July 7th, the final Friday before the All-Star break. The Cardinals continue their 10 game homestand by welcoming the New York Mets into town for a three game weekend set. Carlos Martinez gets the ball for St. Louis; Jacob deGrom for New York. Needless to say, runs were going to be few and far between this evening. Or so we thought. Five innings and six home runs later, the game enters the sixth with a combined nine runs already in, the Mets holding a 5-4 lead. Martinez gives way to John Brebbia, who works a 1-2-3 frame. The Cardinals go quietly in their half of the inning, taking us to the seventh. What followed were the six highest leverage plate appearances of the entire game for St. Louis relievers.
The red hot Brett Cecil and Matt Bowman never took the mound that night. Neither did Seung-Hwan Oh, forcing Mike Matheny and his bullpen orthodoxy to save Trevor Rosenthal for later in the game. Rather than use the threesome of Rosenthal, Bowman, and Cecil, Matheny watched as Brebbia and Tyler Lyons surrendered the run that proved to be the difference. This one is in the books: Mets 6, Cardinals 5. So why were Cecil, Bowman, and Oh all unavailable?
All three had pitched both yesterday and the day before. The potential of the vaunted back-to-back-to-back was enough for Matheny to turn to inferior pitchers with the game on the line. Did he make the right move?
I began by compiling data from every single MLB game this season. A few clicks later, I had found what I was looking for: the results from all 9318 relief appearances in 2017. That gave me a sample of 10,016 innings and 42,805 plate appearances to work with.
Technical notes: Although pitching on three consecutive means that, by default, you are pitching on two consecutive days, the former category's data was isolated from the latter for obvious reasons. Also keep in mind that the sum of these shares will not equal exactly 100% due to rare pitching situations, such as pitching twice in one day or four consecutive days.
Relief Pitcher Rest Breakdown
|Pitching on...||Share of STL outings||Share of MLB outings|
|Pitching on...||Share of STL outings||Share of MLB outings|
|3 consecutive days||2.39%||2.12%|
|2 consecutive days||19.12%||19.75%|
|1 day of rest||33.86%||29.96%|
|2 days of rest||19.52%||21.38%|
|3 days of rest||13.15%||12.24%|
|4+ days of rest||11.55%||14.43%|
It appears that Mike Matheny's philosophy regarding reliever rest falls in line with the rest of Major League Baseball. After all, none of these groups differ by more than 3.9%. Regarding the question of whether or not using relievers a third straight day is acceptable, both Matheny and the rest of baseball agree that there are certain situations where it is justified.
I'm aware that 2.39% and 2.12% may seem like blips on the radar, but look at this way. Those numbers are relative to all relief appearances, not just those who already pitched two days in a row. As of this writing, about 12.5% of all Cardinals relievers who pitched two straight days returned for a third consecutive day, slightly above the 10.8% rate for the league as a whole.
Because of travel days, a 100% return rate is physically impossible. Also factor in that the third game will not always lend itself to a team's better relievers (as you would probably expect, managers seldom deploy their mop-up men three straight nights) entering the game and it becomes apparent that managers are actually fairly comfortable using their relievers on back-to-back-to-back days.
To put this hypothesis to the test, let's play a numbers game. We'll begin by estimating that over the course of a 180-or-so day regular season there are approximately 20 off days. That leaves us with roughly 90% of the relievers who pitched back-to-back that will actually have the opportunity to take the mound on three consecutive days. Let's also assume that of the 90% whose schedule gives them the chance, only around 35-40% of those games will contain close situations, preferably with the team in question leading, that would call for a club's better relievers. If we crunch the numbers on all of that, we find that managers will go with their reliever for a third straight day in about one out of every three realistic scenarios, a far cry from the measly 2.12% league mark in the table above.
The next step was to find out what changes in performance, if any, result from different amounts of rest. Before we dive into my findings, I should note that I only looked at relievers who at one point or another did pitch for three straight days. The reason for this was, like I said in the paragraph above, managers are more willing to pitch their better relievers with less rest. This higher talent pool would prop up the three consecutive days group's stats, thus tainting our results. What I did was eliminate this talent gap by ensuring that each rest group I analyzed had the same raw talent level, making it easier for us to draw conclusions from the data.
