Two Nuggets of Information About Astros’ pitching

Do I like culling through statistics and finding out how they improve our understanding of baseball? Yes, guilty as charged. But most of the readers here already knew that.

Sometimes it can be like sifting through sand to find something interesting. So maybe that’s what we mean by “nuggets of information.”

In this article, I will look at some statistics that might help us understand the Astros’ pitching in some small way. First, I’ll look at pitching z-statistics in a recent Fangraphs article, which may help us predict regression. Second, I’ll turn to Baseball Reference to see what may affect the bullpen’s workload.

Fangraphs’ z-stats through Aug. 10 for pitchers

Even though we are more than 60% of the way through the season, that is still not an overly reliable time period or sample to understand whether a pitcher has under or over-performed. Why do we want to know the answer to that question? We are more interested in what the pitcher will do in the future (e.g., the remainder of the season) than what occurred in the past. It is helpful to know whether current performance will regress in the future.

Dan Symborski, who developed the ZIPS projections, wrote an article at FanGraphs this week regarding pitcher z-stats through Aug. 1, which he calls “leading indicators” of future performance. The z-stats are used to determine pitcher “over” and “under” performers for the three true outcomes (strikeouts, bases on balls, and home runs). The irony is that underlying components can predict Ks, BBs, and HRs better than the actual 3TO events that occurred.

A walk or strike out is the culmination of a number of events during an at-bat—taking strikes and balls, fouling off pitches, and swinging and missing—and the underlying plate discipline stats like “first pitch strike” and various measures of swing% and contact%, can be a better predictor as to whether the existing K or BB rate will improve or regress. That is the concept of the z-stats. The z-stats in theory are more predictive of future performance than previous incurred events during the season.

So, let’s get to the parts of the article which pertain to the Astros. In the article, “overachievers” means the performance is better than the z-stat prediction through Aug. 10, which in turn implies a potential future decline in performance. And vice versa for “underachievers.”

Framber Valdez is an over achiever on z-HR, z-BB, and z-FIP. Accordingly, z-HR says his HR rate through Aug. 10 should have been 3% instead of 2.3%. Valdez’s BB% should be 3% higher. Finally, his FIP (3.29 on 8/10) should be one-half run (0.5) higher.

The regression seems to have already begun, as evidenced by Framber’s two HRs allowed in Monday’s game. After that game, his FIP increased from 3.29 to 3.44. Valdez’s HR/9 rate in August is almost triple the HR/9 for the first half of his season. The tweet below, after Monday’s game, reflects the regression in Valdez’s HR tendencies.

As I discussed in last week’s article, Valdez has allowed high exit velocities this year, particularly in recent months. This is another factor that can lead to more HRs. It is reasonable to expect Valdez’s ERA to regress over the remainder of the season—the unknown is whether it will be a hard regression or relatively mild. Fangraphs depth chart projects a 2% increase in Framber’s ROS ERA, which would be mild.

J.P. France and Cristian Javier join Valdez in the list of BB% overachievers. Normally this would imply a reversion in their walk rate. But the article raises the possibility that the Astros have found a way to improve their suppression of walk rates: “ I have to wonder if the Astros have an approach that other teams haven’t figured out yet; they have three pitchers who rank very highly here despite unimpressive first-strike percentages. Usually, this number is a leading indicator of walk rates, but J.P. France, Cristian Javier, and Framber Valdez don’t fit the usual pattern.”

It’s worth noting that J.P. France is on both an under-achiever and over-achiever list for strikeouts and walks, respectively. Given the weights applied in the FIP formula, the impact should be pretty close to offsetting any effect on France’s FIP. However, France’s current 4.03 FIP could indicate a regression as high as 47% in ERA. France’s x-ERA of 4.23 seems to support that possibility.

Hunter Brown made the list of strike out overachievers. z-SO indicates that his K rate should be 3.9% lower. Some regression has already taken place in August; he had his lowest monthly K rate of the year. Despite this potential negative result, most signs are pointing to reversion in the direction of a lower ERA. Brown’s x-ERA (4.00), FIP (3.96), and x-FIP (3.29) would seem to foretell an ERA lower than his current 4.23.

