In 2015, one of the most enduring memories of the postseason was manager Terry Collins allowing a tiring Matt Harvey to pitch into the ninth inning of World Series Game 5, turning a two-run Mets lead into a Series-ending 7-2 loss in 12 innings to the Royals, and later regretting it.
"I know better than that," Collins said about not calling in closer Jeurys Familia.
In 2016, aside from the Cubs ending their century-plus stretch without a title, you remember Terry Francona (among others) rewriting the rules of bullpen usage to help push the Indians to the World Series, bringing in bullpen ace Andrew Miller whenever needed, sometimes as early as the fifth inning. All told, Cleveland relievers threw 64 2/3 innings, nearly as many as the starters did (69 1/3). It was, in many ways, the tipping point of how pitchers are used to win the biggest games.
Because of that, in the 2017 playoffs, you're going to see the fully-realized evolution of postseason pitcher usage at a new level. Get used to the terms "bullpenning," "piggybacking" and "times through the order penalty," and forget the terms "starter" and "reliever." Postseason baseball often doesn't resemble regular-season baseball, and that's never going to be more true than this October. You'll see short starts by big names, but they won't be by accident. They'll be by design.

This idea -- that pitchers get less effective the deeper they go into games -- has been around for years in some circles, dating back at least to Grady Little's infamous decision to stick with Pedro Martinez in Game 7 of the 2003 American League Championship Series and probably well before that. But it took years for the on-field strategy to truly adapt, and it wasn't until last year that we really saw it come to life in October. The Harvey game was less than two years ago, yet it feels like another lifetime entirely.
Let's take a look at each postseason since the advent of divisional play in 1969, and track the percentage of batters who were facing a pitcher for the third time. For decades, the number hovered between 15 percent and 20 percent, with an average of 16.2 percent. In 2015, for example, 15.9 percent of hitters were facing a pitcher for the third time, essentially the same as in 1973, when it was 15.7 percent. Through 2015, it had never been below 14 percent. Last year? It dropped below 10 percent.

That's a stunning drop in a single year, and it's because the rules of postseason baseball are fundamentally different. In addition to one game of a five- or seven-game series simply mattering more than one game of 162 does, the playoffs come with far more days off than the regular season does. Managers don't really need to worry about length or preserving their bullpens, so much; they just need to worry about 27 outs, and how best to get them.
As time goes on, it's become increasingly clear that the best way to get those outs is not to force a starter to turn over a lineup multiple times, simply because "that's how it's always been done." It's to find the pitcher best equipped to get the out, and the numbers are clear that a fresh reliever is a far better choice than a tiring starter.
2017 Major League times through order outcomes
Starting pitcher, First time:    .314 wOBA   22.5 percent strikeout rate
Starting pitcher, Second time:   .332 wOBA   20.3 percent strikeout rate
Starting pitcher, Third time:    .339 wOBA   18.4 percent strikeout rate

Relief pitcher, First time:       .309 wOBA    23.5 percent strikeout rate
The more times a pitcher turns over the lineup, the less effective he gets. That's due to fatigue (relievers throw 1.4 mph harder on average than starters), due to hitter familiarity, due to worsening platoon splits and other reasons. Put another way, hitters facing a starter the third time produce like Robinson Cano (.338 wOBA), but hitters facing a reliever for the first time are more like Jordy Mercer (.308). Why wouldn't you make that trade, as a manager?

While we're talking about the postseason here, we've seen the beginnings of this happening in the regular season, too, particularly with the Astros and Dodgers -- no team has asked their pitchers to face fewer hitters a third time than the Dodgers have. The math just isn't in their favor.
So how can this year's teams put this into motion? The Nationals aren't going to yank Max Scherzer in the third inning, nor will the Red Sox do the same with Chris Sale, though even they aren't immune. There are always exceptions. What we predict may happen instead is that some of baseball's more progressive staffs could go with something like two "traditional starters," four "tandem starters," two "long relievers" and four "relievers," though the old-school Dusty Baker may challenge that in Washington.
Take the Astros, for example. Justin Verlander and Dallas Keuchel will likely start the first two games of the AL Division Series, and they'll be expected to get through at least five or six innings, but they'll likely be the only Houston pitchers who are. In Game 3, for example, they could potentially tandem three innings from Lance McCullers (.251 wOBA the first time through this year, .447 the third) with three from Brad Peacock (.185 first time, .486 third). Then in Game 4, imagine three from Charlie Morton (.255 first, .331 third) followed by Joe Musgrove (.243 as a reliever, after allowing .372 as a starter), Chris Devenski (.256 wOBA the first time), Mike Fiers or Collin McHugh, or so on.

The Dodgers can do it, throwing Clayton Kershaw and Yu Darvish the first two games, then letting various combinations of Rich Hill, Alex Wood, Kenta Maeda, Hyun-Jin Ryu and Brandon McCarthy handle the next. The D-backs, if they get past the National League Wild Card Game, can have Zack Greinke and Robbie Ray as "starters," and then throw all sorts of Patrick Corbin, Zack Godley, Taijuan Walker and Archie Bradley at you. The Indians are getting ready for another shot at it, too, announcing that starter Mike Clevinger, who has a 2.22 ERA since Aug. 1, will shift to the bullpen. It's not a demotion. He'll get big innings, just in a different way.
With a non-elite starter, there's no reason to push him through a lineup more than twice. Even with an elite starter, as Harvey was in 2015, asking him to do it a fourth time mostly doesn't make sense -- making it a fair question if Madison Bumgarner's complete game in last year's NL Wild Card Game, the only nine-inning complete game of 2016, will be the last we see for some time.
We'll get our first taste of this in the AL Wild Card Game on Oct. 3, which looks likely to feature the Twins throwing a good-but-not-elite starter in Ervin Santana against the Yankees. Santana has made it through five innings in 30 of his 32 starts this year, but like most pitchers, he's less effective after the first time through, seeing his 2017 strikeout rate drop from 20.8 percent to 18.4 percent, and his wOBA against jumping from .267 to .338. His career numbers are similar.
When Santana is lifted earlier than he's been in nearly any other game this year, it may seem like he didn't step up in a must-win game. But remember, postseason baseball isn't regular-season baseball. Having a starter yanked early isn't a bug; it's a feature -- and as we saw last year, it works. At this time of year, that's all that matters, no matter how nontraditional it feels.