Felix Hernandez And True Talent
Today wasn’t one of Felix’s classic amazing starts, but that right there tells you where he’s managed to set expectations. He worked eight strong innings against the Twins, with plenty of strikes and enough missed bats, and as I write this Felix has one of the league’s better ERAs. He has one of the league’s better FIPs, and he has one of the league’s better xFIPs. He’s been one of the league’s better pitchers, and we know that, and we also know that, before today, Felix was in a bit of a slump. Four or five turns in a row he was somewhat mediocre, and so it’s good to see him come out of that. This was Felix pitching like Felix, and afterward, he said he felt great about his performance. We hadn’t heard that for a while.
We also suspect we know why he slumped, or at least why he slumped as long as he did. Granted, fans are always looking for excuses for under-performance, but some time ago Felix got really sick and lost considerable weight. The flu was going around the clubhouse, and it’s not like Felix struggled after a bad night of sleep. I don’t think it’s a reach to assume being sick, and having been sick, affected Felix’s pitching. When you’re sick, you’re weaker, and you get fatigued easily. When you’re recovering, the same stuff applies, to a slightly lesser degree. There’s healthy pitching and sick pitching, and it stands to reason the latter is a lot worse than the former.
This is actually less about Felix, specifically, and more of a general thought. Felix is just who brought it to mind. What Felix did while sick counts on his record. You can’t really just throw starts out, so those innings are just a part of Felix’s career. Now, in analysis, we’re frequently talking about a player’s true-talent level. That is, what his numbers would look like if he had an infinite sample. This is what projections try to approximate. We know that numbers are volatile — what we assume is that they fluctuate around the true talent. Our understanding of true talent is informed by the statistical record.
But the statistical record sort of assumes that everything evens out. We assume that, over time, the signal drowns out the noise. But how do we want to define true talent? Is true talent just overall average performance, or are we talking about true talent when a player is at or around 100%? Do we really want to care about what Felix did when he had the flu? Do we really want to care about what players do when they play through pain? Most recently, Jose Fernandez pitched through a damaged UCL and had a lousy evening. Now he’s officially out for a year. His numbers will always include those five innings and six runs, but why should we care about those when analyzing Fernandez later? Presumably, his true talent ought not involve a torn ligament.
Players will tell you that, especially later in the season, no one’s 100%. Absolutely, that’s correct — the season is brutally long, and taxing, and a player’s condition in September isn’t his condition in April. But there’s ordinary wear and tear, and there’s the more unusual stuff, and if players play through unusual stuff, it can affect their numbers, and it can subsequently affect the perception of their true talent. In some cases this is a minor thing, and in Felix’s case we’re just really talking about two starts involving illness, but I think it’s interesting to consider what true talent means.
Every projection is based on history. You can never be better than 100%, but you can be worse, and many projections will consider history when below 100%. And then they’ll just average that stuff out, such that, if you believe a player is 100%, he should probably be better than his projection. His projection unknowingly accounts for some performance-affecting issues, and if you want to know how good a player is, really, in theory you should strip that stuff out. That is, if true talent refers to when a player is fresh.
On the other hand, if you boost every single projection to put each player around 100%, then pitchers would be better and hitters would be better, and stuff would cancel out, and we’d be back to square one. A pitcher isn’t always 100%, but the hitters he faces aren’t all 100%, and you can’t just half-adjust. This all gets really complicated, and it’s a small-enough deal that it’s not like the projections are systematically wrong. This is me writing words about something insignificant.
Going back to the start: Felix’s numbers reflect, accurately, what’s happened with him on the mound in 2014. They reflect his performance, but they’re not a totally accurate reflection of how healthy Felix has pitched, because there are a few starts in there of Felix pitching while sick or post-sick, and in one of them he didn’t strike a batter out. Non-sick Felix this year has been outrageous, and while even the fresh version isn’t immune to the occasional stinker, it seems that we’re really most interested in how good a player is when he’s on his game. At the end of the year, Felix’s starts will all be grouped together, but depending on what we’re asking, perhaps that’s not quite appropriate.
