For pretty much as long as I can remember, I've had a strong preference for ground ball pitchers. As the maxim goes, a batter can't hit a ground ball over the fence, and that's especially significant with the Blue Jays playing in a home park where the ball flies very well. Preventing home runs goes a long way towards preventing runs; limiting fly balls is pretty much the only way over the long run to prevent home runs.
Of course, I knew that the most important factor driving pitcher is that the most significant driving factor of pitcher success is strikeout and walk rates, so it wasn't a preference at all costs. A crappy ground ball pitcher is still a crappy pitcher. But those things relatively equal, I expected that a GB pitcher would significantly outperform a FB pitcher; and even with significantly worse peripherals could still be comparable valuable. Think about it: in 2015, a strikeout averages -0.25 runs compared to the average at-bat, a walk averages +0.3 runs, but home runs average +1.40 runs. That means one home run prevented is worth about five strikeouts or walks prevented.
And that's in a neutral park, which Rogers Centre is definitely not. We can quantify this by looking at the home/road split of the pitching staff when it comes to their home run to fly ball (HR/FB) rate. It's the same pitchers each year and the same batted profile, but the balls leave the park at very different rates:
2010 | 2011 | 2012 | 2013 | 2014 | 2015 | Avg | |
Home | 10.2% | 11.5% | 13.8% | 14.7% | 10.9% | 11.1% | 12.0% |
Road | 8.7% | 11.0% | 14.3% | 9.8% | 7.8% | 11.5% | 10.5% |
That's about a 15% difference, which makes a huge difference. So I've always looked at a guy like Rick Porcello and thought he'd look great in a Blue Jay uniform. And likewise, I've dismissed the notion of acquiring fly ball pitchers. Someone like...Marco Estrada, who's been an unexpected saviour for the Jays this year. But when he was acquired in December, the general reaction was quite negative. What were they doing bringing a fly ball pitcher with a big home run problem to Toronto and the AL East?
Now, I knew FB pitchers had some redeeming merits: a generally higher strikeouts since they use more four seam fastballs, they tend to modestly outperform their expected home run rates over time, and they generate more popups which are free outs. But until I looked at Estrada, I didn't realize exactly how much that added up to, and I ended up realizing he could be a quality option for the starting rotation.
That made me rethink my position on GB pitchers and FB pitchers, in terms of just how big the actual difference is. So I pulled all pitchers with 500 starting innings from 2002-15 (as far back as FanGraphs has batted ball data), of which there are 259. As a group, they have standard batted ball profile of 20% LD / 45% GB and 35% FB, so there's there's no bias towards one profile among pitchers in that group.
I defined FB pitchers as one standard deviation above average in FB%, basically 40% or greater. Likewise, with GB pitchers, which worked out to a minimum of 50% ground balls. There were 39 and 33 such pitchers respectively. I also looked at extreme version of both, which were pitchers two standard deviations above average, basically 45%+ FB (six pitchers) or 55%+ GB (10 pitchers). Funny enough, Estrada has 495 starting innings so just missed being included, but he's right at the extreme fly ball threshold.
Below is a summary of the results for each group:
Profile | Avg IP | GB% | FB% | ERA | FIP | xFIP | ERA- | FIP- | xFIP- | ||
Extreme FB | 989.1 | 31.3% | 49.0% | 4.27 | 4.52 | 4.60 | 102 | 106 | 108 | ||
FB pitchers | 1020.2 | 36.2% | 43.9% | 4.16 | 4.33 | 4.41 | 98 | 103 | 104 | ||
Average of all | 1116.8 | 44.5% | 35.3% | 4.13 | 4.13 | 4.12 | 98 | 98 | 98 | ||
GB pitchers | 1172.9 | 54.8% | 26.2% | 4.04 | 4.02 | 3.91 | 97 | 96 | 93 | ||
Extreme GB | 1324.9 | 59.2% | 22.8% | 4.01 | 3.99 | 3.89 | 95 | 95 | 93 |
There's a couple interesting takeaways. First, the GB pitchers piled up more innings than average, and the FB pitchers fewer innings than average. I've seen speculation that GB pitchers tend to fall victim to shoulder problems, but it doesn't appear to be the case.
Most importantly, there is not a big ERA difference between the GB pitchers (4.04) and FB Pitchers (4.16), especially looking at the adjusted numbers (97 and 98 respectively). It is more pronounced among the extreme group, but remember the sample sizes there are very small and individuals pitchers can skew things. It think it's useful to have there for comparison, but caveats apply.
The final point is that the fly balls show a definite ability to outperform their FIP and xFIP, whereas ground ball pitchers lag behind those metrics, albeit more modestly. There's good reason for that, which is apparent from breaking down outcomes:
Profile | K% | BB% | HBP% | HR% | BABIP | IFFB% | PU% | HR/FB | |
Extreme FB | 18.8% | 8.0% | 0.7% | 3.4% | 0.276 | 12.6% | 6.2% | 9.9% | |
FB pitchers | 18.5% | 7.9% | 0.8% | 3.1% | 0.280 | 12.3% | 5.4% | 9.9% | |
Average of all | 17.6% | 7.5% | 0.8% | 2.7% | 0.293 | 10.0% | 3.6% | 10.5% | |
GB pitchers | 16.0% | 7.5% | 1.0% | 2.2% | 0.296 | 7.8% | 2.1% | 11.3% | |
Extreme GB | 14.7% | 7.4% | 0.9% | 2.0% | 0.295 | 7.3% | 1.7% | 11.4% |
Let's start with free outs, which come in two forms, strikeouts and pop-ups. Compared to the average strikeout rate, FB pitcher are about 1% higher and GB pitchers a similar amount under. On the extreme ends, it's end more the case, overall it's a solid linear relationship. But popups are lopsided. FB pitchers have significantly higher rates of popups on balls in air (IFFB%), coupled with more balls in the air means a lot more popups: over 5% of all batters for FB pitchers compared to 2% for GB pitchers. That's what drives most of the BABIP difference. Adding popups and strikeouts together, FB pitchers get free outs 23% of the time, versus about 18% for GB pitchers. That's a huge difference.
The flip side, unsurprisingly, is significantly more home runs. 3.1% compared to 2.2% might not sound that bad, but its 40% more. Because FB pitchers allow fewer fly balls to leave the yard and have fewer balls put in play overall, it's not as bad as the raw difference in fly balls (70% more) .In turn of expected runs, the difference in home runs almost perfectly offsets the difference in free outs.
Interestingly, walks are basically a push. I was surprised that GB pitchers are actually slightly better, with hit by pitches narrowing it further. Line drives rates are basically the same too. So we're left with ground balls and non-HR, non-popup fly balls. And that's advantage ground ball pitchers. Ground balls actually have more damage (30 wRC+) than non-HR, non-popup fly balls (approx. 0 wRC+), but the 15% higher GB rate for GB pitchers overwhelms the 9% higher rate for FB pitchers of those fly balls. That drives a small gap of about 1-3% in favour of ground balls. But that's certain smaller than I thought to be the case, and what I imagine is conventional thought to be.
So, in summary, a ground ball profile has a little more value than a fly ball profile. But it's not a huge difference as fly ball pitchers mostly offset a disadvantage in home runs with several smaller advantages. All in all, there should be a mild preference for GB pitchers, but it makes sense for the Jays to acquire fly ball pitchers like Estrada when they're available cheaply and at a discount to fair value. You just have to learn to live with a few more bombs.