There’s a new rivalry brewing in baseball. Unlike most cases of animus or intense competition between teams, it’s not based on being in the same division or geographical region. It’s built, instead, on strong starting rotations and playoff pitching, and while it doesn’t have the history of Yankees–Red Sox, it’s as engrossing right now as any feud other than Dodgers-Giants and Diamondbacks–disobedient batters’ bodies. I’m referring, of course, to the escalating long-distance battle between two of baseball’s best teams: the Detroit Tigers and the Oakland A’s.
According to FanGraphs’ playoff odds, the Tigers and A’s are the most likely World Series winners in the American League, with 17.3 percent and 14.4 percent championship probabilities, respectively. (Most of the difference stems from Detroit’s easier path past the wild-card game.) The A’s and Tigers met in the ALDS in both 2012 and 2013; both times, the series went to five games, and both times, Detroit’s Justin Verlander ended Oakland’s season with a dominant, scoreless start. It’s possible they’ll meet in the ALDS for the third consecutive season, but more likely that they’ll match up in the ALCS with the pennant on the line.
Given the odds that these clubs are on another October collision course, it’s tempting to view the rotation rearming that each did on deadline day as a literal arms race, with A’s GM Billy Beane striking first by adding Jon Lester and Tigers GM Dave Dombrowski countering later the same day by grabbing a lefty ace of his own in David Price. Beane’s post-Price-trade text to Dombrowski and A’s closer Sean Doolittle’s Twitter smack talk (a reference to Verlander’s comments earlier in the month about Oakland’s trade for Jeff Samardzija and Jason Hammel) seemed to support that interpretation.
The idea of Dombrowski and Beane trading body blows by pitcher proxy makes for a fine narrative, and a playoff matchup between baseball’s reigning super-rotations would be an even better one. Realistically, though, teams don’t make major transactions based on the possibility of playing a particular opponent in one postseason series. While playoff matchups may have been in the back of Beane’s and Dombrowski’s minds last week, they presumably made the moves they did out of a desire to improve their own rosters, not to claim or reclaim the rotation crown from another contender.
Whatever their motivations, the two accomplished GMs appear to have positioned their teams well for both the stretch run and the playoffs. As long as Gray-Lester-Samardjiza-Kazmir and Scherzer-Price-Sanchez-Porcello/Verlander stay healthy, neither the A’s nor the Tigers will have to start anyone anxiety-inducing in October, and they’ll also have the option of bumping excess starters to the bullpen. You’ve almost certainly heard, and possibly repeated, the old adage that “pitching wins pennants,” or the related axiom about the importance of aces in short postseason series. The A’s and Tigers would seem to have as much reason to subscribe to those sayings as anyone: In each of the past two seasons, the pivotal game between the two was decided by an archetypal ace beating a talented but unproven rookie. With that history as a backdrop, Rolling Stone’s Steven Lebron wrote that Beane made the Lester trade “because he knows how important pitching is in the postseason,” while Detroit Free Press columnist Jeff Seidel (among others) took the Price trade to mean that Dombrowski “is betting that starting pitching is more valuable than anything else in the playoffs.”1
Dombrowski’s actual statement was, “We thought adding [Price] to our rotation at this point gives us the best chance to [win],” which isn’t necessarily the same as attributing some special playoff multiplier to pitching. It might only have meant that an equally potent position player wasn’t available.
Here’s the problem: It’s hard to prove that those hoary bits of baseball wisdom are true. Before your #SlatePitch sense starts to tingle, let me clarify: I’m not saying that a strong pitching staff isn’t an asset. Nor, to be fair, are proponents of the “pitching wins pennants” position saying that offense and defense don’t matter. What they’re asserting, essentially, is that all else being equal, a team whose strength lies in pitching will be especially well suited for postseason play. If they’re right, then given a choice between two teams of identical quality, one of which has an average staff and a strong lineup, and the other of which has an average lineup and a strong staff, you’d be wise to take the latter.
It’s possible there’s something to the “pitching wins pennants” hypothesis, but if so, it’s hard to see it in the stats. In 2012, Colin Wyers — then the director of research at Baseball Prospectus, now a “mathematical modeler” for the Astros2 — and I looked for evidence that teams with strong no. 1 starters outperformed expectations in the playoffs. We identified the ace of each playoff team from 1995 to 2011, rated each one using a normalized measure of ace-hood, and then checked for any correlation between the strength of each ace and the difference between his team’s regular-season and postseason winning percentages. There wasn’t one, which suggests that once you know a team’s regular-season record, knowing how good its best pitcher is doesn’t add any predictive power. Nor could Colin find any evidence of an effect after rerunning the analysis using the entirety of a team’s playoff rotation instead of its ace alone.
The Astros lead the league in imaginative job titles, if nothing else.
