The Cause of Trevor Cahill’s Downturn

By virtually any metric, it’s hard to look at Trevor Cahill‘s recent performance and conclude that he’s likely to be a good value at his $12M salary for 2015. But that doesn’t mean that he’s not worth more than the $520k the team would be likely to recoup by releasing him. As recently as 2013, Cahill was arguably an above-average starter — and although the results didn’t match, he was substantially the same pitcher in 2014.

Cahill’s 5.61 ERA in 2014 came in 110.2 innings, meaning he bears a pretty big chunk of the blame for the team’s 4.26 ERA last year, which ranked the D-backs 5th-worst in baseball. If you back Cahill’s 7.7% of team innings out of the mix, the team staff would have had a 4.17 ERA, which would have left them ranked…5th-worst. The D-backs staff was, let’s say, not so good regardless of Cahill.

In fact, as we have repeatedly documented but not really explained, the vast majority of the pitching staff underperformed last year, both in terms of our 2014 expectations for each pitcher and in terms of how even the intermediary results that actually played out ended up turning into runs. SIERA ranked the D-backs staff 10th best. The only staff with a bigger negative difference between SIERA and ERA was the Rockies.

Cahill was not so great last year, and it wasn’t just about the very rocky start (7.09 ERA in April) — his 5.66 ERA in the first half was almost identical to the 5.58 mark he put up in the second (in substantially more innings). But the point is that, while he underperformed in a fairly significant way, the rest of the staff did, too. As compared to SIERA (3.59), the staff ERA (4.26) was marked up by about 20%. As compared to SIERA (3.96), Cahill’s ERA (5.61) was marked up by about 40%. The wheels came off for Cahill, no question. But be it pitch sequencing, defensive inefficiency, something different with Chase Field or a combination of those and other things, it seems like a big part of Cahill’s struggles had little to do with Cahill.

Let’s see if we can determine how much of Cahill’s 2014 struggles were due to luck, and what, if anything, we can point to that was different. But as a backdrop, let’s also keep in mind that in terms of the stats that go into Fielding Independent Pitching — walk and HBP rate, strikeout rate, home runs — Cahill actually wasn’t worse. At all.

Behind the Numbers (Are More Numbers)

Armed with these ERA, FIP and SIERA numbers, we can start to identify likely causes for Cahill’s 2014 problems. Given that his FIP actually went down, we can make a good guess that Cahill’s total walks, strikeouts and home runs weren’t the culprit. Since SIERA matches up with FIP, I would normally suspect that batted ball rates also aren’t to blame — although those changed so much in 2014 that I will keep them in the mix. And SIERA’s consistency with FIP helps lock down the idea that although Cahill’s walk and strikeout rates per 9 IP did climb in 2014, that had a lot to do with him facing more batters per inning. The jump from 3.99 BB/9 to 4.47 BB/9 would be meaningful but for the fact that the “9” part changed — on a batter-for-batter basis, Cahill’s BB% rose only from 10.2% to 11.0%.

Unless I’m missing something, the remaining explanations fall into three buckets:

1. Something Cahill was doing that led to his pitches getting hit harder more frequently (not luck).

2. Balls falling in for hits more frequently than we would expect them to (luck, or defense).

3. Inconsistency; SIERA and FIP have guessed wrong because they’ve assumed that Cahill’s bad stuff weren’t as lumped together as they were.

Line Drive Rate

Trevor Cahill saw his ground ball rate, typically very high for a starter, sink to 48.5%, below 50% for the first time since 2009. And most of the missing hits weren’t fly balls, although there was a bit of an increase there — it was Cahill’s LD% that changed the most (proportionally).

See, Cahill has generally had a good LD%. His 2013 LD% was 19.8%, the highest of his career to that date but still not bad; last year that rate would have tied him with Justin Verlander and Madison Bumgarner at 31st among 88 qualified pitchers. Even now, his career LD% is 18.2%, which would have ranked him 8th last year out of 88. In his first year with the D-backs, Cahill had a 16.1% LD% — and last year, that would have ranked him 1st out of 88, ahead of Alex Cobb‘s 16.4%.

Last year’s LD rate wasn’t just worse. It was really bad: 24.4%. Considering that line drives turn into hits about five times as often as fly or ground balls, that makes an enormous difference. And for what it’s worth, only one of those 88 qualified starters last year had a rate that was worse (Ervin Santana, 24.7%). Not good.

I checked in with ESPN Stats & Info to find out: is this just a matter of batters hitting Cahill’s pitches more level, or is this more a matter of Cahill’s pitches getting hit more hard? Turns out that while Cahill’s Hard Hit Average of .173 wasn’t awful (Josh Collmenter‘s was .191, Wade Miley ranked more or less in the middle with .156), it was still pretty bad. That .173 HHAV would have tied him at 77th with two other pitchers among those same 88.

The weird thing, though, is that while the .173 HHAV is not good, it was exactly the same in 2013. And it was higher (.207) in 2012, by a very significant margin. Cahill was not having the cover torn off of the ball in 2014 — not any more than in the past. The change in LD rate was more a matter of launch angle (although, let’s be clear, we’re only talking about a difference of about 12 balls hit into the field). Putting aside (for today) the question of why Cahill was giving up fewer ground balls, we can at least say: Cahill’s downswing was unlikely to be the product of players teeing him up.

Luck or Defense

The 84 line drives that Cahill gave up last year turned into hits (when they were in play; two were bombs) at a .732 clip (BABIP). Compare that to his .253 BABIP on ground balls, and .139 BABIP on fly balls (not including 7 homers). Hits aren’t good for a pitcher — and it looks like Cahill was giving more up.

