an island among themselves. And it’s not like this was just one or two big outliers dragging down the average. This is what the distribution of wins relative to BaseRuns for those 30 teams looks like.

Highest K+ Tier
Team K- BaseRunsDiff
2004 Reds 1.26 10
2011 Pirates 1.16 7
2003 Reds 1.30 6
2003 Brewers 1.19 1
2006 Brewers 1.20 0
2005 Reds 1.26 0
2008 Marlins 1.26 0
2006 Pirates 1.15 -1
2007 Rangers 1.15 -1
2011 Nationals 1.17 -1
2010 Marlins 1.20 -1
2003 Cubs 1.14 -2
2008 Athletics 1.14 -2
2005 Brewers 1.15 -2
2013 Twins 1.16 -2
2006 Marlins 1.20 -3
2011 Mariners 1.15 -4
2011 Padres 1.17 -4
2008 Diamondbacks 1.19 -4
2007 Marlins 1.23 -5
2007 rays 1.23 -5
2014 Cubs 1.19 -6
2013 Astros 1.28 -6
2010 Diamondbacks 1.34 -6
2012 Astros 1.15 -7
2014 Astros 1.17 -7
2004 Brewers 1.25 -8
2008 Padres 1.15 -9
2009 Diamondbacks 1.15 -10
2002 Cubs 1.21 -14

Four of the 30 teams won more games than BaseRuns expected, and another three hit it right on the mark; the other 23 teams all underperformed to various degrees, so 76% of the highest strikeout teams won fewer games than BaseRuns suggested, and a lot of them missed by a good margin. Over this 13 year period we’re looking at, we could say that teams that struck out at extreme levels won fewer games than we’d expect based on their raw totals.

Now, please keep in mind that what I’m showing here is a correlation, not necessarily a causation. Saying that teams that struck out also underperformed doesn’t mean that striking out is why they underperformed; there could very well be a trait that strikeout rate is a proxy for that is actually driving the difference. Much like when it’s reported that people who go to college have longer life expectancies, it’s not that college is promoting better health, but that people who go to college earn higher wages, so they are more likely to be able to afford health care. The same could be true here.

But I will note that the high-K bucket doesn’t appear remarkable in any other type of way offensively. They were exactly average in BB+, ISO+, and BABIP+, and while I thought maybe the high-strikeout bucket would be poor baserunners in a way that BaseRuns wasn’t capturing, they were average by UBR as well. The only real common trait these offenses shared was striking out a lot.

It has been theorized that high-K teams could perform poorly in the clutch, and Jeff Sullivan showed some relationship between the two metrics last year; the fact that extreme strikeout clubs are also underperforming their BaseRuns records by this degree might suggest that the theory has some legs to it. Perhaps teams that stack strikeout hitters together are more vulnerable to losing close games due to their weakened ability to play for one run, or maybe high-K hitters are more prone to being exploited by elite relievers. Or maybe it’s something else entirely.

This isn’t proof of anything, and I would strongly caution against citing this link as evidence that contact is hugely important for a team. For most of the teams in baseball, strikeout rate didn’t impact their records that much, and again, we’re looking at a 13 year sample, so noise could still be an issue. But the fact that the most extreme bucket of high-strikeout teams also dramatically underperformed their BaseRuns by almost three wins a year seems like it could be evidence of an idea that has appeared true to many, and it might be that stacking a whole bunch of strikeout hitters together could lead to some problems that run estimators aren’t picking up on.

Given their own history with performance relative to BaseRuns, it’s interesting that the Orioles are perhaps setting themselves up to be a team that has to overcome this particular link in order to contend in 2016. Perhaps Alvarez and Trumbo won’t see the field enough for the team to push into this kind of tier — they’d need to run a team strikeout rate above 23% to qualify for this bucket, most likely — and the Orioles will settle in as just a normal high-K team, not an extreme one, and maybe this won’t matter at all.

But if the Orioles crash and burn this year, it might be a little less surprising now that they’re banking on so many high-K hitters to drive their offense. Teams that have made similar bets in the past decade have often ended up wondering where all their missing wins went.



Print This Post



Dave is the Managing Editor of FanGraphs.