For this article, I enlisted the aid of Russell Carleton, a Baseball Prospectus analyst whom the Astros haven’t (yet) hired, to perform a variation of the analysis Colin and I did in 2012. For each playoff series from 1969 to 2012, Russell calculated the probability (based solely on regular-season records) that each team would win a single game3 and, from there, the probability that each team would prevail in the series.4
Using Bill James’s Log5 formula.
Using the binomial cumulative distribution function, because I’m sure you were wondering.
Russell found that regular-season records were a significant predictor of postseason success.5 In other words, while playoff results are highly variable, they aren’t completely random: In any given five- or seven-game series, the better team can easily lose, but over a large sample of such series, team quality does tend to win out.
With a p-value of <.001, which signifies that there was less than a one-tenth of a percent chance that the results seemed significant by happenstance.
The next step was to see whether, for the purpose of predicting postseason results, every .600 (or .580, or .560, or .540) team is alike, or whether having good starting pitching adds an extra advantage. Russell examined the first three pitchers who started for each team in each series.6 He also looked at Game 1 starters only. No indicator of the quality of those starters (strikeout rate, walk rate, home run rate, ground ball rate, linear weights) proved to be a significant predictor of a team’s postseason success, after controlling for that club’s regular-season record.7
A team’s first three starters won’t always be its best three starters, but focusing on the first three gives us a good idea of which pitchers might make multiple trips to the mound.
The p-values for every factor but regular-season record were above .50, which (as Russell put it) “in statistical terms is ‘not even close’” to being significant.
So why doesn’t the quality of a team’s top three starters or its ace register as significant? For one thing, the differences between teams are compressed in the playoffs, relative to the regular season: Teams with terrible staffs don’t make it to October, so the gulf between the best- and worst-pitching playoff teams isn’t as stark as we’re used to seeing during the season’s first six months. Perhaps more importantly, there’s more than one way to win baseball games, and even under an expanded playoff format, teams don’t get to October without doing something well. A team with an inferior pitching staff often makes up for its weakness on the mound by being better on offense.
If there’s no clear evidence that pitching acquires extra significance in the postseason, why is the belief that it does so persistent? It might be because it’s so hard not to notice the extent to which scoring is suppressed in the playoffs. There’s no question that playoff games tend to produce fewer crooked numbers: Last season, teams scored an average of 4.17 runs per game during the regular season, but in the postseason, their output declined to 3.78 runs per game, a 9.4 percent reduction. That figure fluctuates from year to year — in 2012, teams scored 19.2 percent fewer runs per game in the playoffs — but the direction of the difference is usually the same: down.8 During the 1995-2013 wild-card era, the gap has been exactly one run per game (half a run per team), or 10.6 percent.
Usually, but not always: In 2011, scoring increased by 14 percent in the playoffs, which probably wasn’t entirely Ron Washington’s fault.
Weather explains some of that effect; playoff games can be cold, and the lower the temperature, the less far the ball flies. Defense also plays a part, since playoff teams tend to be better than average at converting balls into outs. The bulk of the decline in scoring, however, stems from the difference in the postseason pitcher pool. The top line in the table below shows the collective regular-season pitching performance of every playoff team from 1995 to 2012. The bottom line shows the collective regular-season performance of only the postseason pitchers for every playoff team from 1995 to 2012, weighted by the percentage of team playoff innings they pitched. Thus, it omits the regular-season stats of the fifth starters, swingmen, and mop-up men whom teams tend to skip over in October, when additional off days allow for extra rest.
The pitchers on a given team’s postseason pitching staff are generally about half a run better than the same team’s full regular-season staff, and teams generally score about half a run less per game in the playoffs. The postseason scoring mystery is solved: It’s not that hitters lose their mojo once the calendar flips to October, it’s that they face superior opponents.
So in a sense, pitching is better during the playoffs, in that a team’s worst arms generally aren’t invited.9 However, that doesn’t mean it’s more valuable. Every team gets to tell its fifth starter to take a hike (if only to the bullpen) and concentrate its playoff innings among its most effective pitchers. Admittedly, this benefits some teams more than others: The 2001 Diamondbacks, for instance, got to start Randy Johnson and Curt Schilling in 11 of their 17 postseason starts, which was beneficial not only because they were two of the best pitchers in baseball, but because Arizona’s alternatives were Brian Anderson and Albie Lopez.
To a lesser extent, lineups also improve, because the regulars don’t get games off.
