20 October 2015

Miguel Gonzalez Wasn't A Disappointment This Year

Miguel Gonzalez
Miguel Gonzalez.

Yeah, Miguel Gonzalez.

No pitcher was more widely discussed this year than he. Why? Because he outperformed his FIP to an incredible degree last year, leading to a career-best ERA for a division-winning team, only to follow that up with a stinker of a season marked by a 4.91 ERA. Thus, to many Orioles fans, Gonzalez's 2015 was a disappointment, even despite the slight league-wide uptick in offense. The narrative therefore is that Gonzalez struggled this year and every start became a when-will-he-implode bated breath marathon for Orioles fans, press, and (probably) players.

But this narrative is false. Miguel Gonzalez was basically the same pitcher this year as he was last year. In fact he may have even be a little bit better than he was in the Orioles’ 96-win division championship in 2014. How can this be possible? After all, as I wrote above, his ERA jumped by 1.68 runs per nine innings, which is quite a lot over the course of a season. In his 144.2 innings pitched that's an extra 27 runs crossing the plate, or nearly three whole wins, versus if he had put up the same ERA from last year.
Sure, you may say, but Gonzalez pitches in an offense-friendly park and against better hitters than pitchers in the NL. I agree with you. So let’s look at ERA-, which is his ERA adjusted for park and league and compared to the average pitcher, and where 100 is average and lower is better. (Note that ERA- is different from ERA+.)
  • 2014: 82
  • 2015: 119
Even if you believe Gonzalez should be given credit for pitching a lot in offense-friendly OPACY, it sure looks like his ability to prevent runs took a huge hit compared to last year. He must be broken.

But other numbers tell a different story. Taking a look at Gonzalez’s FIP, we start to see some idea of why his 2014-2015 change is closer to "he was the same pitcher" than "break out the pitchforks, let's go on a rampage":
  • 2014: 4.89
  • 2015: 5.01
If we put all our stock in FIP, then Gonzalez’s 2015 appears to be more similar to his 2014 than if we put all our stock in ERA. Instead of 27 runs worse, by FIP Gonzalez would be only two runs worse than last year. Suddenly, the narrative shifts.

Let's take a slight detour here. Talking about FIP brings me to one area in which Gonzalez improved year-over-year. He increased his strikeout rate by 1 percentage point while raising his walk rate only 0.6 pp. As a result, his K-BB rate increased by 0.4 pp. It was still a not-so-good 9.3% this year, but it tells you that overall Gonzalez fooled more hitters in 2015 than he did in 2014. Yet he is vilified by the fanbase and the press.

The other component of FIP, the main one, is home runs. And here, if we measure by HR/9, Gonzalez performed worse than he did last year:
  • 2014: 1.42
  • 2015: 1.49
Gonzalez allowed just one fewer home run this year despite throwing about 15 fewer innings. But that’s not the whole story. Home runs also have a component of random variation that the pitcher cannot really influence.

As a pitcher who doesn’t strike out a lot of batters, what you want to concentrate on is limiting fly balls. And Gonzalez actually did that this year. His fly ball rate (per batted ball) dropped from 41.7% to 35.8%. This is especially impressive considered he allowed fewer batted balls overall, as evidenced by his increased strikeout and walk rates.

So he allowed fewer fly balls, but gave up more home runs? How is this possible?

Sabermetric research would say that Gonzalez just got unlucky. Before this year, Gonzalez had allowed 62 dingers on 551 fly balls for a HR/FB rate of 11.3%. That's a touch worse than average, but given that Gonzalez pitches a lot in OPACY, it’s understandable. It certainly isn’t awful or terrible. 

Then came 2015. Gonzalez allowed 24 HR in 160 fly balls for an HR/FB rate of about 15%. That’s quite an increase, but again, sabermetric research shows that pitchers generally can’t control extreme fluctuations in HR/FB rate. This is why the metric xFIP, or expected FIP, is useful. xFIP debits the pitcher only for the number of home runs he would’ve given up if he had done so at a league-average rate, given how many fly balls he allowed. For our purposes here using the average rate is a bit too low, but it’s helpful and illustrative anyway.

If we look at Gonzalez’s xFIP from 2014-2015, it shows he barely changed as a pitcher:
  • 2014: 4.46
  • 2015: 4.48
Now we're at the heart of the discussion. xFIP shows why 2015 Gonzalez was very, very similar to 2014 Gonzalez. It posits a difference of only 0.3 runs, compared to 2 runs for FIP and 27 for ERA. 0.3 runs is hardly anything at all.