Or maybe it is. I haven’t thought this all the way through, and I don’t know if that’s even possible, or if this is all just circular and infinite. I might be the only person who even gives a shit, and that would be totally fine. But Felix pitched sick. For the most part, when he hasn’t pitched sick, he’s been downright amazing. That’s more like what Felix really is. That’s what I’m most interested in talking about thinking about. We can’t strike games from the record, but we can ask if we should, under certain circumstances.
Comments
13 Responses to “Felix Hernandez And True Talent”
Leave a Reply
You must be logged in to post a comment.
What the fuck, Jeff? TWO articles in a day!? C’mon man! Pace yourself!
I’m comfortable with projection systems that incorporate historical data from outings when a player wasn’t 100%. Because guys do get sick, and guys do try to play through tweaks and soreness and all kinds of crap, so a projection that doesn’t take into account how a guy has played when not at his best is a projection that’s going to be more often wrong than one that does look at everything.
Particularly with Felix, because we all know that when Felix is at 100%, he throws perfect games, and who would ever believe a projection that said Felix was going to go 36-0 with ERA, FIP, and xFIP all at 0?
Adrian Beltre fits this idea well. When he played for the Mariners he always seemed to be playing hurt. Of course not always, but a lot. He had the ability to play through just about anything and, since his defense was so good, the team was willing to put him out there. My suspicion is that playing hurt was as much the cause of his batting woes while in Seattle as was Safeco. Maybe more.
When I think of what Beltre’s True Talent level is, I think of him as a Dodger, a Red Sox, a Ranger. Not as a Mariner. His time as a Mariner dragged down his production, but that seemed more due to Safeco and injury than a *better* definition of his True Talent level.
On the other hand you could maybe say that Beltre’s ability to play through injury and pain is part of his overall package, and you can’t throw that out because that’s who he is.
It depends on the definition of True Talent.
An infinite sample of Felix Hernandez sounds pretty awesome.
For a player like Felix with a multi-year track record, I don’t think a couple starts like this really matter for projections. As you said, any pitcher is going to have days where he isn’t 100%. Felix happened to catch the flu which is a bit fluky.
But for a guy with a short track record like Ackley or Saunders, a period where they played hurt or played with a bad approach messes up projections because they don’t know those PAs are less predictive than others.
I suspect, in most cases, adjusting for health is a mistake. You end up with systems that project Nick Johnson as a 1,000 OPS superstar and that put Eric Davis in the all time top ten of baseball history. (Not to mention Dickie Thon, Ron Gant, and other extreme examples.)
Health is a skill. It’s part of the package, and most often it will harm the accuracy of projection systems to try and adjust for it.
Hurt/sick Felix is better than Joe Saunders.
Even counting his five years in Seattle, Adrian Beltre will have earned his plaque in the Hall of Fame.
Interesting thoughts, Jeff.
A lot of times players won’t ‘fess up to being hurt, so I’m not sure how we could get that info. Conversely, being human, I suspect some players say or feel they’re injured or ill when they don’t perform well.
Felix Hernandez was sick, performed substandardly for a bit, is 2nd in the MLB in pitcher WAR according to Fangraphs.
Jay, did you notice that former M’s #2 draft choice Dennis Raben is still slogging away in the Cal League? I saw his name in the box score and checked to see if it was the same guy. It is.
Yes, I am aware of such things, as he was recently playing against the Mavs and hit a dinger. Was surprised to see him still active after the various microfracture surgeries, etc.
Jeff, I don’t think it’s insignificant. At the theoretical level, the question you’re raising is important. It’s an issue we ought to think about carefully before we decide what practical questions we want/need to ask, because it helps us figure out what exactly we’re trying to find out; specifically, it helps define the terms of the discussion, which is essential if there is to be anything resembling a clear conclusion.