The Rotation Was Good

Entering the 2018-2019 offseason, there remains a contentious debate among many Brewers fans about the need for the Brewers to improve starting pitching. Who can blame these fans? They just spent three weeks watching national analysts bludgeon the Brewers roster construction, bemoaning at nearly every chance that an ace would be preferable to whatever the heck it was that these Milwaukee clowns were doing. And even if other playoff series did not go according to plan (for instance, the Brewers summarily dismissed true ace Kyle Freeland and the Colorado Rockies, and the Houston Astros “all ace” rotation was crushed by Boston. Pitching wins championships except for when hitting wins championships!), there is simply an aesthetic aspect of acehood that resonates with baseball fans. Who can blame them? You want to know who’s pitching when you go to the ballpark, and it’s more fun to talk about pitching using fleshy, breathless language like “a stud” or “a dude” (the Brewers need to get “a dude”, I’m often told during @bpmilwaukee Twitter chats, a demand for which GM David Stearns is unfortunately in the wrong business). Ironically, all Brewers fans needed to do was to consult stats like Deserved Run Average (DRA), a pitching statistic that estimates a pitcher’s runs allowed based on a full array of contextual factors, and their case would be much easier made. But even there the whole story is not told, so it all boils down to an assertion:

The Brewers need starting pitching help. The Brewers need an ace.

┻┳|
┳┻|
┻┳|
┳┻|
┻┳|
┳┻|
┻┳|
┳┻|
┻┳|
┳┻|
┻┳|
┳┻|
┻┳|
┳┻|
┻┳|
┳┻|
┻┳|
┳┻| _
┻┳| •.•) #Brewers didn’t need a SP
┳┻|⊂ﾉ
┻┳|

— BP Milwaukee (@BPMilwaukee) July 31, 2018

Of course this would be the return line for the 2018-2019 offseason, because the line never went away during the season. A large faction of fans were dissatisfied with the starting pitching in April; they were satisfied with the starting pitching in May, “but can this staff beat ‘a dude’ in the playoffs?” (Yes!, it turns out); they were particularly dissatisfied with the starting pitching when the season ended in June and July, and again they were dissatisfied with the starting pitching at the trade deadline. This debate was simply never going to be won, because there is a contingent of baseball fans that refuse to either understand or accept what GM Stearns, pitching coach Derek Johnson, systemwide player development, and the front office are trying to accomplish. For arguably the first time in Brewers franchise history, certainly for the first time in a generation, the Milwaukee system strength is pitching, and not of the sort of high octane, all-risk dreamy profiles that flamed out at the turn of the 21st Century; this is a system that is built on turning a fabulous diversity of pitching profiles into potentially successful MLB profiles (witness the scouting range between Freddy Peralta and Corbin Burnes, for example).

When the dust settled, the system worked. The Brewers rotation was good. It was good any particular way you measured it; it was a good rotation if you divide rotation spots based on overall Games Started and workload measurements; it was a good rotation if you divide rotation spots based on true rotational scarcity (i.e., comparing each spot across the MLB); and it was a good rotation if you separate pitching classes into “true starters” and “replacements,” and measure each set of pitchers against different “spots” or “workloads.” The pitching staff was good if you believe in “Aces,” and it was good if you don’t believe Aces exist.

The Brewers rotation was good by every measurement except DRA, which should be the significant focal point of 2018-2019 offseason analysis in an effort to understand how Milwaukee assembled an elite fielding component in order to prevent runs.

If you do not wish to read the details, the table above summarizes the comparative results from each model. Each Brewers starter was assessed according to their relevant spot, and then compared by Innings Pitched (IP), Deserved Run Average (DRA), and Runs Prevented.

Rotation One: By Games Started
One way to assess a starting rotation is by ranking pitchers according to games started on a leaguewide basis. This ranking method is effective because it approximates the scarcity of both MLB resources (there’s not a whole lot of pitchers that can work full seasons) and roster construction. One benefit of focusing on games started instead of another performance metric is that analysts can reflect the success or failure of an MLB club across games started totals; for example, it matters that Gerrit Cole and Lucas Giolito both started 32 games despite widely divergent performances. The distance between Cole and Giolito is approximately 65 runs prevented, even though they worked the same number of starts, which raises an important question about how different teams assess the importance of effective starters versus soaking up innings. In fact, had Brent Suter and Zach Davies not faced injuries in 2018, they may have forced this question with the Brewers front office, and Freddy Peralta also arguably faced this (along with innings workload concerns) down the stretch run.