However, there aren’t many teams with stars-and-scrubs rotations that extreme, and most clubs don’t have starters who can pitch that frequently and effectively on short rest in an era when pitchers are almost never asked to do so. Russell’s results don’t hint at a huge hidden advantage to top-heavy rotations, and since the pitch-count craze took flight in 2000, the results of short-rest postseason starts have been subpar, which suggests that teams that try to beat the system by bypassing an inferior starter at full strength in favor of an extra outing from a superior starter with diminished stuff might not be getting as big of a boost as they expect. Similarly, teams with elite late-inning relievers can try to get the same sort of advantage by working them hard, but it’s tough to know how they’ll hold up to an unusual usage pattern at the end of a long season. Mariano Riveras are as rare as Randy Johnsons.10
Given how quickly baseball’s bullpen usage has evolved, it might be too soon for statistics to say whether an unbalanced bullpen can help a team outperform its regular-season record in the playoffs. That’s another problem with trying to analyze October: It takes a long time to build up a useful sample.
If the rotations that the A’s and Tigers have pieced together give them an advantage that can’t be captured by their regular-season records, this is where it would lie. Most rotations have at least one of two weaknesses: a lack of elite arms, or a lack of depth. The A’s and Tigers don’t suffer from either affliction. Their rotations go four deep in near-top-of-the-rotation talents, which means that as long as their current starters make it to October intact — something the A’s, who always prudently plan for the worst instead of wishcasting, never take for granted — they won’t have to make the choice that the ’01 Diamondbacks did: Subject our starters to an extraordinary workload, or start Albie Lopez? In the event that the Angels overtake the A’s and force Oakland into the wild-card game, having a fourth top-of-the-rotation starter will give Oakland better odds of avoiding a Jason Hammel start in a single-elimination game. And once the real playoff rounds start, the A’s can keep their arms on regular rest without compromising on quality, minimizing their exposure to known risks.
This is also where we bump up against the limits of postseason stats, however. We can come up with other examples of teams that had A’s/Tigers-esque rotation depth — the 1971 Orioles, the 1997 Braves, the 2010 Giants — but not enough to say, based on prior results, that the key to postseason success is a strong fourth starter. In fact, the better the top three starters on a staff, the less likely it is that the fourth starter will play a pivotal role. When Games 1, 2, and 3 of a five-game series are started by pitchers named Maddux, Glavine, and Smoltz, Denny Neagle might not even be needed.
Because October baseball subjects fans to a disquieting combination of disproportionate importance and exceptional unpredictability, it’s a fertile breeding ground for suspect narratives that attempt to explain small-sample postseason success or failure. Over the next few months, you might hear, for instance, that teams that “back into the playoffs” after a September slump are at a disadvantage against teams that end the regular season on a high note. Not so. You might be told that teams that rely on the home run can’t score in the playoffs, when small ball rules. In fact, the opposite is the case. Surely momentum matters? Uh–uh. And we all know that there’s no substitute for postseason experience — except for a lack of postseason experience, which works just as well. Even skilled prognosticators have failed to crack the postseason code: In 2006, Nate Silver introduced the concept of a Secret Sauce that could sense latent postseason potential, but BP retired it four years later, after it failed to prove predictive. Compared to predicting the playoffs, predicting presidential elections is a piece of cake.
Maybe our mistake isn’t failing to discover the secret to postseason success; maybe it’s presuming that there is one. October baseball has more off days, but at its core, it’s still the same game, with the same goal: outscore the other guys. And no one’s seeking the secret to winning in April or August.
To be clear, the near impossibility of proving that pitchers have magical postseason powers isn’t an indictment of either Oakland’s addition of Lester or Detroit’s pursuit of Price. Both teams are probably better equipped to win with their new starters than they were without them, whether in the regular season or the World Series. Baseball Prospectus ran both teams’ playoff odds before and after adjusting their depth charts to reflect their July 31 trade activity, and according to BP’s simulations, the Tigers and A’s upped their odds of winning the World Series more than any other teams did on deadline day.
As far as those simulations are concerned, though, that’s not because Lester and Price are good pitchers; it’s because they’re good players. If Beane or Dombrowski had upgraded his roster by adding an equally valuable bat instead of a starter, the outlook for his team, statistically speaking, would be either identical today or close enough to it that we’d have trouble detecting the difference. Saving a run simply isn’t much more important than scoring one, even in October.
By constructing two of the strongest rosters in baseball — even before turning good rotations into great ones — Beane and Dombrowski did most of what they could to give their clubs a leg up in October. And by trading for additional aces, they ensured that if there is a slim distinction between a “good team” and a “good playoff team,” their rosters will end up on the right side of the divide. Even so, it’s possible the playoffs will have the last laugh. Four years ago, as the Lincecum-Cain-Sanchez-Bumgarner Giants prepared to face the Halladay-Oswalt-Hamels Phillies, Jim Palmer, the ace of those ’71 Orioles, was asked about the importance of playoff pitching. “It’s a great place to start, but to win, it has to be more than that,” Palmer said. “While you would think it would favor Philadelphia, it’s not as big an edge as you would think because of the fact that the Giants can pitch, too.”
That’s the problem with the playoffs: A team can spend a season assembling a super-rotation, but once it gets to October, it might find that its opponent has a super-rotation, too.