MLB 2018 regular season thread

[Hermes]

Things aren't even, though. Sale's WHIP is noticeably better and he's struck out 22 more batters in 19 fewer innings. The only argument against Sale is he hasn't pitched enough innings.

 

[Baron Scicluna]

And though the innings difference isn't huge, Snell was available to his team every day while Snell missed a month.

Forget wins and FIP and everything else. Snell deserves to win just for being able to accomplish this.

 

[TheSportsPredictor]

Things aren't even, though. Sale's WHIP is noticeably better and he's struck out 22 more batters in 19 fewer innings. The only argument against Sale is he hasn't pitched enough innings.

This is the tl;dr version of my post.

 

[LongTimeListener]

Sale is substantially better in everything besides innings. He is incredibly better in K/9, BB/9 and K/BB (6.73 vs. 3.45).

 

[bigpern23]

Kershaw's ERA and xFIP almost never were that close in his prime. The difference was often near or more than a run. That happened for six or eight years. Then last year his HR rate spiked. And that's because he isn't as good of a pitcher as he was three or five years ago.

That's simply false. Kershaw had only two seasons where the difference was near or over a run, 2009 and 2013. In seven of his 11 seasons the difference has been 0.56 or lower (he had two seasons of 0.73).

2008: 4.26 ERA/3.91 xFIP (-0.37)
2009: 2.79/3.85 (1.06)
2010: 2.91/3.64 (0.73)
2011: 2.28/2.84 (0.56)
2012: 2.53/3.25 (0.73)
2013: 1.83/2.88 (1.05)
2014: 1.77/2.08 (0.31)
2015: 2.13/2.09 (-0.04)
2016: 1.69/2.28 (0.59)
2017: 2.31/2.84 (0.53)
2018: 2.51/3.04 (0.53)

 

[LongTimeListener]

If you're throwing around "within a half run" as some kind of validation of this stat, you're going to have to accept a difference of 0.73 as being near a run. Also: The difference between 2.28 and 2.84 (+0.56) is 25 percent. The difference between 2.79 and 3.85 (+1.06) is 38 percent. Those aren't very close.

If you draw a line at 2014 and go back and look at the difference of ERA and xFIP, weighted by innings, you're going to have an overall difference of at least 30 percent. Even for his career, there's a 24 percent difference between his ERA and xFIP. That is not an indication of ERA and xFIP approximating each other.

 

[ChrisLong]

STOP USING WINS!!!!

Interesting last night to hear Smoltz talking about how important RBIs are.

 

[bigpern23]

If you're throwing around "within a half run" as some kind of validation of this stat, you're going to have to accept a difference of 0.73 as being near a run. Also: The difference between 2.28 and 2.84 (+0.56) is 25 percent. The difference between 2.79 and 3.85 (+1.06) is 38 percent. Those aren't very close.

If you draw a line at 2014 and go back and look at the difference of ERA and xFIP, weighted by innings, you're going to have an overall difference of at least 30 percent. Even for his career, there's a 24 percent difference between his ERA and xFIP. That is not an indication of ERA and xFIP approximating each other.

0.73 is closer to a half run than it is to 1 run, but even if I give that to you, that's still only four seasons. You said it was six or eight. Your statement was false.

Your percentages don't make much sense in the context of ERA and xFIP. If xFIP can give you an idea of what you should expect a pitcher's ERA to be within a half run, that's useful. The percentages don't matter.

You're still fundamentally misunderstanding how xFIP is used. It's not used to look back on a pitcher's season or career and say "see what he could have or should have been?" It's a predictive tool. And if you see an xFIP that is way out of whack with a pitcher's ERA, as Rodriguez's is, it tells you that regression will likely bring those numbers closer together.

Of course, there's room for statistical variance, so pointing out a specific single season or player that doesn't support xFIP doesn't discount the validity of the stat. There's plenty of data available that support the viability of xFIP as a useful predictive tool. People aren't using it for no reason.

You've acknowledged you don't expect his ERA or HR/FB to remain so low going forward, so I don't really get why xFIP bugs you so much. You sound like the guy at a Texas Hold 'Em table who plays 7-2 offsuit and thinks it was a great play because he hit 7-7-2 on the flop.