When you account for all the things Gonzalez should be focusing on as a pitcher -- striking out batters and limiting walks and fly balls -- his 2015 was essentially the same as his 2014, instead of the massive disappointment it seemed to be. The runs that crossed the plate while he was on the mound (or after he'd just been yanked) counted on the board and contributed to actual losses in the standings, but Gonzalez shouldn’t shoulder all the blame for them. He did what he was supposed to do.

There’s a footnote to this story, and for folks like Dan Duquette who are trying to build a contending team, it’s a happy one. While Gonzalez was busy being same ol’ same ol’, the entire league got worse around him. This means that Gonzalez suddenly looks a lot better compared to other arms that could be on the hill for the Orioles.

To see this trend we can look at xFIP-, which adjusts xFIP for park and league while comparing to the average pitcher. Here are Gonzalez's numbers:
  • 2014: 118 (18% worse than league average)
  • 2015: 111 (11% worse than league average)
The year-over-year decrease shows that Gonzalez got better (closer to the average of 100) compared to everyone else. This improvement may be cold comfort to Orioles fans this year because the team isn’t measured in the standings by such relative metrics. And Gonzalez is still not a good pitcher, nor even a league-average one. But for fans looking ahead to 2016, hopefully they are less angry when considering Gonzalez as a part of the team’s plans.

All data for this post taken from FanGraphs.

20 comments:

  1. For me, what is interesting about MiGo is that not only did he perform a run and a half better than what FIP would suggest in 2014, but that he also incredibly outpaced his FIP in 2012 and 2013. As noted in the articles, there has been a lot of thinking about why that is so. I fall tentatively in the camp where this discrepancy exists because of how MiGo pitches in various counts and men on base scenarios. As in he challenged players with no one on base and then works down when men get on. It is a bit more complicated than that, but it basically is about hitting pitches and working the changeup in with respect to count and situation.

    That said, my concern about 2015 is that he lost his command. He was elevating with men on base and cornering himself with poor counts. I am not completely certain on that, but it does suggest that some pitchers are capable of outperforming their FIP with regularity (see: Mark Buerhle). It is rare, but it does seem to happen. From my perspective, MiGo doing it for two and a half years strongly suggests that the outperformance was a skill. Therefore, him dropping back to throwing in line with his FIP is an underperformance for me.

    Moving forward, I think the club needs to be secure with his health and his ability to do what made him successful. If not, a 5 or 6 MM salary on a pitcher giving you a 5.00 ERA is not incredibly useful.

    ReplyDelete
  2. As Jon said, the thing about MiGo is that if FIP explains his performance then he's worthless. His whole value is that he could outperform his FIP.

    His major problem this year was his struggles against lefties with men on base. They seemed to be able to crush pitches whether it was thrown in or out of the strike zone. He gave up five home runs to left-handed hitters with RISP in forty-one chances and probably explains why he got dominated. Why were lefties so successful this year against him as opposed to other years?

    ReplyDelete
  3. I really enjoy reading this site. Thanks for all your thoughtful analysis.

    My concern with MiGo is his performance in the second part of the year. Overall, his stats are quite similar. What happens if you examine these numbers since the All star break?

    ReplyDelete
  4. I guess some might call it all a season of two MiGos. Home runs were not an excessive issue in the second half. Walks were though as well as secondary power. He seemed to largely miss his spots in the second half to a greater degree and his changeup was not as good as it was in previous years. In past seasons, he was pretty tight with his arm action on it and you saw a pitch that varied about 2-3 mph during a game. In 2015, for much of the season you saw him miss spots and have a change that jumped around by 5-10 mph. That inability to hone a motion (which was his strength) is what likely led to his control issues. That all probably has to do with underlying injury, which is what makes me concerned. If he comes out next year and his change is all over the map again, then I would not expect much from him.

    ReplyDelete
  5. Gonzalez wasn't the same since he went on the disabled list with a groin strain earlier in the season. I wrote about it and some of the other concerns for MASN this year: http://www.masnsports.com/orioles-buzz/2015/08/matt-kremnitzer-os-likely-forced-to-hold-steady-during-miguel-gonzalezs-rough-stretch.html

    The shoulder injury is/was more worrisome, as Jon notes, and who knows how he will hold up next year.

