The Brewers’ current rotation consists of Kyle Lohse, Matt Garza, Yovani Gallardo, Marco Estrada, and Wily Peralta. None of them is an “ace” is any real sense of the world, but all of them can be decent, eat innings, and be quite effective. (Entirely excepting last night, of course). Brewers starters currently rank third in the entire league in Quality Starts, and are tied for fourth in average Game Score.
At the same time, waiting in AAA is Jimmy Nelson, the Brewers’ second-round draft pick in 2010. Nelson has been an absolute beast in AAA this year, showing every sign of being ready for the show. Among qualified Pacific League starters, he has the best strikeout rate (9.52 K/9), the best SIERA (2.78), the best FIP (2.44), and above all, the best ERA (1.56). Baseball America just rated him the #38 prospect in baseball.
This means that the Milwaukee Brewers have one luxury most teams do not: six pitchers capable of pitching effectively in a starting role. What do they do about it? One option is to substitute Nelson for the lowest-performing starter at the moment; that person clearly is Marco Estrada.
The alternative, which I think it a better idea, would be to use Nelson to create a six-man rotation, thereby giving their existing starters an extra day’s rest between starts. Why this instead?
Let’s begin with starters as a class. In running some numbers this weekend, I noticed something interesting: over the last few years, league starters overall tend to start the year strong (or at least average), gradually decline in performance through August, and then abruptly improve their performance in September. Over the weekend, I discussed this phenomenon with many thoughtful people on Twitter, and it was agreed that the primary difference in September was September call-ups: that time when major league clubs expand their rosters and call up their best minor leaguers (whose seasons have ended) to contribute during the last month of the year. By calling up these additional bats and arms, managers no longer have to keep their starters in games as long.
What effect does that have? Well, on average, from 1991 through 2013, and excluding strike-shortened 1994, starters improved by 4 points of batter OPS+ between August and September of each year. One thing that correlates well with that performance improvement? An average decline of almost one batter per starter in September games. There is a significant (p<.05) connection between the decreased starter workload in September games and their increased effectiveness. In fact, that decrease in their workload explains almost 20% of the improvement that starters make in September games (r2 = .19).
So, even simply looking at league averages, it appears that workload matters when we start approaching the end of the season.
Now, let’s look at the existing Brewers rotation. Wily Peralta is young and built like a horse. Marco Estrada has been subpar so far, but actually has a history of performing much better in the second half of each season. But the anchors of the rotation are Kyle Lohse, Matt Garza, and Yovani Gallardo. These are the three pitchers the Brewers would count on for post-season play. They are already getting older and probably on the downside of their careers. Moreover, all three have typically shown decline throughout their careers over the course of each season. For his career, Kyle Lohse has pitched 20% above league average in April (80 OPS+ allowed); he declines to 5% below league average by July (105 OPS+), before improving to slightly below league average in September (102 OPS+). Matt Garza has followed a similar path: typically excellent in April (85 OPS+), but he runs into trouble by August (108 OPS+) and is entirely out of gas in September (120 OPS+). Gallardo is slightly more consistent: he tends to be slightly below league average in April (103 OPS+), notably worse by August (108 OPS+), but often rebounds in September to be slightly above average (98 OPS+). Since that data runs through 2013, and those three pitchers are currently a year older, we should probably assume those numbers won’t improve. (Remember: 100 OPS+ is league average).
This year, the rotation seems poised to follow a similar trend. In March / April, Brewers starters had a 3.01 ERA. In May, they had a 4.08 ERA; below average. In June, they had a 4.08 ERA again, so again below average.
So, what do we know? (1) Starters in general tend to decline in performance over the course of a season. (2) The Brewers’ primary starters have shown similar tendencies. (3) Starters perform better later in the year when their workloads go down.
That brings us to Jimmy Nelson. He is certainly ready to face major league hitters, but he has also struggled to adjust each time he was promoted to a higher level, needing a few months to get his ship right. In 2012, his walk rate ballooned from 8% to 18% when he transitioned from Brevard (A+ ball) to Huntsville (AA ball). Once he mastered AA hitters, Nelson was promoted to AAA in 2013, during which his walk rate doubled from 6% to 13%. Nelson once again got his walk rate under control at AAA, but walked 12% of his batters in his lone big-league start this year. In short, Nelson has needed time at each level to adjust to the hitting and umpiring: he likely will have a bit of a transition at the major league level as well.
Adding Nelson to the rotation, without taking others out, allows the existing starters to get an extra day of rest. Adding Nelson to the rotation, without taking others out, allows him to make his adjustments (if he is going to make them) while the Brewers are multiple games up in the division, rather than while they are possibly neck-and-neck at the end of August. Adding Nelson is also unlikely to cause an overall decline in the projected performance of the rotation: Steamer projects him to have a 3.85 ERA and a 3.94 FIP for the rest of the year. That compares favorably to Steamer’s projections for Lohse (4.34 ERA, 4.37 FIP), Garza (4.13 ERA, 4.03 FIP), and Gallardo (4.20 ERA, 3.93 FIP). It also allows Nelson to be truly ready if in fact one of the other starters gets hurt or becomes completely ineffective; if that happens, the Brewers would simply revert to a five-man rotation with Nelson in it.
Most fans have instead been thinking that Nelson should replace Marco Estrada now, sending Estrada to the bullpen. I think this is the lesser of the two options. First, Estrada is still projected to be reasonably effective for the remainder of this year (4.05 ERA, 4.38 FIP). Second, as noted above, Estrada has a history of finishing seasons very strong. Finally, I think the “Estrada to the bullpen” argument proves too much. If your argument is that Estrada is too homer-prone, putting him in the bullpen is hardly a solution. If there is one thing bullpen arms should not do, it’s give up home runs. If Estrada’s really not good enough for the rotation, then he’s probably not good enough for the Brewers in general. And although I’m perhaps the only one who feels this way, I think we’re at least another month of starts away before making a decision like that.
Would starters actually benefit from the extra rest? Well, that’s hard to predict. But, we may have gotten a preview of the effect last week, when the Brewers got two days off around a Toronto road trip. What happened? Matt Garza’s fastball in Cincinnati was the hardest and most effective it had been all season, at 94.32 mph, per Brooks Baseball. Yovani Gallardo’s four-seam fastballs were 1.5 mph faster, just short of 94 mph — just like the Yo of old. Two outings do not by themselves prove anybody’s point, but there certainly is no evidence that extra rest posed either Garza or Gallardo any problem. With a six-man rotation, the Brewers can potentially continue that trend throughout the summer months.
Follow Jonathan on Twitter @bachlaw.
All data from Baseball Reference.