0-and-2016: Sample sizes are working against Twins bullpen

The term “small sample size” gets used so often in baseball (and other areas) these days that it is in danger of losing some of its meaning, context and relevance — just as many catchy phrases or buzz words have in the past (R.I.P. whatever thinking outside the box was originally intended to actually mean).

But sample size is still a relevant thing, and I’m here to make it great again by showing how it relates to your 0-6 Minnesota Twins.

If the bullpen had a pecking order going into 2016, it probably looked like this: Glen Perkins at the top as the closer; Kevin Jepsen next, as the eighth-inning guy; Trevor May close behind as the guy who would come in to get the big outs in the 6th or 7th as needed (a role that is sometimes more important than those other two roles given that he figures to come into games with runners on base and in potentially game-defining at-bats); Casey Fien as the known commodity who would pitch in medium-leverage situations (Twins down by a couple of runs but still in a game; Twins up by a few runs but not wanting to burn May or Jepsen, etc.); Ryan Pressly as an emerging but not fully trusted arm; Fernando Abad as the lefty specialist; and Michael Tonkin as the last man standing/long guy by default/out of options and wasn’t even supposed to be here.

The first four guys are arguably the most important guys. The way hitters work counts and starters expend energy these days, six good innings is often all you can reasonably ask for from a starter. From there, it might take four relievers to finish the job.

And those first four Twins relievers have struggled to varying degrees this season. In a larger context, their failure — insomuch as one frustrating week of baseball in a six-month marathon can be labeled a failure — is a product of the Twins failing to properly judge sample sizes.

*With Perkins, the Twins took the long view. This is a pitcher who was primarily very good over the last five seasons and was virtually unhittable in the first half of 2015. He’s been named to the past three All-Star teams. But his performance plummeted in the second half of 2015 (the ugliest number being the 7.32 post-break ERA), a descent attributed to injuries.

The Twins gave him his closer role back in 2016. They might have been privately concerned with his velocity dip during spring training, but the message from Perkins and others was that he would be fine once the lights came on. Instead, he’s looked shaky in two outings — the most notable of which being Sunday’s blown save in his first opportunity. His velocity is still a few ticks below peak performance — a fastball that once routinely hit 95 mph is now settling in around 91 or 92. In counting on the old Perkins based on the large sample size instead of the small one, the Twins made their bullpen vulnerable.

*With Jepsen, the Twins took a small sample size — lights-out pitching in the final two months of the season last year, a performance that kept the Twins in the playoff chase — and perhaps overvalued it. Jepsen has been a good relief pitcher for much of his career, but there is nothing in his history that suggests he is a lights-out guy over the long haul. His career ERA is 3.67, though it is much better (2.47) in 2014 and 2015 combined. He’s a top-four bullpen arm, but his track record makes him far from a certainty to deliver clean eighth innings almost all the time.

He did deliver that clean eighth on Sunday in front of Perkins’ blown save, but his previous two outings were clunkers: entering a tie game opening day and giving up the winning run in the 9th, then turning a 3-2 lead in the eighth Friday against Kansas City into a 4-3 deficit. Did the Twins look at a small sample size and extrapolate it too much?

*May was great in the bullpen last season after a second-half move to shore up the relief group. He posted a 2.87 ERA and struck out 10.6 batters per nine innings while finding extra velocity on his fastball when he could cut it loose for 15-20 pitches instead of 100. It wasn’t folly for the Twins to come into 2016 thinking he could be a top bullpen arm, and indeed his overall numbers this season — a 1.93 ERA and 8 strikeouts in 4.2 innings — suggest dominance.

But May, too, has struggled in his two highest-leverage performances: coming into a tough spot with runners on the corners and nobody out in the seventh inning against Baltimore on Thursday, he let both inherited runners score to turn a 2-1 lead into a 3-2 deficit, then gave up his own run in the eighth. On Sunday, a bout of 10th-inning wildness gave Kansas City three free bases (two on a botched pickoff and the winning run on a wild pitch). May is the pitcher I’m actually least worried about in this whole group based on the first week, but again: his sample size is only half a season, and his role this season has an element of hope to it.

*Fien’s job is to hold down the fort, and that hasn’t happened in two of his three outings. On opening day, with the Twins scrambling because of a rain delay, he gave up two runs in one inning (the fifth) in an eventual 3-2 loss. And he turned a 4-0 Kansas City lead into a 7-0 lead on Saturday, effectively putting that game way out of reach.

Fien was a nice find for the Twins in 2012, posting a 2.06 ERA with few expectations. Since then, he’s had three nearly identical season with one big exception: His ERA (between 3.5 and 4) and innings have stayed the same (low 60s each of the past three years), while his strikeouts have gone from 73 to 51 to 41. He’s a fine guy to have around, but the Twins might be overvaluing him relative to his role. Indeed, it could be argued that Pressly (two good outings, one OK outing so far) has already swapped spots in the pecking order with Fien or is at least being used in similar situations.

The sum total is four games out of six in which better bullpen performances could have resulted in wins: opening day, which was tied in the ninth; Thursday against Baltimore, when the Twins led 2-1 in the 7th; Friday against Kansas City, when the Twins led 3-2 in the eighth; and Sunday, when the Twins led 3-1 in the ninth. You’d like to think they would win at least two of those games, maybe three. Instead, they of course won zero and are 0-6 heading into Monday’s blustery home opener.

Ah, but the last piece of this comes back to the theme, and it tilts in favor of the Twins: all of this is SO FAR this year, which is an exceedingly small sample size. It’s fair to judge the Twins’ bullpen construction solely in the context of this season, which is only 1/27th complete (if this was an NFL game, we’d be at around the first drive of the third quarter of the first game).

As such, there is still plenty of time for the Twins to be proven right about their judging of sample sizes — or at least not as wrong.