I've been going full bore on a fantasy league the past few weeks. I mocked up a cursor with miniscule baseball laces, outfitted the home page with CSS and a flash game, Photoshopped
this when we got Moncada. I've got two spreadsheets' worth of personal projections and notes as compared to FantasyPros and ESPN rankings. There's a sticky note on my monitor displaying league average rates for o-swing, whiffs, first pitch strikes, etc. My opinion is still worthless, but I feel almost as though it's my turn to bat here.
Ascertaining the tangible value of a quarterback in football is impossible. Take Ryan Tannehill. Fifth in completion percentage, fifth in success rate. 14th in QBR, though; 18th in expected points added per play; 28th in yards per attempt. His receiving corps (tenth-dropsiest) and run game (second in yards per carry) must be taken into account. Moreover, it took him months to fully acquaint himself with Bill Lazor's schematic. From Week 7 onward Tannehill's QB rating was among the best in football. To top it off, PFF graded the Dolphins O-line as being worst in the league for a second straight year.
Tony Romo will attest, a good line is paramount.
So is Tannehill top-ten? Who the hell knows? Football is so inherently complex that we might as well judge quarterbacks by wins and rings.
(As an aside, I think now's the worst time to be investing heavily in middling QBs. I'm anticipating an influx of real-life J.D. McCoys real soon who will all have been running some variant of the spread offense since they were six. I expect the floor for QB play to rise like it has for relief pitching in baseball. They'd be running back fungible!)
Six-hour NASCAR races complete with pit stops are determined by fractions of a second. If you dive to the middle line and nobody follows you there, you'll plummet like a stone. And if a teenager eight rows ahead clips another driver, your day and the days of ten others will be ruined through no fault of your own. It's impossible to win in NASCAR without a healthy dosage of luck, almost comically so.
Baseball isn't like these games. Baseball is intensely individual. Pitchers pitch, batters bat, repeat ad infinitum, and it's super awesome. It lends itself to statistical scrutiny like no other. National elections, binary as they are, can be
extremely easy to prognosticate. Baseball is much further toward this side of the spectrum. Statistical analysis is my spirit animal. I wouldn't love the sport half as much without it.
Baseball analysts can control for every variable in precisely the way football analysts cannot. If Pedroia doubles with none on and two down, historically that would have been worth about
a fifth of a run. One out, runners on second and third, home team trails by a run in their half of the seventh. A strikeout, and their win probability (derived the same way, from mathematical permutations and thousands of previous outcomes) would fall from
56% to 40%. We know this because baseball and rigorous analytics of said baseball are two sides to the same coin. Baseball
is math, more readily so than the other team sports.
So I put a lot of stock into weighted runs created plus (
wRC+) and skill-interactive earned run average (
SIERA). They're the two best freely available metrics that consider component parts under this microscope. There were 8,137 doubles in 2014 worth on average
this much . So Pedey's, which bounced over the right field wall in this imagining, would have
this concurrent impact on wRC+. We've discerned that pitchers cannot control certain things—batting average on balls in play for one, homerun-to-flyball ratio for two. So SIERA approaches that which they
can influence by case—K rate, walk rate, groundball-to-flyball ratio—and spits out an exacting number scaled to match league average ERA. SIERA's a much better predictor of future ERA than is present-day ERA itself! Park- and league adjustments are applied to each.
Incidentally,
here are the scoring settings for that league I mentioned.
In lieu of the hyperkinetic NASA stats over which Cherington pores in my imagined Sox front office—batted ball velocity, infielder range factor in meters, flyball trajectory, bat-to-ball proximity on sliders while behind in the count—these are my best bet. I look at hitters' wRC+ and pitchers' SIERAs and try to divine from context which direction they're likely to head. If a hitter sees an inordinate number of first pitch strikes, for instance, that would explain a low walk rate. Because sabermetrics excel in another facet: controlling for
luck.
If pitchers have no dominion over BABIP and HR/FB%, and in a huge majority of cases they won't, then it stands to reason that any pitcher with a worse BABIP than league average experienced some bad luck. Strand rate, the percentage of baserunners allowed who fail to come around and score, is similar. It's been shown to hover within a few ticks of 74%. If a pitcher fails to strand three quarters of baserunners, he's unlucky and his ERA is substantially inflated.
Infield hit percentage variance can be impactful. Giancarlo Stanton muscled his way to a .288 average this season, partly on the strength of 13 infield hits. Somehow Bigfoot legged out a proportionally higher percentage of grounders than did Kinsler, Adam Jones, or Ben Revere. Control for his career figure of 6.2% or something I cleared the calculator and his average reads, ".280."
I've even devised three of these things on my own which I hadn't seen elsewhere, all for pitchers. The first is
leadoff OBP versus OBP overall. Leadoff hitters don't always, but typically will carry the most sway in the course of an inning. (Hearkening back to Pedey's double: if it had occurred with none out, it would have been worth .6 runs.) If you perform worse than usual against them, you're probably experiencing some degree of bad luck with sequencing. The second is home start percentage. One in ten pitchers making 35 starts will have 22 or more take place on the road. This is significant because pitchers across the league will perform appreciably worse during road games. The third is RBI/HR. I don't know the league average figure, but I'm certain RBI/HR would be useful if I did. Carlos Carrasco allowed seven home runs last season, five solo, two two-run.
So you see, advanced stats are most definitely illuminating.
Case in point: Johnny Cueto. He sucks and the Reds are idiots for not having traded him. Motherfucker
fell into their lap and they dealt Latos instead. Cueto posted a 2.25 ERA last season and the conventional wisdom even round these parts is that he's a bona fide ace. He is not. He posted the second-highest strand rate in baseball and the lowest BABIP. His SIERA, for its part, is a full run higher—an outlier career best, 3.15. But the SIERA is heavily influenced by a top-ten K rate, easily his best ever, and an above-average walk rate—which are ludicrously unsustainable for a guy who throws the sixth-fewest strikes of anyone, with average chase and whiff rates. Cueto isn't great(o). I want him least. We have enough number three types. If Cueto's markedly above average next season I won't post on this site again
Edit: Pitch framing metrics are new and momentous.
This is what 97 looks like, and turning balls into strikes is an
immensely valuable trait. Baseball Prospectus posits that Christian Vazquez is worth five wins (48 runs) above average over the course of a 10,000 frame season.