clock menu more-arrow no yes mobile

Filed under:

The Six-man Rotation: The Abominable Snowman of roster management

The Mets are about to try and capture this mythic creature. Here's a look into what the numbers say about this elusive and possibly beautiful beast.

Noah K. Murray-USA TODAY Sports

The six-man rotation is like the Abominable Snowman. Or Yeti, or Sasquatch, or whatever your preferred euphemism. It is occasionally sighted, never confirmed. Is it a beast to be feared, or is it a misunderstood secret beauty waiting for us to be ready to accept its wondrous gifts? So many questions.

The New York Mets may actually have the means to give us some answers on the subject this year.  It looks like they will go forward with a six-man rotation when Dillon Gee returns to the club, possibly running with it until August according to recent reports. It would seem to be a benefit to capitalize on the team’s pitching depth as they are looking to limit innings for their young hurlers. With Steven Matz waiting in the wings and Zack Wheeler looking to return from Tommy John surgery next year, the dream of having a 'staff of aces' feels tantalizingly close. Six-man staffs are standard in Japan and Korea, so there is some precedent.  Now that this mysterious phenomenon is about to step out of the tabloids and into our lives, it has to be asked: Is it a viable long-term strategy moving forwards?

The biggest anecdotal concerns about six-man rotations revolve around pitchers being ‘creatures of habit.'  A number of them (including Matt Harvey) have stated that they don’t like the extra rest, it throws off their routine and rhythm, and there is the worry of pitchers being ‘too strong’ with additional rest and therefore losing some of their command. And of course it seems counter-intuitive to have your best pitchers take the hill less often when you are trying to win now.  With so much money on the line, hesitancy to change or experiment and therefore undertake risk is understandable.

So do the numbers support the idea for a six-man rotation or the concerns about them? It’s notoriously hard to do research on something that has yet to happen. We do have some data related to the subject, but it is imperfect. As you may imagine from imperfect data, the results are a mixed bag.

In 2006, the book called The Book: Playing the Percentages in Baseball took data from almost 15,000 starts spanning from 1999-2002. It posited that the ideal number of days of rest for a pitcher was 5 days, since the wOBA against on such rest (.346) was lower than 4 days rest (.351) or 6 days rest (.355). The problems with looking at the numbers for 6 days rest here are multi-fold: this is the smallest sample size of the three (645 games); it does not account for why the pitcher got the extra days of rest (hidden injury or other reason why he was pushed back in the rotation, lesser starters are often the ones pushed back, etc.); therefore the small sample size is likely to be polluted with a higher number of lesser starters. Pairing groups of ‘equal ability’ or looking at individual player’s performance may be more indicative here, but again would make sample sizes even smaller.

In 2009, Eric Seidman of Baseball Prospectus ran fastball velocity and pitch F/X data over various days of rest during the 2008 season, breaking up 150 pitchers into groups of ‘hard throwers’ (93+ mph), ‘medium throwers’ (90.5-93 mph), and ‘soft tossers.' There were some slight differences noted—with extra rest, all throwers got a slight increase in velocity; medium throwers had a slight increase of movement, while hard throwers and soft tossers had slight decreases in movement—but overall, the results did not seem to indicate anything particularly significant either way. Again there is an issue of selection bias here, as there is no way to show that any given pitcher has the same proportion of starts within the group on 4 days rest as 6 days rest. But the fact that the numbers remain so close regardless is intriguing.

In 2010, J.C. Bradbury (then of and Sean Forman of Baseball-Reference did a paper on pitcher workloads. Using all pitcher starts with at least 18 batters faced from 1988 to 2009, they hypothesized that with each day of extra rest, ERA would improve by .015 points, which nevertheless would not be statistically significant.

What also has to be taken into account with all these numbers is the irregularity with which a pitcher gets 5 and especially 6 days rest: but again given the fact that nothing significantly negative can be gleaned from existing evidence, one has to wonder if the numbers could actually improve somewhat if pitchers were given a structured 6-day schedule. But questions still remain: how might a change in schedule affect individual pitchers?

So let’s take a look at the numbers for a group of successful pitchers with a sizable track record that are familiar to Mets fans, using some indicators available through Baseball-Reference:

GS 4 days rest GS 5 days rest GS 6+ days rest BABIP 4 days BABIP 5 days BABIP 6+ days OPS 4 days OPS 5 days OPS 6+ days K/W 4 days K/W 5 days K/W 6+ days
Tom Seaver 445 81 59 .259 .259 .282 .616 .633 .675 2.65 2.56 2.34
Nolan Ryan 378 104 99 .265 .266 .272 .611 .572 .609 1.98 2.48 1.94
Jerry Koosman 315 84 47 .287 .280 .270 .691 .683 .625 2.19 1.87 2.22
Dwight Gooden 253 94 56 .291 .292 .297 .656 .654 .744 2.58 2.29 1.88
David Cone 231 111 63 .290 .275 .272 .684 .644 .661 2.41 2.27 2.25
Ron Darling 217 90 40 .287 .270 .256 .723 .688 .708 1.72 1.81 1.72
Sid Fernandez 160 72 58 .237 .253 .274 .615 .601 .721 2.56 2.35 2.06
Bob Ojeda 151 72 52 .288 .283 .281 .702 .682 .717 1.77 1.62 1.56
Bret Saberhagen 204 93 57 .290 .289 .275 .677 .684 .662 3.75 3.48 4.12
Al Leiter 189 119 71 .289 .291 .278 .716 .704 .672 1.60 1.87 1.75
Orel Hershiser 249 143 50 .277 .277 .282 .659 .697 .693 2.26 1.72 1.58
Tom Glavine 434 164 59 .283 .290 .278 .687 .720 .705 1.71 1.83 1.55
Pedro Martinez 199 161 47 .285 .278 .291 .609 .607 .675 4.08 4.84 3.70
Johan Santana 160 92 26 .274 .278 .278 .645 .638 .702 3.69 4.45 2.46