 

[LongTimeListener]

@bigpern23 you also say Randy Johnson and Pedro Martinez's numbers are close. Martinez's really aren't based on percentages (2.93 vs. 3.44 is a 17 pct difference). But Fangraphs doesn't have xFIP or HR/FB data before 2002. So those guys' best years -- and years in which they allowed fewer home runs (in Pedro's case, barely 10 a year) -- are not included.

 

[bigpern23]

@bigpern23 you also say Randy Johnson and Pedro Martinez's numbers are close. Martinez's really aren't based on percentages (2.93 vs. 3.44 is a 17 pct difference). But Fangraphs doesn't have xFIP or HR/FB data before 2002. So those guys' best years -- and years in which they allowed fewer home runs (in Pedro's case, barely 10 a year) -- are not included.

Quite honestly, I simply thought of a couple dominant pitchers off the top of my head who aren't currently pitching and checked their career ERA vs xFIP. I didn't look at their season by season numbers.

That said, Pedro's FIP was always within (and often lower than) a half-run of his ERA in those pre-2002 seasons, so I'd expect his xFIP probably supported his ERA as well, but fair enough if you want to discount Johnson and Pedro.

 

[LongTimeListener]

0.73 is closer to a half run than it is to 1 run, but even if I give that to you, that's still only four seasons. You said it was six or eight. Your statement was false.

This is fairly OOP-like. Take the data from 2008 to 2014 and see what the difference looks like. It's going to be close to a run. That's seven years.

Your percentages don't make much sense in the context of ERA and xFIP. If xFIP can give you an idea of what you should expect a pitcher's ERA to be within a half run, that's useful. The percentages don't matter.

It doesn't matter that there's a 30 percent difference in the runs the guy gives up vs. what he is "expected" to give up? Uh, OK. I think we're seeing who doesn't know how to apply the numbers.

You're still fundamentally misunderstanding how xFIP is used. It's not used to look back on a pitcher's season or career and say "see what he could have or should have been?" It's a predictive tool. And if you see an xFIP that is way out of whack with a pitcher's ERA, as Rodriguez's is, it tells you that regression will likely bring those numbers closer together.

People said this for like six years about Matt Cain. They were finally right -- when his elbow exploded. It's starting to happen with Kershaw. It happened at the tail end of Pedro's career. Know why? Because they weren't/aren't as good. That's not random chance. That's age.

You've acknowledged you don't expect his ERA or HR/FB to remain so low going forward, so I don't really get why xFIP bugs you so much. You sound like the guy at a Texas Hold 'Em table who plays 7-2 offsuit and thinks it was a great play because he hit 7-7-2 on the flop.

Don't knock it till you try it! The psychological benefits last forever. You can run the table from that point forward because people have no idea what the fork is in your hand!

 

[bigpern23]

This is fairly OOP-like. Take the data from 2008 to 2014 and see what the difference looks like. It's going to be close to a run. That's seven years.

You said "The difference was often near or more than a run. That happened for six or eight years." That statement is objectively false. It only happened in four seasons. Don't move the goalposts and try to say that it averaged almost a run across seven years. Please, man, come on.

It doesn't matter that there's a 30 percent difference in the runs the guy gives up vs. what he is "expected" to give up? Uh, OK. I think we're seeing who doesn't know how to apply the numbers.

No, the percentage doesn't matter because if the tool can tell you what to expect a pitcher's ERA to be within a half run, that's helpful. If you were to rely on percentages, the range of possible ERAs expands as the numbers go higher. The difference between a 3.00 and 4.00 ERA is 1 run. That's not as useful as knowing that the range should be somewhere between 3.00 and 3.50. Using percentages doesn't make sense in this context.

People said this for like six years about Matt Cain. They were finally right -- when his elbow exploded. It's starting to happen with Kershaw. It happened at the tail end of Pedro's career. Know why? Because they weren't/aren't as good. That's not random chance. That's age.

Of course it's not random chance. It's how they pitched. And for Kershaw, his xFIP was around a half-run different from his ERA for SEVEN of his 11 seasons, including three of those first seven that you want to cherry pick. As long as we're cherry picking seasons, the difference in ERA and xFIP was around a half run or less in 2014, '15 and '16, his age 26, 27 and 28 seasons when he was absolutely dominant and in his prime. That's age?

Don't knock it till you try it! The psychological benefits last forever. You can run the table from that point forward because people have no idea what the fork is in your hand!