    ReplyDelete
  6. Wait a minute. There's one piece missing from the analysis (maybe it makes a difference -maybe not). It was alluded to but not detailed statistically. My gut tells me the Oriole relievers were not stranding inherited runners like they did the year before. Since inherited runners count against the pitcher allowing them not the pitcher they score against, the ERA of the prior pitcher is affected but not the reliever. Since the O's relievers had great ERAs yet we know some of them weren't as good (ref Matusz, Brian), could the increase in ERA for the entire starting rotation be affected by worse relief work? I can't remember how many times Tillman left the mound having given up only one or two runs and his final line showed four or five. Could Miguel's overperformance of his peripherals be enhanced by good relief work in prior years?

    ReplyDelete
  7. Three pitchers were above average...Matusz and Hunter barely so and Brach by quite a bit. Everyone else was above average. I doubt that had much to do with it.

    ReplyDelete
    Replies
    1. That was confusing. Three pitchers were worse and all the rest were better.

      Delete
  8. I think there's some evidence that O's relievers did allow more inherited runners to score than last year, pushing up starters' ERA. I looked at this over at Camden Chat earlier in the year. The evidence is in RE24 for the bullpen - it's a tad lower this year than last year. When a reliever allows an inherited run to score, it's the starter's ERA that goes up, but with RE24 it's the reliever that is penalized. We'd have to look at situational splits to really confirm this, though.

    ReplyDelete
  9. Just raw numbers:
    YEAR / Inherited Runners Scored / Inherited Runners / %
    2012 / 77 / 239 / 32% (average 29%)
    2013 / 72 / 221 / 33% (average 30%)
    2014 / 59 / 238 / 25% (average 28%)
    2015 / 73 / 226 / 32% (average 30%)

    Just looking at that, I doubt we will see anything of note for MiGo.

    ReplyDelete
  10. I knew breach was bad with inherited runners! I could not understand why they would march him out there over and over with 2 on, one out in the sixth or seventh. He was great to start an inning. I'm glad it wasnt in my head but I want to know why he was used in those situations after he kept failing in them. That alone cost us a few wins.

    ReplyDelete
  11. Probably because he's still an effective reliever and is unlikely to be that bad at stranding runners going forward. And you can only use O'Day and Britton so much.

    ReplyDelete
  12. Why does it seem like managers (not just Buck) have a "rule" that they don't replace a pitcher in the late innings until the tying run comes to the plate? That kind of thinking alone creates extra high leverage situations. When I'm watching the games, it seems obvious after the first baserunner allowed in a late inning that the pitcher has "lost it" - especially if it's a leadoff walk. Also, it seems like many relievers take a few pitches to find the strike zone. So if you wait until there are two on base, the probability is higher that the bases will soon be loaded - or worse. I am absolutely sure I saw this several times - especially with Tillman.

    ReplyDelete
  13. Hang on here...... Between 2014 and 2015 there were 12 more inherited runners in 2014 and 14 fewer scored. that's a potential 26 run difference there. That's 3/4 of the way to the 40 runs that would have made this a playoff team..... How can that not be huge???

    ReplyDelete
  14. One...that math is a little screwy. Two...we are talking within the context of MiGo.

    ReplyDelete
  15. Granted - the math was a little screwy because the "12" are runners not runs. But still it seems like 14 fewer inherited runners scoring would significantly affect Miguel's ERA especially as he had more innings. That leads me into a potential micro-statistic. How would the Orioles pitching staff had fared if they hadn't pitched their last inning or partial inning this year compared to last year? Was there something about Buck's pitcher management that changed?

    ReplyDelete
  16. Those numbers are for the entirety of Baltimore pitching, not just MiGo. The resolution of the data, as you mentioned is poor, so there may or may not be something there. It would certainly go into the line of bullpen management and whether a pitcher was pulled too late or early. It really is wide open as to whether their is really anything of significance there. Although we know 14 runs is a lot, we tend to know that from much much larger sets of data. I doubt we have that here.

    ReplyDelete
  17. MiGo pitched 15.8% of all Orioles' starter innings last year. Presuming that relievers gave up a proportionate amount of inherited runners in his starts, that would imply that they cost him an extra two runs. Two fewer runs allowed would decrease his ERA to 4.79. Five fewer would decrease his ERA to 4.60. Does a 4.60 ERA as opposed to one that's 4.90 change the story?

    And given his numbers in the second half with runners on base, he'd probably be better off with Chris Davis taking over for him then actually pitching himself.

    ReplyDelete
  18. And...to add to that ballpark answer...relievers also inherit runners.

    ReplyDelete