What stands out here on the positive front is the remarkable stability of both BABIP and OPS over 4 and 5 days rest.  Considering the majority of starts in a six-man rotation would be on five days rest, this would be a heartening sign.  If we look at BABIP as an indicator of quality of contact allowed, arguably only Fernandez performed markedly worse, and even then still stands almost 50 points below league average, while his OPS allowed actually dropped.  Only Glavine and Hershiser saw their OPS go up appreciably, while Ryan, Cone, and Darling saw a significant drop in OPS.

Things get a little more muddled on 6+ days rest.  6 of the 14 pitchers saw a significant rise in OPS as opposed to 4 days, while 9 of 14 saw a notable drop in their K/BB rate, which may be indicative of problems with command.  Still, BABIP was stable or improved for 11 of the 14 hurlers.  Again, the irregularity with which 6 days of rest occurred (or in some cases more; BR didn't specify how many starts came on more than 6) has to be considered as a mitigating factor, as does possible reasons for such a layoff.

Of course, it's also impossible to control for quality of opponent in these numbers, or ballpark effects, or run support, or any number of subtle things that could affect performance.  Yet overall you get a sense that there is some individuality in the way people react to extra rest, and that there will be bumps for some pitchers adjusting to 6 or more days rest—as you would expect for any person (athlete or otherwise) who has their schedule changed.

How have the very players who have prompted this inquiry performed on various days of rest? Let's check in with the numbers for the Mets' staff:

GS 4 days GS 5 days GS 6+ days BABIP 4 days BABIP 5 days BABIP 6+ days OPS 4 days OPS 5 days OPS 6+ days K/W 4 days K/W 5 days K/W 6+ days
Bartolo Colon 224 176 71 .292 .306 .297 .722 .766 .697 2.64 2.66 2.23
Jon Niese 71 59 27 .328 .306 .320 .768 .713 .724 2.66 2.74 2.49
Dillon Gee 55 31 21 .282 .300 .277 .712 .733 .729 2.32 2.26 2.32
Zack Wheeler 26 16 7 .267 .344 .310 .630 .739 .772 2.43 1.88 1.89
Matt Harvey 24 12 9 .289 .293 .227 .637 .530 .436 4.39 5.36 5.27
Jacob deGrom 16 12 4 .317 .287 .265 .621 .597 .610 3.30 4.32 3.71

With relatively small sample sizes for everyone except Colon, it's hard to go too nuts either way with these figures, especially since they are a bit of a jumble.  But it is fun to point out that the early returns look really good for "enemy of rest" Matt Harvey and one of the other arms the Mets hope to protect, Jacob deGrom.  Harvey has a 1.07 ERA with 79 Ks in 58.2 innings in the 9 starts with 6+ days rest.  It also might be a little disturbing how much trouble Zack Wheeler has had in starts on more than 4 days rest, which is nearly half of his career total.

It's difficult to know what to make of all these numbers since none of them actually capture what it's like to be in a six-man rotation.  On the one hand, you know that five days of rest works out just fine, and keeping a schedule which varies by a single day (in this case, after 4 or 5 days rest) holds no concern.  Could the same effect be achieved if you regularize 5 and 6 days rest for your top pitchers?

Even if you hypothesize that, say, weakening muscle memory may be to blame for some pitchers' trouble on 6+ days rest, you would have to imagine that trainers could structure off-day programs to account for that.  The fact is, the human body is incredibly resilient and malleable; it’s really good at adjusting to changing conditions, especially if given time to adjust. General agreement about sports science, like any science, is and has been subject to change. After all, the five-man rotation itself is a relatively new phenomenon in the context of the game's history, with widespread use coming into vogue 40 or so years ago. Before that, a four-man rotation was the norm, and before that three was the magic number. And before that, it wasn't unheard of for a pitcher to throw 500 innings per season.

The ideal way to approach an enterprise as revolutionary as a constant six-man rotation would involve an offseason of setting up a well-considered regimen, which should ameliorate problems or concerns. It looks like the Mets are instead likely to make an experimental laboratory out of ballparks around the nation this summer, but there doesn't seem to be any clear physiological or statistics-based indications that this wouldn't work over a long term basis.  It is a bold step into the unknown, but the Mets' roster has conditions ripe for a more than cursory investigation into this matter.