Running 2017 MLB regular season thread

[Dick Whitman]

No, it measures how often a pitcher let a runner cross home plate, which is the fundamental point of the game. Yes, league and park factors can skew ERA, but that has been addressed. These extrapolated stats are just hypothetical measurements. Nobody can say with any certainty why there is any divergence between ERA and xFIP for a single pitcher in a single season. ERA is a measure of what a pitcher is supposed to do, not let the other team score. That's all that matters over the course of a season. If you want to say Joe Blowe cannot do it next year, fine. It doesn't mean he didn't do it this year.

The point of an offense is to score runs. I know of absolutely no one who would argue that the most important statistic in evaluating hitter performance is runs scored.

 

[justgladtobehere]

ERA is as contrived a measure as any other, in that what constitutes an "earned run" -- and which pitcher it is attributed to -- is entirely the result of a long-standing practice rather than some empirical fact.

The issue isn't whether ERA is empirical fact, but whether xFIP and other stats is a better measurement of performance than ERA. We don't know for a single pitcher in a single year. If you want to argue that a guy won't pitch as well going forward because his K rate is too low, fine. But don't claim with absolute certainty that he sucked because his xFIP is two runs higher than his ERA.

 

[justgladtobehere]

The point of an offense is to score runs. I know of absolutely no one who would argue that the most important statistic in evaluating hitter performance is runs scored.

That's silly. Whether a batter scores is much more dependent on his teammates than whether a pitcher allows a run. Nobody dismisses ERA as a useful measurement like people do runs scored.

 

[doctorquant]

The issue isn't whether ERA is empirical fact, but whether xFIP and other stats is a better measurement of performance than ERA. We don't know for a single pitcher in a single year. If you want to argue that a guy won't pitch as well going forward because his K rate is too low, fine. But don't claim with absolute certainty that he sucked because his xFIP is two runs higher than his ERA.

As I understand it, if you're trying to forecast next year's ERA, you'd be about as well-served by using this year's ERA as this year's xFIP. The correlations are different, but not dramatically so (say, 0.37 to 0.33, which sounds huge, but we might be talking about only 25% more variance (of a small amount) that's accounted for). The predictive power of one or the other doesn't really interest me. The argument that one is a contrivance and the other isn't ... that does.

 

[justgladtobehere]

How much of a contrivance is the issue. One takes an actual event (runs scored) and uses certain rules to calculate a number to measure how often that event happened. The other takes individual events that may or may not lead up the actual event in question and calculates a number based on what should have happened in some analytical vacuum. The latter is more contrived.

 

[Dick Whitman]

Do pitchers have a particular skill at limiting runs scored, independent of the components - K rate, BB rate, GB rate, etc. - that go into limiting runs? If not, I'd prefer to evaluate them based upon the components you can isolate that are within their control.

 

[justgladtobehere]

We don't know in any particular situation. If pitcher A gets a K with two men on while pitcher B grooves a fastball in the same situation, I don't care that pitcher A has the worse K rate or HR rate. He did better at that moment. That is what ERA measures.

And these projected ERA stats derivative of individual stats aren't magical. The people who create them cannot even agree on one.

 

[Dick Whitman]

We don't know in any particular situation. If pitcher A gets a K with two men on while pitcher B grooves a fastball in the same situation, I don't care that pitcher A has the worse K rate or HR rate. He did better at that moment. That is what ERA measures.

That's fine, but the question is whether Pitcher A is particularly skilled at getting the K in that situation as compared to Pitcher B.

If he isn't, then ERA isn't measuring anything worthwhile.

 

[justgladtobehere]

That's fine, but the question is whether Pitcher A is particularly skilled at getting the K in that situation as compared to Pitcher B.

If he isn't, then ERA isn't measuring anything worthwhile.

But, even imprecisely, it measures what happened, not what some equation says what would have likely happened. That is the value. Pitcher A did better at preventing runs than Pitcher B over a certain period. Does that mean I want Pitcher A over Pitcher B going forward? No.

 

[Dick Whitman]

But, even imprecisely, it measures what happened, not what some equation says what would have likely happened. That is the value. Pitcher A did better at preventing runs than Pitcher B over a certain period. Does that mean I want Pitcher A over Pitcher B going forward? No.

I flip a coin 10 times and get eight heads. My friend flips a coin 10 times and gets four heads. Am I "better at" getting heads?

 

[Hermes]

I never thought getting head would be boring, but here we are with this discussion.

 

[doctorquant]

Maybe a better analogy is this ... Company A had net income of $1 per share. Company B had net income of $1.10 per share. The two companies are in identical lines of business and have very similar asset/liabilities profiles, but Company B boosted its EPS by using working capital to trade in very risky derivatives markets.

By the doctrine of "what actually happened", Company B outperformed Company A ... I ain't sold.