It’s been quite a week for both the New York Mets and Washington Nationals, though the two teams have seen very different results. The Mets, currently riding a five-game win streak that included a sweep of the Nats, are back in first place in the National League East for the first time since early June, despite the returns for the Nationals of injured starters Ryan Zimmerman and Jayson Werth. Many factors allowed the Mets to sweep the Nats this weekend, including their fireballing young trio of Matt Harvey, Jacob deGrom, and Noah Syndergaard, the overall – and sudden – ineptitude of the Nats’ offense (see figure below, courtesy of ESPN, for an overview), the addition of two-time Home Run Derby champion and 4.2 WAR outfielder Yoenis Cespedes to a suddenly-deep Mets lineup, and the magma-hot bat of Mets first baseman Lucas Duda, who had an incredible run that saw nine home runs hit in an eight-game span, and was awarded co-National League Player of the Week for his efforts.
Washington’s offense needs to pick it up if it plans to get back into first place.
But the most frustrating cause of the Nats’ defeat for sabermetricians – and Nats’ fans – was the mistake-filled performance put on by their manager, Matt Williams. His first mistake was not electing to bat Bryce Harper, the best hitter in the National League, second, which could have helped at least marginally. He also elected not to use either Drew Storen or Jonathan Papelbon the entire series, for phantom reasons mysterious to even Albus Dumbledore (their ERAs of 1.56 and 1.69, respectively, are the best among qualifying pitchers on the Nationals this season).
The most instantly apparent mistake by Williams, however, came in the second game of the series on Saturday night, in the eighth inning of a 2-2 ballgame. “Lefty specialist” Matt Thornton was on the mound, and righty Mets newcomer Yoenis Cespedes was on his way to the plate with a man on second and one out. Williams elected to intentionally walk Cespedes to face Duda, with the reasoning, as stated postgame, that “we have confidence in [Thornton] every time he faces a lefty.” His decision came from a spot of conventional baseball wisdom: with first base open, he could walk Cespedes, one of the league’s top hitters, to get to a lefty-lefty matchup, and even give himself a chance for a double play.
There were numerous flaws in Williams’ decision, that he (or any other member of his coaching staff) could have found if he’d done the research that should be necessary for someone who’s being paid to manage a major league ballclub.
With a run in scoring position at second base, any hit would give the Mets a lead. Though Matt Williams considers Thornton a lefty specialist, he’s allowed hits on 10 of 51 at-bats vs. lefties this season (a .196 BAA) compared to nine of 44 against righties (a .205 BAA). A quick statistical t-test shows that his success rate against lefties isn’t statistically significantly better than that against righties this season. So, for the purpose of not giving up a hit, which was the goal in this situation so as not to give the Mets the lead, Thornton is not a lefty specialist at all.
But even if Thornton were a lefty specialist, pitching to Duda wouldn’t be preferable to pitching to Cespedes. Though he’s a lefty, Lucas Duda is actually hitting .307 against lefty pitchers this year, compared to .226 against righties. This is rare among MLB batters (hence the conventional wisdom), but amazingly, Cespedes embodies the same phenomenon: the Mets’ new outfielder is hitting just .184 against lefties, compared to .321 against righties, this season, despite being a righty himself.
Clearly, Matt Williams didn’t know, or simply chose to ignore, these numbers. But there’s still the issue of intentionally giving the Mets an extra baserunner with only one out, perhaps in hopes for a double play. A quick glance at TangoTiger’s run expectancy matrix – predicting the number of runs scored in each of 24 possible situations based on the occupancy of the bases and the number of outs – based on 24 years of data, shows that historically, the expected number of runs given up with first-and-second, one out, is .902, compared to .678 for just a man on second.
So, walking Cespedes was risky to begin with, even if Thornton did have a higher probability of getting Duda out than Cespedes – which is the opposite of what the numbers suggest.
And even with this overwhelming evidence against Williams’ choice, there’s still the fact that, whether or not you believe in a “hot hand,” the Nats decided to walk a man who was 0-for-3, and feeling the jitters of his Mets debut in front of a raucous crowd of fans at Citi Field, for a man who not only had already hit two home runs that night, but had eight dingers in his past seven games – in fact, each of his last eight hits had left the ballpark! Was it possible Williams was completely ignorant of Duda’s hot streak, or did he again choose to ignore the information?
No matter how you slice it, Matt Williams made a hugely illogical and truly indefensible decision in that fateful eighth inning, and like poetic justice, Duda shot an RBI double into the gap into left center field to give the Mets a lead they wouldn’t relinquish. According to Greg Stoll’s win expectancy model, the Nats had a 30.6% chance of winning before the walk, and an 8.3% chance of winning after the double – the decision wasn’t just the wrong one, but it was in an extremely high-leverage situation, and may have cost the Nationals one of the most important games of the season.
by Derek Reifer, Northwestern University