Bullpen Usage/Workload

Bullpen workload has been a frequently cited Astros concern. This was the reason that the Astros actively engaged in the relief pitcher market at the trade deadline, acquiring Kendall Graveman and leaving at least one additional reliever trade on the table when San Diego opted to become buyers rather than sellers.

A complicating factor for analyzing bullpen usage is its close relationship to starting pitcher usage. The Astros’ preference is to structure a rotation that pitches deeply into the game, thereby reducing bullpen workload. However, if a starting pitcher fails to pitch deeply into the game, that can upset the bullpen apple cart.

Whether the Astros’ bullpen is overworked or not depends, in part, on whether usage is measured by innings or appearances. The Astros’ bullpen has the third-lowest total innings pitched and third-lowest innings pitched per game. The Astros are somewhat higher (ranked ninth) in total appearances.

But these team stats don’t tell the whole story, because the Astros starters pitch more deeply into the game than most teams, which means that fewer appearances and innings will be required. A more granular view indicates that workload concerns may exist for certain pitchers.

Several of the Astros relievers have high rankings in appearances but far lower rankings in innings pitched. Bryan Abreu has the 2d highest number of appearances (56) in the majors but is ranked 51st among relievers in innings. Phil Maton is 9th ranked in number of appearances but is ranked 60th in reliever innings. Other Astros reliever appearance rankings: Neris, 21st; Pressly 33rd; Graveman 35th; Montero 59th; and Stanek (105).

Based on appearances, the Astros’ reliever usage is concentrated in the top four of Abreu, Maton, Neris, and Pressly. To some degree, I think this reflects the Astros’ confidence in those pitchers' ability to handle high-leverage situations. The Astros bullpen faces a higher than average average leverage index (LI) than other teams. The average LI for the Astros’ bullpen is 9th, which probably means the bullpen is dealing with more late and close games. This is shown by the Astros’ 9th rank in the number of tie games when a reliever enters.

Ryan Pressly’s LI (1.79) is ranked higher (22d) than the number of appearances (33d), which is probably appropriate for a closer. Neris and Abreu also have higher than average LI (above 1) and are ranked 52nd and 60th, respectively.

The lack of multiple-inning relief appearances partially explains the Astros relievers’ higher ranking in appearances vs. innings. Some statistics to support this position:

• The Astros are ranked 30th (last) in number of relief appearances with more than 3 outs. The Mariners are ranked 29th.
• The Astros and Mariners are 30th and 29th in relievers pitching more than 1 inning. The Mariners are 30th and the Astros 29th in innings per relief appearance.
• The lack of multiple-inning appearances is explained in part by starters pitching deeply into the game. But it also reflects a difference in usage for late-inning relievers. Some teams routinely use 7th and 8th-inning pitchers for more than one inning, while others like the Astros and Mariners, do not.
• Even though the Astros’ bullpen has a lower workload in terms of total innings, the Astros are ranked above average (9th) in the number of reliever appearances with 0 days of rest.

The tendency to use relievers in one-day stints contributes to appearances with fewer days of rest. The downside of multiple-inning appearances is that relievers are more likely to be unavailable longer.

The Astros’ bullpen is not well-stocked with long relievers. Currently, Parker Mushinski and Seth Martinez (who exchanged positions in Sugar Land) fill the multiple-inning role when they are on the active roster. Ronel Blanco can also fulfill that role if he is recalled after Sept. 1. However, the Astros seem to be moving toward a partial six-man rotation. I use the term “partial” because J.P. France and Hunter Brown have taken turns acting as a potential multiple-inning reliever in the bullpen. This may become the standard practice in order to constrain bullpen workload, as well as reduce starter innings pitched by Brown, France, and Javier.

I am less concerned about the bullpen workload since the trade for Graveman. He should help reduce the high number of appearances by Abreu and Maton. It also may help the late-inning workload to work Ryne Stanek into more 7th or 8th-inning appearances. Stanek’s average LI is well below one, indicating that he isn’t used to high-leverage situations. But he has performed well lately. He has allowed a zero OPS over the last 14 days. He has allowed only a .363 OPS over the last 28 days.