Relief Pitcher Performance by Rest
|3 consecutive days||3.99||4.33||4.17||4.16|
|2 consecutive days||3.69||4.04||3.28||3.67|
|1 day of rest||3.82||3.83||3.83||3.82|
|2 days of rest||3.83||4.02||4.02||3.95|
|3 days of rest||3.80||4.08||3.66||3.85|
|4+ days of rest||4.33||4.32||4.18||4.28|
The two consecutive days through three days of rest rows turn out rather similar numbers, especially when you consider the fact that ERA is the most volatile (and least indicative of a pitcher's true success) of the three metrics. However, we do notice drops in performance among the two most drastic rest categories, three straight days and four-or-more days of rest.
Why is this? We can assume that a pitcher's stuff deteriorates with less rest, although a plethora of PitchF/X data would be needed to definitively confirm or deny this. On the other end of the spectrum, my best guess would be that relievers accumulate a good deal of "rust" when they don't pitch in an actual game for an extended period of time. This of course would in turn make them "less sharp" their next time on the mound, though once again, further research would be required to test this theory.
Recall that Mike Matheny uses his relievers for three consecutive days at a slightly above league average rate. One could argue that the if the talent spread among your relief core is wide enough, it would still be worth it to use your best relievers, even if we can expect their performance to take a dip. But take a look at the Cardinals top five relievers by innings pitched through the lens of various ERA predictors. Pay close attention to the column labeled "average".
St. Louis Cardinals Bullpen Breakdown
Outside of Trevor Rosenthal, who has clearly proven himself to be the Cardinals best reliever in 2017, the difference in performance between Bowman, Cecil, Oh, and Siegrist is rather slim. If anything, there should be even less incentive for the Cardinals to use their relievers (with the possible exception of Rosenthal) for three consecutive days. This list doesn't even include John Brebbia, Tyler Lyons, or Sam Tuivailala, all three of whom further stress the point that the greatest virtue of Matheny's bullpen is its depth. Compare the Cardinals bullpen to that of their former division rival.
Houston Astros Bullpen Breakdown
The Houston Astros hate using relievers for three straight days. In fact, they have only done it a grand total of two times the entire season (once with Michael Feliz and again with Chris Devenski, by far their most valuable relief pitcher). In many ways their bullpen hierarchy resembles that of the Cardinals. Both teams have a clear cut top reliever while numbers two through five have performed fairly similar to one another.
Now let's compare the Cardinals to another former division rival, this time from the old NL East.
New York Mets Bullpen Breakdown
The New York Mets' share of relief appearances coming for the third consecutive day is more than double the MLB average. And when you take a look at their bullpen composition you'll understand why. It's no secret that the injury-riddled Mets have seen their pitching depth evaporate in 2017, leaving them with a big drop-off after established closer Addison Reed, veteran lefty Jerry Blevins, and rookie right-hander Paul Sewald. Terry Collins would much rather place the game in the hands of a running-on-fumes Reed than a Josh Edgin or (World Champion) Fernando Salas, who both grade out as roughly replacement level arms according to FanGraphs.
Looking deeper into the Mets bullpen, their sixth through ninth most used relievers (Josh Smoker, Hansel Robles, Neil Ramirez, and Rafael Montero, who was recently moved back to the rotation) have all posted at or below 0.0 fWARs in relief. Needless to say, Collins, unlike Matheny, can use his top-heavy bullpen to justify trotting out his relievers three nights in a row.
That's not to say that pitching on three consecutive days is ideal. Nor am I letting Matheny off the hook because he's adhering to the herd mindset. Fortunately, the Cardinals haven't been burned too badly this season.
However, just this past Saturday, Matt Bowman was charged with allowing the game-winning run in the Cardinals' 3-2 loss to the Cubs. The appearance was Bowman's third in as many days, and all the more frustrating when you consider that quite literally every single arm in the bullpen was more rested than his.
In what was the most important at-bat of maybe the most important game of the season (as FanGraphs' leverage index metric attests), Matheny's more optimal options (Rosenthal, Oh, Cecil, and arguably Brebbia and Siegrist) watched as Kris Bryant knocked home an RBI single against a worn down Bowman.
If the numbers tell us anything, it's only a matter of time before the Cardinals shoot themselves in the foot...again.