Twins Pitching Woes: 2013 and 2014

Thrylos, do you think Pelf or Milone is currently in the lead for the 5th starter spot? Did Trevor May do enough in his four innings of perfect baseball today to put himself back in contention?

A great article was written this week by Jonathan Judge and published at The Hardball Times, " FIP in Context." This introduces an new metric, called cFIP, or contest-adjusted FIP that attempts to "estimate the pitcher’s true pitching talent during a particular season".I'm always interested in new pitching metrics development, and not only because I have myself taken part in the endeavor. This is an interesting metric, albeit much more complex than PE and xPE. It also correlates well with SIERA, which along with xPE (because it is easy to calculate) are my two favorite predictive metrics regarding pitching performance.

I will not steal Jonathan's thunder, please read that excellent article, but I will present his framework and then present his work as applied to the Twins' pitchers. (He calculated cFIPs for every pitcher in the league for the past 4 years, including Jamie Carroll.) The cFIP scale is normalized to 100 for average, just like OPS+ and ERA+, but it is a minus scale, meaning that less is better, like ERA and FIP and SIERA and all similar metrics. (It should have been called cFIP-, but that is a different story.) So 100 is average and less is better. Jonathan Judge has the following buckets of pitchers, according their cFIP:


70 and less = superb
70–85 Great
85–95 Above Avg.
95–105 Average
105–115 Below Avg.
115–130 Bad
130+ Awful

Let's put the 2014 Minnesota Twins' pitching staff in those buckets. For reference, players that are not still with the team are in parenthesis. I am also including the 2014 cFIP numbers of the newcomers this off-season. They have an asterisk behind their names.

Superb:
Phil Hughes 70

Great:
Glen Perkins 74

Above Avg.:
Casey Fien 89
Tim Stauffer 91*
(Yohan Pino 94)

Average:
Aaron Thompson 98
Logan Darnell 99
Ricky Nolasco 100
Trevor May 101
Ervin Santana 101*
Michael Tonkin 102
Caleb Thielbar 103
Blaine Boyer 103*
Lester Oliveros 105

Below Avg.:
(Jared Burton 106)
(Kris Johnson 106)
(Sam Deduno 107)
Stephen Pryor 108
Kyle Gibson 109
(Anthony Swarzak 111)
A. J. Achter 112
Ryan Pressly 112
Brian Duensing 114
Tommy Milone 114

Bad:
(Matt Guerrier 116)
(Kevin Correia 119)

Awful:
Mike Pelfrey 132

A few observations:

• According to this, in 2014, the Twins had one superb pitcher, Phil Hughes, one great pitcher, Glen Perkins, and 13 pitchers (that is a full MLB staff, ladies and gentlemen) who were average, above average, great or superb. Mike Pelfrey (who tied for worst in the majors in this metric) was the only awful pitcher in the Twins' staff.
• But, the Twins had the second worst bullpen in the majors according to xFIP and SIERA and the third worst rotation in the majors, according to SIERA (fourth according to xFIP).
• Other than Yohan Pino, who was an unfortunate loss, the Twins' front office seems to behave pretty well according to this metric. The pitchers they let go were all below average or worse. They did keep a few below average pitchers, and they did keep Mike Pelfrey, who is better suited for the pen and was injured. Other than Duensing who had a down season, the other below average pitchers are all young.

The big issue in the big picture here: The Twins had a whole staff worth (13) pitchers who were average and above, yet they managed to be almost at the bottom of the league in pitching. Those things seem to be in conflict.

Let's dig deeper and check out the 2013 Twins' cFIP buckets that Jonathan Judge calculated. For reference purposes, players who left after 2013 are in parenthesis and I added Ricky Nolasco (with an asterisk) as well.


Superb:
Glen Perkins 63
Casey Fien 67

Great:
Nobody

Above Avg.:
Jared Burton 91
Caleb Thielbar 91
Ricky Nolasco 93*
Michael Tonkin 94

Average:
Anthony Swarzak 97
Brian Duensing 97
(Shairon Martis 105)

Below Avg.:
Mike Pelfrey 109
(Liam Hendriks 110)
Ryan Pressly 111
(Cole DeVries 114)
(Andrew Albers 115)

Bad:
Kevin Correia 116
Samuel Deduno 116
(Josh Roenicke 118)
(P.J. Walters 122)
(Vance Worley 124)
Kyle Gibson 125
(Scott Diamond 129)

Awful:
Nobody

This is some really interesting data. Here is what I see:

• I think that I either underestimated the Twins' front office use of metrics in personnel decisions in building the team or Jack Goin should buy me a beer this week at Hammond Stadium, because this tool really describes what the Twins are doing regarding personnel decisions. Recently, they have tended to get rid of below average and worse pitchers and add average and above pitchers. Hughes was around 100, but I did not add him here. This is a stop the presses type of statement. I'm close to nodding my head in approval of what the front office is doing.
• For 2013 this tells a tale of two categories: all the average and above pitchers were relievers. All starters were below average or worse but not awful. And Pelfrey was the best of that lot.
• Enough with 2013. What happened in 2014, relative to 2013? Every single reliever from Perkins down regressed, while the starters (save the injured Pelfrey and replacement-level Correia) improved. This is fundamentally interesting, because it breaks some commonly accepted narratives. One excuse for the decline of the Twins' pen in 2014 was that they were too tired because the rotation was so bad. This data turns this upside down. The Twins 2013 rotation was worse than the Twins 2014 rotation, and the 2014 Twins' pen made the 2014 Twins rotation worse. So a bad pen can make a rotation worse. Like a reliever coming in with two outs and the bases loaded to give a grand slam and four runs to the starter's record. What a concept...

I am starting to really like this metric. So, and this is really hard for me to say, the front office did make some improvements for 2014 that actually seem to be supported by real data, but the pitching tanked compared to 2013. Why?

I'd love to hear your theories after this, and this is what I am thinking:

• Look at that 2014 list up there. In your mind, normalize it for playing time. That would shift the buckets heavier to the below average. How many games did Pino or May start compared to Pelfrey, Correia, Deduno? Why was Burton used in high leverage situations over better relievers? Yes. Do the same normalization for playing time for the 2013 data. And you are looking at evidence of what has been written here, loudly and clearly, about mismanagement of the Twins' pitching staff by Gardy and Andy.
• This has to be part of the reason for the pen decline in 2014. And it had to be fixed. And, yes, metrics can be devised to normalize and approximate defense independent pitching, but I have not yet seen one that could estimate the madness of the Twins' 2014 OF. (What is the range factor of a bucket? )

This actually makes me more hopeful, because it seems like the Twins are making an effort to address some things that need addressing. So, what do you say?

Click here to view the article

 

Why should I buy you a beer for finally realizing we use stats in personnel decisions more than you realized?  I think you owe me a beer. I drink Leinie's.