This is the second of a series of posts looking at Jonathan Judge’s new pitching statistic DRA. Read part one – where I looked at the 2016 pitchers that DRA loved – here. As I explained in my previous post, while FIP only controls for fielding and timing, DRA controls for a mess of other factors on top of those.

One thing I noticed from playing around with BP’s leaderboards was that most of the guys that DRA likes are going to overlap with the guys that FIP likes simply because fielding is still the biggest thing that DRA controls for. Fielding is the fat eye in the centre of the DRA/FIP Venn diagram. Usually DRA and FIP tend towards the same conclusion but with varying magnitudes. That’s the case with Michael Pineda: both metrics agree that he was substantially better than his ERA indicates, but DRA goes balls the walls.

Cases like Pineda are intriguing, but what really got me thinking was the cases where FIP and DRA pull in opposite directions. In these cases, unless FIP and DRA rely on completely different sets of fielding statistics as inputs (which is entirely possible), you would assume that even after controlling for fielding and timing, DRA put enough weight in other factors that were out of the pitcher’s control to move the needle in the other direction.

It’s worth noting that in every case where FIP and DRA disagreed, FIP was more bearish than ERA and DRA was more bullish. Logically, that means that these are pitchers who benefited from defence and timing (hence their bearish FIPs) but even so suffered to a greater degree on a combination of other factors out of their control.

I won’t pretend to fully understand the gory math behind DRA, but what follows is a somewhat methodical attempt to explain these discrepancies. In each case, I’ll first use Def (for defence) and WPA/LI (for timing) to justify the FIP, and then I’ll look at the “other factors” that may have led DRA to disagree with FIP. Those “other factors” may include:

  • Park factors
  • The quality of opposition
  • The quality of pitch framing
  • Whether the pitcher is pitching at home or away

The Guys Who DRA and FIP Disagreed On

To identify the cases where this happened, I calculated the ERA-FIP and ERA-DRA gaps for every starter and then subtracted those differences from each other to get an absolute value that indicates the spread between FIP and DRA in relation to ERA. I then sorted that column (Disagreement: ED-EF) and cut it off at ED-EF > 0.5.

disagreement-leaders

After outperforming his FIP by an average of 1.10 runs from 2013-2015, the Chris Young fairy tale finally burst in 2016. He was so bad that it doesn’t really matter just how bad he was. DRA didn’t really rate him as being better than he seemed, it rated him as being less terrible.

1. Cole Hamels is the most intriguing name on this list. My gut tells me luck had a lot to do with this discrepancy given that he pitched for the Texas Rangers, who posted 95 wins despite having a negative run differential. The 3.32 ERA was in line with his career average, but his WAR dipped below 4 for the first time since 2011.

Justifying FIP’s Negativity

  • Defence: Hamels benefited substantially from the defence behind him. The Rangers’ 7.7 Def ranked 13th in baseball.
  • Timing: Hamels graded as barely average by his 0.87 WPA/LI, indicating that he was a worse pitcher by context-neutral standards and benefited from timing.

Justifying DRA’s Positivity

  • Framing: Hamels threw the bulk of his pitches to a mix of Bobby Wilson (69 IP), Jonathan Lucroy (67.1), and Robinson Chirinos (45.1). According to BP’s Framing Runs Above Average (FRAA) stat, Lucroy was above average (+8.8) Wilson (-0.4) was about average, and Chirinos (-7.7) was one of the worst receivers in baseball.
  • Park Factor: Globe Life Park in Arlington had the sixth-highest park factor in 2016 (1.156).
  • Quality of Opposition: Hamels pitched 60% of his innings against .500+ teams.
  • Home/Road Split: Hamels pitched 55% of his innings on the road.

2. Francisco Liriano looked like a completely different pitcher after moving to Toronto at the deadline. He cut his walk rate from 13.5% to 8.9% from the first half to the second half, and bumped his strikeout rate by 6%. Still, his ERA never fully recovered as he posted a career-high 1.44 HR/9.

Justifying FIP’s Negativity

  • Defence: It’s tough to tell whether Liriano got a boost from his defence. The Pirates were the worst defensive squad in the NL (-40 Def) while the Blue Jays were among the best in the AL (19.6 Def)
  • Timing: Liriano’s -2.92 WPA/LI was the third-worst mark in baseball and a career-low. No doubt FIP docked him a few points on account of his context-neutral suckage.

Justifying DRA’s Positivity

  • Framing: You couldn’t ask for better battery mates than Russell Martin (+12.2 FRAA) and Francisco Cervelli (+8.4), but Liriano did toss the other 40% of his innings to a mix of Chis Stewart, Eric Fryer, and Erik Kratz – all of whom are below average pitch framers.
  • Park Factor: PNC Park slightly inflated run-scoring in 2016, and the Rogers Centre was tied with Texas for the sixth-highest park factor in 2016 (1.156).
  • Quality of Opposition: Liriano pitched 56% of his innings against .500+ teams.

3. Carlos Carrasco has yet to surpass 200 IP in a single season. Maybe that reflects well on Terry Francona and the Indians’ pitching staff, but they’ve had no problem riding Corey Kluber when they’ve needed to. Steamer’s projecting 193 IP, and I think that’s generous. In addition to being a constant injury risk, Carrasco lost over half a tick on his fastball and allowed a 169 wRC+ off his heater in 2016. According to DRA, Carrasco was as good as Chris Sale in 2016, but I’m not jumping to buy shares as he carries too much risk at his price tag.

Justifying FIP’s Negativity

  • Defence: There’s no question that Carrasco got a boost from playing in front of Francisco Lindor and the Cleveland defence. The team’s 41.6 Def ranked fifth in the majors.
  • Timing: Carrasco’s 0.45 WPA/LI was his worst mark since breaking out in 2014. It’s fairly average relative to the league, but suggests that he benefited from pure sequencing luck.

Justifying DRA’s Positivity

  • Framing: Roberto Perez earned high marks for his framing work in the World Series, and BP’s stats indicate he was just as good during the regular season (+9.4 FRAA). Buuuuut the Indians’ started the season with Yan Gomes (-5.7) starting behind the plate and also gave Chris Giminez (-3.1) substantial playing time.
  • Park Factor: Progressive Field had the third-highest park factor in 2016 (1.207).
  • Quality of Opposition: Carrasco pitched 61% of his innings against .500+ teams

There’s a common thread here. All of these guys benefited from either good defence, good timing or a combination of both, while simultaneously suffering from pitching the majority of their innings to poor pitch framers against good teams in relative bandboxes. Again, I don’t claim to have a solid grasp on the inner workings of these stats. There are many other factors at work that I don’t know enough about to analyze. That’s why I didn’t look at Chris Sale.

If you trust DRA over FIP, Hamels is an interesting fantasy play for 2017. Brad Johnson had him as the #25 overall SP in his latest rankings at Rotographs, below Felix Hernandez, Gerrit Cole, and Carlos Martinez. Considering he was one of just 15 SPs to surpass 200 IP last season, I think he’s being undervalued.

I don’t think Liriano will be much better than replacement level in 2017, but a full season of Blue Jays defence and Russell Martin framing can only do him good. Carrasco is another guy who will get an automatic boost from throwing more of his innings to a better pitch framer (assuming Perez holds on to the starting gig). He had a ridiculously tough schedule in 2016, and if you’re betting on a velocity rebound and are brave enough to draft him as your #2 starter, he has the upside to provide #1 value.

Advertisements