Dusty! Use This for Selecting Relief Pitchers!

One of the biggest criticisms targeted at Dusty Baker is his use of the bullpen.

Having worked on the following for over a week I have constantly seen the Astros violate the recommendations here and then watched the sub-optimal reliever fail. It is extremely difficult to constantly watch bad decision be made and knowing there is a better way.


Out of utter frustration, I decided to give a guide of how the relief pitchers should be deployed based on an overly simplistic model of:

1. Batting Situation- are the next three batters LH or RH or a mix and what is the profile of the potential PHs?

  • All three hitters are right-handed, and no left-handed strong pinch hitters are available

  • Two of the next three hitters are right-handed

  • The next few batters are a mix of right-handed, left-handed, and evenly split switch-hitters

  • Two of the next three hitters are left-handed

  • All three hitters are left-handed, and no right-handed strong pinch hitters are available

  • For this analysis, switch hitters in this system would be assigned a “side” based on if their performance is better in one direction. If not, perhaps the middle scenario should be chosen

2. Game Situation/ Leverage

  • Extremely High (EH) - 8th- 9th inning 0-2+ Runs

  • High (H)- 6th- 9th inning -2 to 4+ Runs, extra innings

  • Low or Early (L, E)- <6th inning, 6th- 9th inning <-2 to >4+ Runs

  • Blow Out (BO)- 4th- 9th inning <-5 to >7+ Runs

The leverage wording may be a challenge, so perhaps this table is more clear to show how we are applying leverage in this context.

Before we lay out the usage tables let’s look at the bullpen wOBAs vs. RHH and LHH and how those rank among all of MLB’s relief pitchers. (Data from Fangraphs.com)

I have added Javier and Urquidy into the numbers here to illustrate how they could be important to a post season bullpen. The comparison data was set to a minimum 15 IP vs RHH (therefore 235 RP in data set) and 10 IP vs. LHH (therefore 240 RP in data set.) The numbers for Baez are set to his career average until he can pitch a statistically relevant number of PA.


Conclusions

  • Some of these split numbers are really bad for some of the relief pitchers. This shows how the Astros bullpen will be challenged to be an elite bullpen until Javier and Urquidy are fully available and until Baez pitches to his historical level (if he does.)

  • Urquidy and Javier may be as critical to the 2021 postseason bullpen as other recent starters have been in the postseason.

  • The Astros have many pitchers that have huge platoon splits, and this REALLY limits their usage in leverage situations. Basically, only Pressly can be fully trusted to get LHH AND RHH out at an elite level.

  • I said at the trade deadline that the Relief Pitchers that the Astros got needed to be better than Abreu (RHH 0.291. LHH 0.321).

  • By this metric, Montero and Maton are not.

  • They aren’t better than the career average of James who is now blowing away hitters in AAA.

  • James Click was quoted as prioritizing getting pitchers that could get RHH out. Graveman and Yimi Garcia do that but Montero and Maton don’t.

  • Problem Areas

  • Raley and Taylor vs. RHH- The Astros LHRPs are some of the absolute worse against RHH. This dramatically reduces the capacity to confidently use either in leverage situations. Taylor is elite elite vs. LHH and could be deployed in 2oo3 LHH situations.

  • Maton can strike people out and that is cool. He is bad when batters make contact (BABIP 0.344) and his 4.35 BB/9 needs to improve significantly. His K/9 for 2021 is 12.02 but only 6.97 with the Astros. I believe Maton will be sent to Sugar Land when Urquidy is available. Abreu may be given a September call up spot as well.

  • Ryne Stanek vs. LHH- Go back and watch when Ryne Stanek has had problems. Most of the time the big hit is coming against a LHH. Stanek cannot be used in high leveraged situation in an inning where a LHH can tie or win the game.

  • Yimi Garcia and Pedro Baez may deserve more opportunities in mixed type situations (both RHH and LHH)

So how would I manage the bullpen? The following table shows the way I would deploy the Astros relief pitchers based on the upcoming batters and the game situation.


What may be even more important than what I would do is the following DO NOT use chart for the extreme high and high leverage situations. Staying away from these traps are critical to not blowing games.



Notes regarding these recommendations:

  1. I have chosen to be overly simplistic in this because as I have trial run this over the last week, I have seen the recommendations here be ignored and then seen the pitcher fail.

  2. Part of the simplicity here is to base the decision matrix on wOBA. I believe wOBA gives the best synthesis of ERA, FIP, WHIP, and other metrics to give the best options. I believe that wOBA is also an early warning indicator that would be needed here. Basing this matrix purely on ERA would not be wise.

  3. The mixed platoon wOBAs are based on the weighted average of the RHH and LHH wOBA stats.

  4. I used career data for Baez in this data set. As he gets to ~10 IP for each RHH and LHH or if his performance indicates clear trends, this can be updated.

  5. My tables could lead to an overuse of the same pitchers over and over

  6. Obviously, part of the key here is to NOT use the high leverage pitchers in low leverage situations- ever.

  7. The table includes several options for most cases and if the positive table is exhausted, then the answer is to NOT use one of the pitchers on the negative table.

  8. My tables basically show Montero and Maton should be rarely used. Yes, that is correct. Their performance backs up this conclusion. If their data improves by being used in low leverage situations, then their recommended usage would increase.

In an ideal world, these tables might be step one on a decision path that looks something like this

  1. Are multiple innings needed? If so, using the mixed batting situation recommendations would be best.

  2. What is the Batting Situation for the next three batters? If any of the batters have career and 2021 reverse splits treat them as the opposite handed batter.

  3. All three hitters are right-handed, and no left-handed strong pinch hitters are available

  4. Two of the next three hitters are right-handed or a LHPH is available

  5. The next few batters are a mix of right-handed, left-handed, and evenly split switch-hitters

  6. Two of the next three hitters are left-handed or a RHPH is available

  7. All three hitters are left-handed, and no right-handed strong pinch hitters are available

  8. What is the Game Situation/ Leverage

  9. Extremely High (EH) - 8th- 9th inning 0-2+ Runs

  10. High (H)- 6th- 9th inning -2 to 4+ Runs, extra innings

  11. Low or Early (L, E)- <6th inning, 6th- 9th inning <-2 to >4+ Runs

  12. Blow Out (BO)- 4th- 9th inning <-5 to >7+ Runs

  13. Find the group of pitchers recommended for both criteria and eliminate any unavailable due to workload and needed rest

  14. Does the game situation dictate a strikeout or two is needed or are the next batters easily prone to strikeout? Use the pitcher among the group recommended with the highest K/9

  15. Are the next batters power hitters? If so choose the recommended pitcher with the lowest HR/9

  16. If none of the recommended pitchers are available

  17. Chose a pitcher NOT on the Do Not Use table or

  18. Find the table with wOBA for all RP for the batting situation and choose the lowest wOBA pitcher available.

All of these would complicate and possibly adjust what is a general answer from the first table. I believe the challenge for many Astros fans is that answers are not really given to explain the decision that have been made even when those decisions fail and were clearly wrong by this system. Therefore fans, especially those with at least a general idea of these tables, are left to believe the decisions that are being made are just unwise.


Watch a few games with these tables nearby. It may revolutionize how you think about bullpen usage.


As always, PLEASE give me your feedback.

84 views0 comments

Recent Posts

See All