15 June 2015

Miguel Gonzalez Is Still Cheating His FIP

In the off-season, a number of Camden Depot writers wrote articles discussing how Miguel Gonzalez has outperformed his FIP. This year has turned out to be no exception so far as he has an ERA of 3.33 and an FIP of 4.60 based primarily upon a .235 BABIP (2nd best out of 106) and an 82.2% LOB (7th best out of 106).

What is interesting is that this year he is outperforming based on performance against hitters when runs are in scoring position. He’s faced 54 batters in that situation and he’s allowed 8 walks (2 intentional), 4 home runs and caused 7 strikeouts. This is pretty awful because pitchers that give up more walks than strikeouts usually struggle and giving up a home run every ten batters on average is awful. He’s allowed seventeen fly balls and given up four home runs for an HR/FB of 23.5%. Simply put, those numbers are dreadful. Two of the remaining thirty-five batters have hit doubles which is pretty reasonable and shows that hitters aren’t struggling to hit for power against him when runners are in scoring position. The area where Miggy has excelled is that he’s only allowed one single with runners in scoring position. That’s absolutely incredible and has allowed him to put up a BABIP of only .083 with runners in scoring position. In addition, he has only given up one unearned run so it’s not like runners are scoring but not counting against him. Batters have been able to put up strong power numbers against him, but can’t seem to hit for average.

With runners in scoring position, Gonzo has a Hard% of 30.8% (75th out of 118 qualified pitchers), which indicates that twelve balls have been hit hard against him. His Med% of 51.3% (51st out of 118) indicates that twenty balls have been hit with medium force against him. His Soft% of 18% (68th out of 118) indicates that only seven balls have been hit softly against him. He isn’t causing weak contact against them, which might explain his low BABIP. Rather hitters are hitting the ball with adequate power but that hasn’t translated to them hitting singles.

ESPNs TruMedia tool claims that Gonzo has thrown 206 pitches with runners in scoring position of which batters swung at 106. Gonzo only had 19 called strikes out of 100 pitches where a batter didn’t swing. Of the 39 pitches put into play, only 8 (21%) were outside of the strike zone.  This seems to indicate that batters aren’t struggling against him because they’re swinging at pitches outside of the strike zone. It seems like they’re swinging at pitches that they’re able to hit.

Part of the reason why Gonzo is doing so well was touched upon by Ryan’s earlier article about Gonzo. Gonzo has thrown a total of 81 combined sliders and splitters with runners in scoring position, allowed 11 batters to put the balls in play and only allowed 1 hit total (a double) with those pitches. By comparison, Gonzo has thrown 13 curveballs with runners in scoring position and allowed both batters to put the ball into play to get hits, thrown 11 changeups and allowed 1 of 3 batters to put the ball into play to get hits and thrown 98 fastballs and given up 3 hits on 23 balls in play. Hitters are simply unable to convert when facing his slider or splitter and have struggled against his fastball. Ryan argued that Gonzo’s improved slider and splitter have allowed him to strike out more batters, but so far it also seems that they’ve helped him by causing a lot of unproductive contact.

The other part of the reason of why Gonzo is doing so well is that he has allowed a low but reasonable six line drives with batters in scoring position of which three were hits. He also has allowed 17 fly balls for which he’s given up four home runs and no other hits, 14 ground balls of which zero were hits and 2 bunts that also didn’t result in a hit. So far, batters have been able to hit fly balls out of the park with a staggering amount of success, but they’ve had limited success when hitting line drives and zero success when hitting fly balls that remain in the park or a ball hit on the ground. Either Miguel Gonzalez has found a way to have batters hit the ball towards fielders or has been the beneficiary of nearly flawless performance by his fielders with batters in scoring position.

It seems that it is unlikely that this will continue. Even if the slider and splitter have legitimately improved, it seems unlikely that batters will continue to struggle as much as they have against his fastball. If the defense has merely over performed, then it is unlikely that they’ll continue to do so to this extent. Then again, a number of writers have been saying for years that Gonzo can’t continue to be this successful and he’s done an excellent job of finding a way to outperform what his peripherals suggest his performance should look like. If some pitchers like Rick Porcello can find a way to mostly underachieve despite strong peripherals then it stands to reason that a few pitchers could find a way to overachieve despite weak peripherals. When it comes down to it, Gonzo has been good the past few seasons and is well on his way of having another strong season. Sometimes it’s just better to be lucky than good and no one is going to care about his FIP if he continues to perform at such a high level.