According to this measurement, there are approximately 10 rotation spots discernible by workload throughout the 2018 MLB, as well as emergency starters (who started one game; I will always assess emergency starters as their own category). On the surface, this is a pleasing model; the top starters by workload typically are the best starters in the game, even if there are differences between guys like Cole and Giolito, as discussed above.

On this model, it is clear that the Brewers succeeded because of their depth. A critique about the top of the rotation could be true in terms of DRA, as the contextual performances of Jhoulys Chacin and Chase Anderson were not comparable to top workload pitchers across the MLB. The importance of the depth should not be understated, from Wade Miley and Peralta to Brandon Woodruff and even Gio Gonzalez. If you’re reconsidering Gonzalez’s trade cost, not only should the veteran lefty’s surface performance be assessed, but one should not that, marginally, he was worth seven runs better than his median workload.

Another benefit of using this model is that analysts can assess “phantom” runs prevented where teams “miss” particular spots. For example, Chacin may not measure up to the median Top Spot prototype, but having his performance was better than not having a heavy workload pitcher whatsoever (in theory; Giolito’s performance would obviously have not validated a heavy workload benefit for the Brewers). If a team was missing a Top Spot, they theoretically would be punished 16-to-17 Runs Prevented. Milwaukee did not use a Five, Seven, Eight, or Ten workload, each of which approximately ranged from 2 to 4 runs below average; one could argue in this way that the Brewers also received 10 “phantom” Runs Prevented by avoiding these typical workloads.

This should help to validate the ideal that there are a couple of different ways to construct a rotation. A team could indeed bank on a Jacob deGrom type atop the rotation, and seek a 30 run advantage from their top workload. One must be careful of the cost for this type of pitcher, however, as if considerable resources are spent at the top of the rotation, they may be diminished at the bottom of the rotation. The Brewers demonstrated the “bottom-up” approach: they lost out on the Yu Darvish sweepstakes, and Alex Cobb did not bite on a one-year deal, so they proceeded with Chacin and Miley, plus their developmental pipeline. That internal pipeline was worth approximately five runs (better than their median workload) to the 2018 Brewers, while external candidates were worth more than 15 runs (better than their median workload). It was not flashy, there were no “dudes” on the marquee, but it worked.

Rotation Two: By Team

Of course, even the preceding model is relatively clean or “idealistic,” for MLB teams do not necessarily construct their rotations according to the same ideal. An additional method for assessing rotations is to judge each team’s rotation spot by turn; since two pitchers literally cannot start the same game, this method goes spot-by-spot, start-by-start for each MLB team. The benefit of this method of rotational assessment is that it reflects team preference, or injury and ineffectiveness circumstances, across the league. Some teams attempt to duct tape 13- or 14-pitcher rotations together, whether they are contending or tanking, while others attempt to yield more mileage from each spot. By giving each team one exclusive spot for each turn (until their pitchers run out), this type of rotational model can allow teams to be analyzed against attrition across the league.

Every team in the MLB required at least six rotational turns throughout the season, but this model demonstrates the divergence of team strategies one they hit six starters. Some teams preferred to give replacement starters two or three starts each, while others leaned on emergency starters.

The Brewers front office, coaching staff, and pitchers did a fantastic job weathering 162. They hit the right buttons in replacing some starters at certain points in time (such as resting Peralta down the stretch, or [arguably] “shuttling Woodruff between Triple-A and MLB), while giving starters room to breathe at others point in the season (this also applies to Peralta, who was given some time to adjust from rough starts, as well as Junior Guerra). By spitting on rotation spots 10 through 14, the Brewers also arguably saved 16 “phantom” runs, as the club would not have found effective pitchers (on average) digging that deep into league or organizational resources. (This line could be argued with further research, however, as one could note that someone like Corbin Burnes could have been effective in two starts, for example).

