For the same two minutes every game, Yu Darvish is an awful pitcher

The ace of the Texas Rangers is very, very good. But he has a singular weakness -- and you can set your watch by it. AP Photo/Tony Gutierrez

Editor's Note: The piece was first published on May 4. Since then, Darvish's weakness hasn't gone away, but he has been shipped to L.A.

Dodgers fans, take heed.

In the first inning Saturday night, Angels leadoff man Yunel Escobar dug into the batter's box and stared out at a starting pitcher who has been, in his career, one of the worst in the game. Escobar took a fastball far outside, fouled off a center-cut fastball, then took three more balls. He jogged to first with a walk, and the Angels had a first-inning rally.

Moments later, the No. 2 hitter, Kole Calhoun, dug into the batter's box and stared out at a starting pitcher who has been, in his career, one of the game's very best. Same guy, actually, but if you believe the statistical record, those few seconds made all the difference in the world. Yu Darvish got Calhoun to fly out. Then he struck out Mike Trout for the first of 10 K's in six innings of mostly dominant pitching. Darvish and the Rangers won.

It's often said of aces that you have to get to them early. In one of the odder statistical quirks going, Darvish has exaggerated that cliché nearly beyond belief. Since he debuted in 2012, he owns the seventh-best ERA+ in baseball, better than that of David Price and Madison Bumgarner and Corey Kluber and Stephen Strasburg and Jon Lester. Only one starter in history -- Jose Fernandez -- struck out batters more frequently over a career. Darvish has made the All-Star team every season that he has been healthy. He's a bona fide ace.

But for about two minutes every night that he starts, Darvish gets lit up. No pitcher in baseball has been worse to start games, relative to the rest of his pitching, than Darvish. He allows an OPS almost 270 points higher to the first batter of the game than to everybody else. Here's how his pitching compares to the rest of his body of work:

It makes sense that Darvish -- or any pitcher -- would get hit slightly harder in the first inning, when the heart of the order is likely to bat. But it makes much less sense that the first batter of the game would crush him like this, considering how many of these leadoff hitters are Coco Crisp/Endy Chavez/Chone Figgins/Jemile Weeks types. (Along with the occasional 2012 Mike Trout or 2016 George Springer.) Last year, the league's leadoff hitters were worse than six other lineup positions.

What's more, Darvish "fixes" himself immediately. The far superior second batters of the game hit only .187/.286/.306 against him. The third and fourth batters of the game have hit .191/.292/.317.

What's going on?

Let's acknowledge, first off, that in small samples, all manner of mischief occurs. Darvish has started 106 games, so he has faced the first batter of the game 106 times. Drawing conclusions about this is like drawing conclusions about your favorite hitter based on what he has done this season: No, Ryan Zimmerman isn't this good, and no, Dansby Swanson isn't this bad, so no, Darvish probably isn't really a Double-A talent against the leadoff man. I'd still take Darvish over most pitchers against the first batter of a game.

But I'd be slightly nervous about my choice. Darvish has been struggling out of the gate since his major league debut. Facing Figgins, who was coming off a season in which he hit .188 and slugged .243, Darvish allowed a walk on four pitches, all fastballs. (It was the first of five four-pitch walks he has issued to leadoff men.)

"Nearly all of Darvish's missed targets were the result of faulty timing," pitching mechanics expert Doug Thorburn wrote after that start. "He repeatedly waited too long after foot strike to initiate trunk rotation, preventing his arm from reaching full extension at release point and generating pitches that trailed high to the arm-side. Darvish would then overcompensate with an early trigger, resulting in over-rotation of the shoulder axis and producing dirty baseballs. There were instances where Darvish lined up his delivery and painted [Mike] Napoli's target without budging the glove, with the difference between a well-timed delivery and one that was off-kilter coming down to hundredths of a second."

Darvish had a big windup that day against Figgins. He lifted his arms high over his head as he started the windup, just as he had in Japan. A week later, when he made his second start, he still had the windup, but he kept his hands down in front of him. In his third start, he was still in the windup with his hands in front of him, but he added a mid-windup hesitation, long enough to take a breath. In his fourth start, he wasn't in the windup at all but was pitching exclusively from the stretch.

"He was all over the place, but more importantly, the pace of the game was horrible," Rangers color analyst Tom Grieve said during that fourth start. "[Pitching coach Mike Maddux] said all of a sudden we put him in the stretch, you took the hesitation out of it, and he's throwing strikes."

Since then, Darvish has stayed in the stretch, but he has continued to tinker with his delivery. He has added and removed pauses and hitches in his delivery. Sometimes he starts hunched over, then after a few starts is upright. Some months, his leg kick is higher than others. Sometimes he's on the third-base side of the rubber, sometimes the first. His arm slot fluctuates from year to year:

This leads to hypothesis No. 1: Darvish tinkers so much that almost every start is a brand-new experience, and it takes him just a few pitches to find a sense of balance on the mound. Thorburn praises Darvish's pitching motion generally. But in 2012, he noted that the pitcher had been inconsistent and uncomfortable at the beginning of his starts. "Darvish appeared to be clearly out of sync, adjusting his shoulders between pitches to loosen up the joints. ... He has done a much better job of repeating the delivery overall, but Darvish still takes a while to find his timing in-game. To quote Rangers team President and former ace Nolan Ryan, 'I've seen him struggle in some early innings before falling into a rhythm.'"

This is especially noticeable on his fastball, a pitch Darvish throws for strikes less often against the first batter than overall -- and throws far more of against the first batter than overall. In one start, after Darvish fell behind 3-0 on three outside fastballs, his catcher started to throw the ball back, then stopped and made a hand motion out to Darvish. The pitcher replied with a two-handed miming of what appeared to be a tunnel, or a track, toward the strike zone. (Darvish threw the next three fastballs in the zone before walking the batter.)

That he's throwing more fastballs against these batters leads to hypothesis No. 2: What makes Darvish special is his pitch mix, a boggling array of offerings that emerge unpredictably like splatter paint. But early in the game, he's more typical and less special: He establishes the fastball and feels around for which secondary pitches are sharpest that day. "I think a lot depends on how the game is going and what's working for him," Grieve said before a recent start. "He'll settle in to probably a fewer amount of pitches, the ones that are working best at the time." When we see Darvish start a game, then, we're seeing him pawing about, trying to get his bearings for the day.

That is related, somewhat, to hypothesis No. 3: The first batter of the game has an approach perfectly suited to beating Darvish. Most batters, aware that Darvish has so many strikeout pitches, want to attack early. But leadoff hitters are trying to work deep counts, see pitches and let their teammates see pitches. They are willing to let Darvish work himself into trouble. Usually he does: Barely 40 percent of the first pitches he throws are strikes. (Overall, 63 percent of his pitches are.)

We can go back to our caveat: This phenomenon might be a fluke of sample size. We can say with near certainty that Darvish has been very bad against the first batter of the game, but we can't say that he has been bad because it was the first batter of the game or that he was predisposed to be bad against the first batter of the game. This phenomenon might stop here. He might get the next 20 leadoff men out. Which means there are two possible stories baseball is telling here: One is that one of the best pitchers in baseball can be, with just a slight crosswind, one of the worst for two minutes a night. The other is that one of the best pitchers in baseball can be totally undone for two minutes a night by complete numerical chance lacking any semblance of cause or design.

Either explanation is enough reason to keep watching and see what happens next.