By Eno Sarris
Whether you're evaluating your fantasy team or thinking about the future of your real-life team, now is the time for your favorite projection system to shine. Using past numbers, weighted to generally favor the most recent year but also to take input from early in a player's career, these systems try to tell us the most probable outcomes for the upcoming season. Right now, they're all we got other than spring training updates and impossibly useless results.
Unfortunately, those projection systems know a good deal less about one half of the game than the other. Pitchers remain a mystery.
If you test projections against actual outcomes, you'll find that they do a decent job for hitters. Projection systems can explain roughly two-thirds of the general variability for their production, year to year. That's much better than looking at just last year, which only tells you about half of what's about to happen this year.
Pitching? You're lucky if your fancy projection system gets anywhere close to explaining half of what's coming this year. Many projection systems don't do better than explaining one-third of the variability in the projected year.
Look at last year's best teams and how they did versus the projection systems. The Cardinals, Red Sox, Braves, Athletics and Pirates all beat their projected win totals by double digits. Those teams beat their pitching projections by an average of 90 runs, and only beat their hitting projections by 30. The Pirates, one of the biggest surprises in the game, beat their pitching projections by 111 runs and came in 36 runs *worse* than their hitting projections.
|Real RS||Real RA||Projected RS||Projected RA||Diff RS||Diff RA|
If you're looking for a team to surprise this year, look for a pitching staff that you think will improve. And if you just want to hang your hat on a contender, gravitate towards the team that will score the most runs -- just because that's the team we know best. Sort this list of projected results by runs scored, and that gives you the Rangers, Red Sox, and Tigers of course. But could the Blue Jays, Rockies or Angels surprise? They've got the offense -- they just need their collection of arms to work out.
We're still left wondering about the why. Why are pitchers harder to project? Ask ZiPs creator Dan Szymborski, and he admits "Why is really, really damn hard." Pressed for a guess, he says that he suspects that "pitching performance is simply inherently more volatile than hitting performance" even if that's "a completely unsatisfying answer."
We've talked some about pitchers get injured more often and stay injured longer, and how their arsenals and mechanics can lead to those injuries. Maybe some day those things can be integrated into the projection systems to improve their performance. But Szymborski isn't sure that injury risk is the real driver in the difference. It's not just about projecting innings pitched totals, it's about the performance in those innings.
Right now is the time for updates about how every player is in the best shape of their life, and how every pitcher is working on a new pitch. Those updates are oft-derided, and maybe when it comes to fitness, that's for a good reason. A 30-year-old athlete is by definition post-peak and almost certainly not in the best shape of his life, really.
But a new pitch? A new pitch could be a big driver behind lurches in performance. Look at some of the recent projection-busting veterans, and many have made adjustments to their arsenal. A.J. Burnett added a two-seamer (and started throwing a second knuckle curve from a new arm slot, pictured below).
Ubaldo Jimenez featured his split-finger more often. Anibal Sanchez started throwing his fastball less and less often. Adding a new pitch is adding a dimension that seems almost impossible to do on the other side of the plate.
Basically, what we might be saying here is that we know less about what happens at the mound than we do at the plate. We want line drives and walks from our hitters, but pitchers have many different ways to coax an out. They can get there with a tiny walk rate, or an extreme ground-ball rate, or a great strikeout rate. Some can even do it with an average skill set across the board. And for every advancement in theory, there are always asterisks. Pitchers may not be able to control balls in play, a theory known as DIPS or Defense-independent Pitching Statistics -- except there's evidence they can control pop-ups, which are automatic outs.
In order to help their projections know more about what's happening on the mound, Steamer has added velocity data, zone contact information, and pop-ups to their pitcher projections. That helps with pitchers that were difficult to project before, particularly pop-up monsters like Colby Lewis, Jered Weaver, Justin Verlander, Matt Cain, and Jon Lester. In the present iteration of Steamer, they've added some adjustments to reflect the pitcher's defense, and his impact on the running game.
These little changes help us understand pitching better, but they work on the order of millimeters, not kilometers. Last year, the Steamer team was able to explain about half the variability in the pitching population with their projections. That's better than it has been in the past, but just barely.
Jared Cross, one of the founders of Steamer projections, had a couple ideas about why the gap still remains. One is that perhaps pitchers have less sway over your average plate appearance than hitters do. Though some work by Steve Staude suggested that the pitcher and hitter were equally responsible for a strikeout, maybe other components don't show that tie. Maybe a hitter has more to do with where the ball goes when it's in play, for example.
The last possible reason for the split is perhaps the most compelling, but also the strangest. If you look at the spread of talent across the league, you'll find that pitchers have been grouped more tightly than hitters over the last decade. Only two of the last twelve years have produced spreads that were tied or were larger in the pitching population. If talent is more closely grouped together, it's harder to project. Consider the ridiculous case of a league where every player was equally good -- it would be impossible to project the season!
One last wrinkle is that hitters have generally become tighter-knit in terms of talent over the last decade, too, and so it might eventually get harder and harder to project them.
But for now, it's still true: You know more about what the hitters on your team are going to do this coming season than you know about the pitchers -- fantasy or real.
* * *
Eno Sarris writes about baseball at FanGraphs www.fangraphs.com most of the time. He also started BeerGraphs www.beergraphs.com for the beer nerds out there. He doesn't always play daily fantasy, but when he does, he plays it at DraftStreet.com.