Compare to his 2013 BABIPs: .674 on line drives, .211 on ground balls, and .175 on fly balls. Literally every kind of batted ball fell in significantly more frequently in 2014 than they did in 2015. And yet we have an indication from the ESPN Stats & Info hard hit stats that it’s probably not the case that every kind of hit was being hit harder — especially since Cahill’s hard hit rate didn’t go up.

Putting it differently: if Cahill’s downswing was caused by a dropoff in defensive efficiency and/or poor luck, this is exactly what it would look like.

Let’s turn to defense. A simple way to check on this is to see if the entire pitching staff saw a similar BABIP change on each type of batted ball; if that’s out of whack with HHAV, that difference would not be the fault of the pitchers, and it’s almost impossible that an entire staff was significantly unlucky for an entire season.

This isn’t a perfect method, in that it includes Cahill — whose effect on total innings is smallish, but whose big swing in BABIP versus HHAV is something we’ve already noted. In addition, the method is a lot less telling with respect to fly balls, since HHAV would include home runs on fly balls but BABIP wouldn’t. Still, it doesn’t look like there’s anything here.

To blame defensive efficiency, we’d be looking for changes in BABIP to be more positive than changes in HHAV. Instead, it looks like the D-backs were significantly more efficient at converting line drives into outs in 2014 than they were in 2013. We can’t say the same about ground balls — the team did do a little worse. But it looks like the team more than made up for that in being much more efficient at turning fly balls into outs (thanks, A.J. Pollock and Ender Inciarte!). There were more ground balls than fly balls, but the difference almost completely evens out.

So it’s probably not defense. If there’s a way to prove bad luck, I don’t know what it is; Cahill did have a left on base (LOB) percentage of 62.6%, which is quite a lot lower than his career LOB% of 71.8% or his 2013 LOB% of 73.5%. If it’s luck, we can really only point timidly in that direction, and only if we can more or less rule everything else out. But while the LOB% would be what it is if luck was at play, it’s also possible that we can blame Cahill for it.

I’ll share one other indicator that luck may be at play: Cahill’s results on ground balls with runners in scoring position. Remember, Cahill’s BABIP on ground balls rose a fairly significant 29 points between 2013 and 2014, but with runners in scoring position, BABIP on ground balls jumped from .216 to .390, a difference of about 5 hits in the 2014 context — those were almost all undoubtedly run scoring plays. And Cahill’s BABIPs in those situations had hovered in the low .200s in previous seasons, as well.

Inconsistency

I’m probably not alone in feeling like Cahill was a totally different person for different stretches, even within the same games. It seemed like either he was on and it was all working, or he’d lose the sinker and he’d look like any out was the result of divine intervention. But let’s see if the numbers back that up.

If this was a matter of losing command, walks — or better yet, balls — would probably be where we’d find that. I’m frankly not sure how to measure a pitcher’s “lumpiness” in terms of walks or balls, let alone how to compare whatever we came up with against the rest of the league. But we can at least make some educated guesses.

A walk here and there is not such a big deal; the only way to prevent them is to pitch in the zone almost exclusively, and that would be a poor strategy unless your stuff was literally unhittable. But sometimes, one would think, a pitcher would care more about walks. With runners on, for instance.

But we don’t find that with Cahill. In 2013, Cahill walked 9.5% of batters he faced when there were runners on, 8.3% in 2012. In 2014, he was right at the same level: 8.7%. It does appear, though, that this was a good thing to look at — the fact that all of these walk rates are a touch lower than his walk rate overall seems to suggest that he was trying to avoid walks in those situations.

I have no way of comparing this total to league average (that I know of), but I did go through the game logs to find instances of Cahill throwing multiple walks in the same inning:

3/22, 5th inning (walked two, then pulled in favor of Collmenter)
4/13, 3rd inning (walked three — first two walked runners stole a combined three bases in three separate plays)
4/26, 8th inning (after Joe Thatcher put a runner on with no outs — ROE — Cahill came in as a reliever and walked Marlon Byrd, sandwiched 4 hits around a strikeout, then walked another, but intentionally)
5/21, 12th inning (walk, double, and a very understandable intentional walk with the walkoff run on third)
5/25, 8th inning (walked two, then pulled in favor of Thatcher)
6/5, 7th inning (came in as the third pitcher in the inning with two outs, walked two, was pulled before third out)
8/15, 3rd inning (after walking Donovan Solano with one out, walked Giancarlo Stanton then closed inning with GIDP)
8/26, 4th inning (two walks to open inning, strike out, then five consecutive hits before getting pulled)
9/1, all game (in a six-walk, 4.1 IP performance, Cahill gave out two free passes in the third and two more in fourth — but, oddly, his last walk of the game was one he got as a hitter in the top of the fifth before giving up two hits to lead off the bottom of the fifth, getting yanked)
9/7, 6th inning (a homer-and-walk sandwich)
9/26, 2nd inning (two walks; first player scored on a GIDP)

That’s 12 multi-walk innings, and, oddly, 4 of them were at the hands of the Dodgers. If we back those 13 walks and 12 innings out of his totals, we’re left with:

12 multi-walk innings; 42 one-walk innings; and 56 no-walk innings — or something like that. Cahill had 110.2 IP for the season, but he pitched in something like 125 different MLB innings, what with getting pulled mid-inning and some of his relief appearances. That kind of distribution is well within what would be likely if they were a matter of chance.

No, we’re still at luck. The bad thing about that is we can’t necessarily explain it. The good thing, though, is that we don’t have a reason to think that his bad luck will continue — especially if increased shifting could help lower his BABIP on ground balls. Stay tuned.