How an Ace Performance Impacts Reliever Workloads

The running theme of the 2017 playoffs has been the heavy reliance on bullpens to get hitters out, particularly in the fourth inning–or even sooner. The practice of #bullpenning is being greatly complimented by some, and hated by others. One thing’s for sure, it’s making games a lot longer.

One of the main reasons for pulling the starter from the game is to avoid the dreaded third-time-through-the-order effect, the hitters becoming more familiar with the pitcher or the pitcher becoming fatigued and losing his effectiveness. In the regular season, leaving a starter out a bit longer to save the bullpen makes sense; it’s a long season, and you’ll get ‘em tomorrow. However, in the playoffs, there often is no tomorrow. You need to stop rallies fast and get hitters out. There’s no punting a game with hopes you’ll just win one the next day.

Elite starters who can give you length in a game are incredibly helpful for your chances of winning, both because they reduce the workload in your relievers and because they offer the opposing team fewer looks at your relief arms. So why doesn’t every team just go out and draft the next Justin Verlander? Well, that’s because guys like Verlander are about as freakish as you will see in professional sports. To play at such a high level for so long and remain relatively free of injury is next to unheard-of. And to pair that durability with such elite velocity? This is a throwback pitcher in the modern era, with the pitching arsenal to still be ahead of the velocity curve.

So, then why doesn’t every team just go out and sign elite starting pitching? For one, it’s extremely expensive. Looking at SporTrak salary reports for all pitchers in the 2017 postseason, the average salary for starting pitchers is $9.5 million, and the average relief salary is $4.0 million. While elite closers command huge salaries ($9 million for Andrew Miller, $15 million for Aroldis Chapman), young, controllable relief pitching gets outs at a very low cost. Chris Devenski, Ken Giles, Carl Edwards Jr. and Archie Bradley are all making less than $600,000 per year, but they have recorded outs in some very important innings this postseason.

If you look back on the history of advanced statistics and Moneyball, the main reason to use these numbers was to find undervalued commodities that came cheap and helped teams win on a budget. Yes, workhorse starting pitching is expensive, but that doesn’t mean teams with smaller budgets can’t compete.

To quantify workload, I have moved away from the traditional metrics of games, innings pitched, and pitch counts. There is extensive scientific literature that states these metrics do not serve a protective effect for pitchers. The metric I have created is called Fatigue Units, which aims to quantify the cumulative muscle fatigue of the forearm a pitcher will develop over the course of a game and during a season.

Keep in mind, muscle fatigue is the amount of force a muscle loses during exercise, not a level of tiredness. A predicted muscle fatigue level is calculated for each inning, summed over the course of the game, and then multipliers for how hard a pitcher throws are applied. The FUs for a game then are multiplied by a correction factor representing the number of rest days the pitcher had before appearing in that game. This metric has outperformed cumulative innings pitched when looking at risk level for Tommy John surgery in the following seasons.

To give you an idea of who were the most heavily worked pitchers in the 2017 season, here are the regular-season Fatigue Unit leaders.

As you can tell, relief pitchers are heavily prioritized by this metric. There are eight relievers and two starters in this top 10. If you’d like to read more about injurious workloads, check out this article on RotoGraphs.

How does the starter’s performance influence relief pitcher workload? What happens to relievers when the starter doesn’t make it out of the fourth inning?

In the 2017 regular season, 3,616 relief appearances occurred when the starter didn’t make it into the fifth inning. Compare this to 11,734 appearances when the starter did make it into the fifth. The proportion of appearances for relievers on different days of rest doesn’t look too different. You can see in the table below there are more pitching appearances where a pitcher had more than four days of rest during these “failed” start, which seems to be when a manager will send in his mop-up guy who might need to get some work in. (As a Blue Jays fan, we call these Liriano days.)

During the regular season, games in which the starter didn’t make it into the fifth inning accounted for 26.5 percent of the total workload experienced by major league relief pitchers. In these failed starts, the average workload for relievers was about 17 percent higher when compared to appearances when the starter did make it out of the fourth inning.

Now, in the 2017 playoffs, we have a much different story.

Starters are getting pulled from the game much earlier, and as a result, 39.4 percent of all relief appearances have come during games where the starter did not get out of the fourth inning. Relievers entering the game in the fifth inning or sooner represent 46 percent of the total workload experienced by relief pitchers in the 2017 postseason. Relievers entering the game in the fifth inning or sooner experienced 23.7 percent higher workloads when compared to relievers entering the game later.

Is bullpenning effective? A lot of the results indicate it may very well be (looking at you, Luis Severino and the Wild Card-game Yankees). Is bullpenning putting a higher workload on reliever arms? It most certainly appears that way.

So, time for a bit of a thought experiment. What would have happened if Verlander didn’t have his stud Game Two performance in the ALCS? What would have happened if he made it through only five innings, leaving Chris Devenski and Ken Giles to pitch an additional two innings each in that game? Where would that leave the Astros bullpen heading into the World Series?

This is what things currently look like. Verlander has the third-highest workload of all pitchers in the 2017 postseason.

To simulate what would happen in this alternate world, I just replaced the sixth- and seventh-inning pitches Verlander threw with Devenski’s name and did the same with Giles in the eighth and ninth.

This move (let’s say Giles and Devenski were awesome and the result was the same, a 2-1 win for the Astros on the back of Carlos Correa’s and Jose Altuve’s heroics) had baseball success–a win, probably some strikeouts, and happy fans going home from Minute Maid Park. Under the hood, this move increased Devenski’s pitch count by workload by 199 percent and Giles’ by 212 percent while reducing Verlander’s output by just 16 percent.

This research has shown just how important elite starting pitching is. In the playoffs, bullpenning is about doing whatever it takes to win the game, right there in the moment. In the regular season, it’s about trying to make up for budget or talent deficiencies. In both cases, it looks like it will elevate pitcher workloads, which may lead to an increased risk of injury.

References & Resources

• Thomas Karakolis, Shivam Bhan & Ryan Crotin, The Journal of Strength & Conditioning Research, “An inferential and descriptive statistical examination of the relationship between cumulative work metrics and injury in Major League Baseball pitchers,” 27(8), 2113-2118
• Mike Sonne, RotoGraphs, “An Introduction to Fatigue Units: A New Method for Evaluating Workloads”
• M. W. Sonne & J.R. Potvin, Ergonomics, “A modified version of the three-compartment model to predict fatigue during submaximal tasks with complex force-time histories,” 59(1), 85-98.