So far in 2015, Colorado Rockies pitchers, as a group, have played out an all too familiar script: being bad, basically. The boys in purple have pitched to a 5.08 ERA in 2015, 29th out of 30 teams (Thanks Blue Jays!). Really, most casual observers would probably conclude that it’s pretty much the entire team that’s following script thus far. Rockies batters are being “so Rockies” this year, too, posting a .765 OPS, 6th best in the league.
But whenever a savvy Rockies fan sees mainstream hitting and pitching statistics like ERA and OPS, he or she knows that additional analysis is necessary. Coors Field looms on both sides of the ball. We’re lucky enough to have some pretty good context neutralizers for batters. wRC+ happens to be my favorite, and by that measure, the Rockies offense, at a mere 96 wRC+, looks closer to average (16th, in fact). Sounds about right, I think.
The players on the other side of the equation – those rascally Rockies pitchers – have always been a bit tougher to figure. There is so much noise in a number like ERA. The ballpark and defense matter, of course, but plain old luck and all sorts of other factors matter, as well. The net result is that ERA borders on useless, either as measure of what the pitcher actually did, or as a measure of what the pitcher is likely to do going forward.
DIPS Theory – basically, the idea that the pitcher has very little to do with what occurs once a ball is put into play – helped a lot. It led to measures like FIP, which is a much truer estimate of pitching talent. FIP is also intuitively lacking. Even if we slap a ‘+’ or ‘-‘ on there and make the FIP statistic park and league adjusted, there is such a thing as inducing weak contact, isn’t there? Sure feels like it, at least. SIERA, which tweaks the formula and includes some batted ball data, was an improvement both logically and in terms of statistical verifiability, but even then, we know that the classification of batted ball data is somewhat arbitrary. And in all cases, again, there are just so many other variables at play…
But dag-nabit if the Sabermetric community hasn’t improved our sorry lot yet again! Two new pitching metrics – Deserved Run Average (DRA) and Context Fielding Independent Pitching cFIP – have emerged just over the last couple of months, and while neither purports to be the be all or end all, they represent another giant step towards The Truth. Both are based on the idea of using “mixed models” to isolate the importance of several variables simultaneously at play with one another. I’ll ask that you click on each link above for the full explanation (Seriously: do it. They’re amazing.), but suffice it to say that both of them a) account for a mind-blowing amount of factors, and b) are statistically proven to be two of the most robust pitching metrics ever created. We’re getting closer, people!
With that in mind, let’s take a look at our Rockies pitchers and see what lies beneath their surface level statistics. I’ve included both of the new statistics and, for purposes of comparison, their “old” counterparts: ERA and FIP. I’ve also included the players’ league-rank for each. Not unlike ERA and FIP, a shorthand approach to digesting DRA and cFIP is to look at the former primarily as a measure of what has happened, and the latter as a cleaner estimator of underlying talent, and therefore a better predictor of what will happen in the future. There are 434 pitchers in the pool, which includes everyone who’s faced at least 10 batters. I was able to set the threshold so low because, as you know from reading the links above (you did read them, right?), DRA and cFIP, unlike ERA and FIP, tell us something useful even in small samples. The table is sorted by DRA, which you’ll note is much more bunched together than the ERAs. This makes sense, since the whole point is that pitchers in general have less to do with overall run prevention than ERA might make us think; nobody truly deserves a 0.00 ERA, nor do they a 90.00 ERA. (Note: all numbers heretofore and to follow are as of May 4th.)
Let’s start at the top. Jordan Lyles is the player that looks most different in light of the new metrics, and in both extremes. DRA absolutely loves what he’s done thus far, deeming him the 29th best pitcher in the league. Part of DRA’s upgrading of Lyles vis-à-vis ERA surely comes from what we might consider the common park adjustment, but remember, DRA doesn’t apply a blanket park adjustment in the way that the standard ‘+’ and ‘-‘ adjusts ERA, FIP, etc. In those cases, the park adjustment is applied 50/50 Coors/road. DRA is much more nuanced – not all road parks play the same, of course, and not all pitchers play exactly half of their games at home, especially early in the season before the schedule has enough time to balance out. DRA also considers the quality of competition all the way down to the individual batter. Knowing this, perhaps it’s not surprising that DRA likes Lyles so much, given that only one of his five starts (at L.A. Dodgers) occurred in a pitching-friendly venue, and even that start came against the team with the best wRC+ in the league.