Rotation Three: By Type
During my time writing at Sportsbubbler (RIP) and Disciples of Uecker, I published annual starting pitching rotation rankings based on the decision point of 100 IP. If a pitcher worked 100 or more innings with 50 percent of their games as starts, they were a starting pitcher; if not, they were replacement depth. On this model, I attempted to assess pitchers according to Runs Prevented, with the ideal that (a) working a lot of innings should be worth more as a starter, and (b) rotation spots could be designated based on the resulting Runs Prevented rankings. I’m no longer certain of this method’s veracity, as I believe there are better ways to assess rotational scarcity and usage across the MLB. But, here we are, testing the Brewers 2018 rotation, so let’s assemble the pitchers.

Wouldn’t you know it, the 2018 MLB did not have many “true” rotation spots: there were only 129 pitchers across 30 teams that fit the first criterion listed above. This is not enough pitchers to fill a true five man rotation, and it’s hardly enough to fill a four man turn.

Yet, those Runs Prevented totals present some order to the universe. There are aces, even if there’s only a couple of them. There are nice middle of the rotation “dudes” that you can really sink your teeth into; 150 IP and 2 Runs Prevented feels like a solid effort for a team. Every contender would accept that workload (every MLB team would, for that matter).

Where there are not full-time starters, replacements are necessary, and MLB really dug deep in 2018: there were 227 replacement starters, including 57 Emergency Starters, across 30 MLB teams in 2018. Basically, on average, MLB teams were using more replacements than they were using regular starters. The Brewers are no different here, and in fact, that’s partially how they gained their value. Viewing the range of Runs Prevented across each of these roles should demonstrate the importance of having a solid organizational pitching strategy; replacement starters need not simply be the pitching equivalent of throwing spaghetti against the wall. Tampa Bay demonstrated this with their genius “Opener” strategy, and they produced one of the elite Runs Prevented units in baseball. The Brewers accomplished their success by using long-term replacements like Miley and Peralta, but they also received value elsewhere across their high-floor organizational depth.

If you previously thought the idea of “Phantom Runs Prevented” by not using a rotation spot was a suspect idea, this seems to be your chance to pounce on the Brewers for not using an “Ace” or true “Number One” starter. By this model, the Brewers were gutsy, punting nearly 63 runs prevented at the front end of the rotation. Yet, the club also did not use a true “Number Four” or full-time starter that should have been replaced, which bought the club another 43 runs prevented. All told, the Brewers rotation of regular starters lost the club approximately 19 runs here, thanks to their cavalier strategy.

Of course, the Brewers used every Replacement typology except a “true swingman,” and this is where the club torched the league. Gio Gonzalez and Wade Miley covered the lack of an “Ace” or “Number One” starter, and demonstrated the value in not having a Regular Four, either. Peralta, Dan Jennings (yes, Dan Jennings), and Woodruff gained significant Runs Prevented advantages in the replacement ranks as well. On top of these depth successes, the rotation was not bad overall; Suter and Guerra were close to true Number Three starters, and Anderson was close to a true Number Two starter. Chacin was better than a typical Number Two starter, boasting a Runs Prevented performance that almost placed him in a phantom “Number One” role for the club.

What is startling on this model is that the Brewers typologies also worked according to DRA. Once an analyst accepts that the club did not have a True Ace or True Number One starter, the threshold for assessing DRA is lowered significantly. Witness Chacin, for example, who was assessed against DRA that were significantly better than 4.00 on the first two models; his DRA performance looks much better on the final model, because once you stop comparing him to Aces, the comparison becomes more realistic. The Brewers deep organization also performs very well against median DRA requirements on this model, which raises a question about which model’s expectations one should use going forward.

The last remaining criticism for fans and analysts rests on how one interprets Deserved Run Average results for the Brewers rotation. On any model one chooses, be it based on Games Started, Team Rotational Turns and Scarcity, or Actual Runs Prevented performance, the Brewers’ rotation was good in 2018. Now it is worth digging through these models during the offseason, in order to gain important lessons for Corbin Burnes, Woodruff, and Peralta during their potential first full workloads in 2019, and even for reworking Jimmy Nelson. Milwaukee has proven the success that can come with aggressive rotational swings and an organizational pitching strategy, coupled with elite, efficient fielding.

An underrated 2018-2019 offseason would find David Stearns making moves to further improve the fielding (such as improving Right Field, and then working Christian Yelich primarily as a Left Fielder), which should in turn help boost the pitching depth strategy going forward. As it stands, the Brewers do not even need an external pitching move; this makes potential offseason moves even more interesting for speculation.