Ryan Romano said...

Good stuff, Matt. Long live Gonzalez.

Anonymous said...

There's really no reason to believe that the residual of FIP is luck. Beware treating your metrics as anything more than they are, lest you fall into the vicious cycle of confirmatory-search-bias. SABRmetrics has done this with FIP for years, to their detriment.

Related (and worth a read, though most of the site should be taken with a grain of salt): http://lesswrong.com/lw/iw/positive_bias_look_into_the_dark/

Matt Perez said...

The point I was trying to make with the luck statement is that managers won't care about how a pitcher succeeds as long as he does. If Gonzo were to load the bases every single inning but not allow a single run then he'd have a long career because when it comes down to it his teams would win despite the statistical improbability.

But I do think Gonzo has been lucky with men on base. He's given up a lot of extra base hits so it's not like he is giving up a a lot of weak contact. I mean, he's clearly getting hit hard. He's allowed a lot of walks while recording a minimal amount of strikeouts so it's not like guys aren't putting the ball into play. He's just found a way to somehow avoid giving up singles. How can a pitcher that is getting dominated in so many facets find a way to avoid giving up singles?

Anonymous said...

I think you are overfitting.

It is not that he has "found a way of avoiding giving up singles" in particular so much as, as per most of his career, "found a way to avoid giving up hits." It is less-burdened, I think, to suggest that the fact that more of the hits he has given up this year with men on base have been big hits is simply noise and he's essentially the same pitcher as he has been since he started with the Orioles.

Without his track record, I would be inclined to agree that this is not sustainable. Most pitchers are not able to limit hits with men on base. But he has done this, more or less, for three years now.

Matt Perez said...

But he IS giving up hits. Allowing 4 HRs in 54 TBF is terrible. The 2 doubles are at least average. More importantly, he's giving up the types of hits that signify a pitcher is giving up hard contact.

In 2014, he gave up 15 singles out of 138 TBF (10.9%). Interestingly, he also only allowed 3 HRs or fewer than he's giving up in all of 2015. In 2013, he gave up 22 singles out of 145 TBF (15.2%). This is relatively new.

I do agree that he's had low BABIPs with runners in scoring position the past two years. But it's still been on average around the .250 mark. A .083 is considerably lower.

He didn't do anything like this in either 2012 or 2013.

Anonymous said...

The "types of hits that signify a pitcher is giving up hard contact" are subject to a lot of variance, though. How many of his outs were hit hard? Over this sample size, do you have any reason to think that he hasn't simply had a large number of weakly-hit balls to go with his 6 hard-hit ones? This is what I mean by overfitting.

If we had access to the hitFX data, I think one could do analyses like this with a lot more confidence (we could, for example, examine whether his outs have come on hard-hit balls). As it stands, I'm not sure there's really much here, to be honest.

For the record, I don't think .083 BABIP is sustainable by anyone under any circumstance. I also don't think that you're going to see this rate of home runs allowed w/ RISP for the rest of the year, either. Put me down under "he'll end the year with numbers that look like his career numbers" until there's really a compelling reason to predict otherwise.

Matt Perez said...

First of all, you'd expect some weakly hit groundballs to go for hits or some weakly hit fly balls to go for bloop singles. That hasn't happened (provided you trust the contact categories defined by ESPN). But the Soft% data reported by BIS shows that he's not giving up a lot of weak contact.

What I wanted to do is look at each of the 39 balls put into play and see whether there was good defense. At 30 second a play it would take roughly 20 minutes. Unfortunately, that ESPN tool appears to be broken.

I agree that the sample size is small. And I'm not sure there's so much here either. We know that these are some really odd results for 39 balls in play. I'd probably expect his numbers to look like his career numbers as well and that this is just what happens when dealing with limited cases. But it's always fun to see an outlier.

Anonymous said...

Fair enough. I do think that the only real way to do analysis on small trends like this really is to look at detailed information on every data point. But, yes, this is a pretty weird stretch.