But much of the shine comes off when one’s eyes slide to the right and take in that cFIP number. Traditional FIP doesn’t care much for Lyles, either – mostly because those strikeout and walk rates of his are atrocious – but FIP doesn’t bang on Lyles as much as it otherwise might because Lyles has managed to limit the home run damage (just 1 in 29 2/3 innings). Years and years of statistical history tell us that even a groundball-orientated pitcher is unlikely to keep up that kind of HR rate. We’ve had xFIP around for a while now to account for unsustainable HR/FB rates, but cFIP considers that plus a whole lot more. And cFIP hates Lyles. In fact, the disparity between the two new metrics (DRA: the run prevention credit he deserves based on what’s actually happened, and cFIP: the run prevention we should generally expect) is the largest in the league. In other words, he might be the sport’s biggest smoke and mirrors show in the young season so far. This is bad news for Rockies fans.
Everything we just discussed with Lyles – at least with respect to projection going forward – also appears to be true for fellow young guns Tyler Matzek and Eddie Butler. And the reasons are probably largely the same: the sort of walk and strikeout ratios that are hard to get away with for long. And, once again, cFIP gives us even more cause for alarm than does FIP, despite the park-adjustments incorporated into cFIP. In other words, the more context we consider, the worse their statistically-derived likely futures appear. Look, none of these metrics can interpret these guys’ processes or pedigrees, and even cFIP is a less helpful forecaster than the full-bore forecasting systems, most of which are more optimistic on these guys than cFIP. I personally remain quite bullish on both Matzek and Butler, even in the short term. Still, it’s one more little statistical tidbit that makes the meal we’re eating just a tad less appetizing.
In sunnier news, we have Rockies stalwart and Coors Field ledged Jorge De La Rosa. We didn’t need any new fancy stats – or even old fancy stats – to know that De La Rosa wasn’t going to pitch to a near-league worst ERA all season. He’s got too long a track record and, just by the plain old eye test, he’s simply looked too good. (Who knew Jorge still had 96 MPH in him!?) FIP agrees we should expect better run prevention, and cFIP is downright manic in defense of this view. I’ve personally always been a big fan of De La Rosa’s – sometimes even irrationally so, perhaps – which makes that cFIP number the prettiest set of digits in the chart to me.
Also encouraging is that our relief aces – Adam Ottavino, Rafael Betancourt, and John Axford – look good across the board, old fancy stats and new, looking forward and looking back. They’ve all largely deserved their stellar ERAs, with Betancourt actually coming in on top in the DRA race, and the performance of all three appears sustainable. Ottavino has been downright elite this year – again, both by the stats and the eye test – but of course the Baseball Gods just decided to take him from us. Sigh… at least we’ll have the memories:
See you in 2016, Ottavino’s Slider.
The Christians Bergman and Friedrich both represent the same interesting case. DRA tells us that their ERAs are largely what they seem (Bergman’s raw DRA number is a couple points higher, but mostly because, again, DRA “bunches.”), and traditional FIP suggests they’re really earning it. cFIP doesn’t buy any of it for either Christian.
cFIP takes the exact opposite view of Logan and, especially, Oberg. DRA might not disagree with ERA with respect their run prevention thus far, but cFIP thinks these two are much better pitchers than the numbers – even plain FIP – suggest.
Brooks Brown is the Jordan Lyles of the relief corps.
And, finally, we have Kyle Kendrick and Jorge Rondon. Rondon has the worst DRA in the league. DRA cuts him some serious slack in comparison to that ERA number (also worst in the league), but even more damning than that DRA number on its own is the gulf between Rondon and the second worst DRA in the league, the Phillies’ Severino Gonzalez at 6.96. Once again, DRA tries hard to squeeze to the middle. Rondon cares not – he has looked Math straight in the eye, and broken its will.
The new fancy stats tell us nothing new about Kyle Kendrick. His story changes least of all. There is no relief. There is no hope. Actually, that’s not quite right: when I ran these numbers the first time on Friday, Kendrick had a DRA of 6.44. So that means that, as of Sunday’s start when he lasted 5 innings, giving up 6 runs on 6 hits including 2 home runs, he’s officially trending in the right direction.