Why Aaron Cook And His Two Strikeouts Are A Nightmare For SabermetricsS

Aaron Cook, who's now pitching in Boston after years with the Rockies, is doing something amazing this year. Cook has pitched 29.2 innings and struck out two batters. Two. (Both came in one start against the Mariners, naturally, and one was—again, naturally—Chone Figgins.)

If Aaron Cook had struck out 10 batters in 29.2 innings, he'd have a remarkably low strikeout rate. But Aaron Cook has not struck out 10 batters in 29.2 innings. Again, he's whiffed two. His strikeout rate per nine innings is 0.61. At that pace, he won't even reach 10 strikeouts by the end of the season.

No one has had a strikeout rate as low as Cook's, in as many innings, in more than 70 years. The last guy to do it was Lou Fette in 1940. He split his time between the Boston Bees and the Brooklyn Dodgers, and he wasn't good. Only four other men ever did it—Dick Braggins, Cy Slapnicka, Ted Wingfield, and Red Peery—and all pitched before 1930, and poorly.

The funny thing is that Cook is pitching well. He has a 131 ERA+. Only five pitchers in history (with as many innings as Cook's thrown in 2012) have managed ERA+ figures that good with strikeout rates under one per nine. The most recent among them is Pete "The Polish Wizard" Jabolonski, who did it for the 1927 Cincinnati Reds. Even if we increase our strikeout threshold to two per nine innings, we'd have to go back 10 years to find someone who had success like Cook's: Andy Van Hekken, who had five starts for the 2002 Detroit Tigers.

Van Hekken was not just the last man before Cook to have a strikeout rate under two and an ERA+ of 131 or better. He was also the last man before Cook to pitch 30 innings, period, with a strikeout rate that low. It's no coincidence that Van Hekken—who never appeared in another big-league game, although he's still pitching in Korea today—achieved his feat on the 106-loss 2002 Tigers, pitching in a rotation with Steve Sparks and Jose Lima. Teams don't keep sending strikeout-averse starters out to the mound unless they've given up all hope.

Every team craves strikeouts. It's why every year, some team signs Erik Bedard, Rich Harden, Ben Sheets, Kelvim Escobar, Rich Hill, or Oliver Perez, and why teams recently paid big trade prices for Max Scherzer and Jonathan Sanchez. Guys who can't strike anyone out get fewer chances than the strikeout wizards do. That's part of why Cook—who had the second-best spring-training ERA among Red Sox starters—didn't get a chance to start until after Daniel Bard had failed. Now he's been Boston's best starter.

Among the statistical set, the strikeout has been king ever since Voros McCracken devised his controversial theory about batted balls in 1999. McCracken argued that pitchers had no control over the outcome of batted balls. Therefore, the argument went, teams should hunt for pitchers who acquired as many of their outs as possible through strikeouts. So Fielding Independent Pitching (FIP)—a crude stat that evaluates a pitcher's performance solely on home runs, walks, and strikeouts relative to innings pitched—replaced ERA as the thinking person's metric for a pitcher's performance. (xFIP, which replaces a pitcher's actual home run rate with the league average home run rate, also took off.)

McCracken's idea was brilliant and revolutionary, but it wasn't entirely right. (See this from Tom Tippett, who, coincidentally, now works for the Red Sox.)

Aaron Cook, with his boulder of a sinker, can prevent hits on balls in play better than, for example, his teammate Vicente Padilla, whose fastball is fat and straight and has been for years. Duh. But Padilla's ERA lags behind his FIP by a run; Cook's ERA is outperforming his FIP. The post-McCracken statistical consensus tells us that Cook's ERA will rise as the season goes on, and Padilla's will fall. Padilla, after all, has all those strikeouts. Cook may keep getting hitters to ground out weakly, and he may keep giving up fewer runs than Padilla. But Padilla is, sabermetrically speaking, an inherently better pitcher, even if he pitches worse.

Perhaps you imagine this is is unfair to stats-oriented writers. But then a piece like this comes along from Fangraphs seemingly every day:

Axford has also given up home runs at an insane rate this season. Axford is giving up home runs on 22.2% of his fly balls. That's the sixth worst rate among qualified relievers. The good thing is, that rate is unsustainable over a full season, and Axford has never been a homer-prone pitcher in his short career. There's a chance that some of his struggles with home runs has just been bad luck. At the same time, it's clear that Axford is not pitching like the dominant reliever we've grown to expect.

The rest of the piece proceeds along these lines. Essentially: John Axford is pitching worse this year than he ever has before, even though his strikeout rate is up, so here are a bunch of numbers that indicate he's pitching worse than he ever has before. The numbers suggest no underlying explanation for why Axford's struggled, no solution, so Fangraphs just throws its hands up, and offers something halfhearted about regression to the mean. Then they splatter more numbers on the page, and we get a little bit dumber, even though the numbers are supposed to make us smarter.

Meanwhile, Aaron Cook is pitching really well despite striking out almost no one. (He's also walking no one, which helps.) A lot of Internet stats writing has no idea what to do with that. Tellingly, Fangraphs hasn't written about Cook since his call-up, and when the site did write about Cook, when he was still in the minors, Eric Seidman wrote:

Doubront and Bard have been bright spots and shouldn't be moved, while Lester and Beckett are bound to improve their peripherals as the season progresses. ... [T]he best course of action is to use Cook as a reliever.

We now know that Aaron Cook deserves better than the bullpen. The widely consulted statistical models would tell you that's crazy, but they'd tell you, too, that Aaron Cook's whole career is crazy. Cook's career ERA is higher than his FIP and xFIP—which would ordinarily indicate that he underperformed the statistical models—but before this year, his home ballpark was Coors Field. Rockies pitchers almost always have inflated ERAs, relative to FIP and xFIP, because none of those stats takes into account the ballparks' effects.

Had Cook gotten out of Colorado earlier, he'd be a more obvious statistical marvel. Then again, the Rockies need groundball pitchers more than any other team. With his low strikeout numbers, Cook might not have even made the big leagues if he'd been drafted by another franchise.

What Aaron Cook teaches us—and what Scott Diamond, Kyle Lohse, Mark Buehrle, and Matt Harrison also teach us—is that starters can succeed without strikeouts. No shock there! But the stats-inclined sportswriter often barks, "Regression! Regression!" before examining what's really going on. We can learn from the models, sure, but we can also learn from Aaron Cook.