Sort by:   newest | oldest | most voted
mtsw
Member
Member
mtsw
2 months 18 days ago

The theory here is actually fairly sound and based on accepted Sabermetric knowledge:

1) High K% batters perform slightly-but-measurably worse against high K% pitchers (as established in “The Book”) than typical players do.

2) High K% pitchers are disproportionately used in high leverage situations because elite relievers are almost universally high K% pitchers.

3) Therefore, teams with high K% will underperform their expected win totals because they will struggle (slightly) more in high leverage situations.

Maybe I’m missing something but I don’t think it needs to be more complicated than that.

rosen380
Member
rosen380
2 months 17 days ago

Last year, the top third of qualified batters by K% [highest] had a weighted average 114 wRC+; the bottom third had a 106, so I think you also have to figure in that the high K% batters are also generally better in the first place…?

Att Scotchison
Member
Att Scotchison
2 months 18 days ago

Dave, are the strikeout-heavy teams scoring fewer runs than would be expected given their offensive component stats, or are they winning fewer games than their runs scored totals would suggest? Or is it both?

chipmunknunchucks
Member
chipmunknunchucks
2 months 18 days ago

I’d be curious to cross-reference this data w/ how many double plays each “bucket” hit into and how those numbers matched projections.

One would expect high K guys to hit into fewer double plays for obvious reasons, but any sort of variance on the matter could play serious chaos w/ said expectations.

JerryF1983
Member
JerryF1983
2 months 18 days ago

College graduates have longer life expectancies not because of better health insurance, but because being a college graduate is selective for better personal decisions that impact lifespan. To name the two most obvious negative impacts on life spans, college grads are much less likely to smoke and much less likely to be obese than those without a degree.

Bingo Short
Member
Member
Bingo Short
2 months 17 days ago

Inheritance is a good personal decision, too.

Jason B
Member
Jason B
2 months 17 days ago

I am super confused as to this comment’s relevance to the article. It’s not a bad comment per se – it’s not an unhinged rant about the Phillies’ ownership group, for instance – yet still, I’m just totally bewildered.

Yanks123
Member
Yanks123
2 months 18 days ago

Could it be something with productive outs that we’re not considering? Or having extra runners on base for HRs since OBP is boosted?

Yanks123
Member
Yanks123
2 months 18 days ago

Should be fewer runners since OBP is lowered.

Twitchy
Member
Twitchy
2 months 18 days ago

Since your sample starts in 2002, I’m wondering if a fair # of the high K guys are also defensively challenged – your Adam Dunn types. Could that be an issue? Or where the missing wins come from? As in a team is so bad that the pitchers look good by FIP but bad by ERA, and baseruns would say “Hey, you should have allowed fewer runs” which is why they look good by baseruns but bad in runs allowed (and thus, underperform)?

That was my first thought – that the high K group is also poor defensively. The O’s have a great defensive IF, but that OF could be a trainwreck defensively if Kim/Jones are merely average and Trumbo is well….a guy who should never be allowed in the OF.

pdunes
Member
pdunes
2 months 18 days ago

Has someone calculated the value of either moving a runner up, sacrifice fly, or RBI ground out versus a strikeout? Or is this difference imbedded in the value of a strikeout?

It seems like the sequencing of a runner on third with one out, fly ball then strikeout should provide more value than a strikeout then fly ball. I didn’t know if this is captured at all.

MGL
Member
2 months 18 days ago

High K guys may be more likely to be poorer base runners with fewer SB. Probably are. That’s why I don’t like using BaseRuns for this kind of research. It ignores base running, SB, and on the defensive side, WP and PB. BaseRuns takes the component stats and assumes league average base running, including SB.

Does your BaseRuns formula include ROE? Most that I’ve seen do not. If it doesn’t that’s a huge factor as high K batters obviously have fewer ROE.

Also, does your BaseRuns include GDP or does it assume a league average GDP? If it uses GDP but not “productive outs” then it is over-stating high K players’ run production.

Basically BaseRuns is pretty useless for this kind of research, IMO.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 18 days ago

Per and earlier comment from dave, Fangraphs uses a BaseRuns formula that includes SB and CS.

MGL
Member
2 months 18 days ago

What about base running (not SB/CS) and GDP?

Ernie Camacho
Member
Member
Ernie Camacho
2 months 17 days ago

Most of the “intermediate” formulas I’ve seen (e.g., from Smyth or Patriot) include GDP, but not other outs on the bases or going first to third, scoring from second, and other running by runners already on base. Some also exclude ROE.

To repeat an earlier request, it would be great if Dave shared the formula Fangraphs uses so we can see if missing events explain some of the findings.

David Appelman
Admin
Member
2 months 17 days ago

A = H + BB + HBP – HR – .5 * IBB

B = (1.4 * TB – .6 * H – 3 * HR + .1 * (BB + HBP – IBB) + .9 * (SB – CS – GIDP)) * 1.1

C = AB – H + CS + GIDP

D = HR

We’re using this one, so no RBOE.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 17 days ago

Thanks David!

Hurtlocker
Member
Hurtlocker
2 months 18 days ago

Outs are not the same, putting a ball in play opens up many possibilities, striking out opens almost none. (passed ball maybe??) There’s nothing worse than watching a guy strike out with one out and a runner on third.

Hank G.
Member
Member
Hank G.
2 months 18 days ago

There’s nothing worse than watching a guy strike out with one out and a runner on third.

One out, runners on first and third (or bases loaded), guy hits into a double play. That’s worse.

Shirtless Carson Cistulli
Member
2 months 18 days ago

Slamming your toe into the door and having the nail pull back is worse than both of those.

maumannts
Member
maumannts
2 months 18 days ago

There were 1,878 double plays in the National League in 2015. There were 19,165 strikeouts. You tell me — which one is the more likely outcome in that situation?

Your chances of hitting the ball somewhere other than into a double play (and giving the runner at third the possibility of advancing) would seem a better option than giving the defense an advantage without needing to make a play.

Neil
Member
Neil
2 months 17 days ago

maumannts: Your numbers are ridiculously misleading. It looks like you’re trying to imply that a GIDP is an incredibly rare outcome. It most certainly is not.

The GIDP rate most years is about 10%; over the past 3 years, the K% has been almost exactly 20%. The groundball rate has also been around 45%, so we’re talking about somewhere between 1 in 4 and 1 in 5 groundballs being turned into double plays.

Now, I’m not sure whether this constitutes an argument for or against one approach or the other, but it certainly isn’t an easy answer.

Johnston
Member
Member
Johnston
2 months 17 days ago

“There’s nothing worse than watching a guy strike out with one out and a runner on third.”

You ever play the game? A failed squeeze bunt with the batter and runner both out is much worse.

bsball
Member
bsball
2 months 18 days ago

3 of the teams dragging down the bottom bucket are the rebuilding Astros (2012-14). Those teams were very low quality teams. It might not be surprising if their production didn’t fit a function based on MLB-quality players. How many of the other teams fit that profile?

Also you note that the highest K+ bucket has exactly average BB+, ISO+ and BABIP+. But if you have teams with high K then wouldn’t you expect them to compensate with above average performance somewhere else? Is that another suggestion that these are not quite MLB quality teams?

Only glove, no love
Member
Only glove, no love
2 months 18 days ago

This is excellent. I wonder if total bases and RISP as well as GIDP versus strikeouts might be useful. Also, the distribution of Ks within the the team may have a significant effect. For example, it looks like Theo in CHC went out of his way to drastically reduce the number of K’s overall but there are still some poweful K generators in that line up.

Danbowski
Member
Danbowski
2 months 18 days ago

Dave,
There are two questions here
– what drives the disconnect between base runs and runs?
– what drives the disconnect between run differential and wins?

I think breaking up your analysis is likely to help make it considerably more likely to find relationships. That is, if you are looking at the correlation between base runs and wins, you are introducing much more noise.

Oblarg
Member
Oblarg
2 months 18 days ago

This sure is an awful lot of stuff written about a set of data that doesn’t look convincingly different from noise.

Re-run the analysis with the data arbitrarily partitioned in half, see if the two halves don’t say completely different things (or each say nothing at all).

In the future, FanGraphs would do well to adopt a policy of “don’t publish articles on trends like this unless they were found in one data set and confirmed in a second, independent data set.” Partitioning is your friend, if you want to avoid overfitting.

wpDiscuz