Not to turn this column into an echo chamber, and not to get too serious about a junk stat I've invented as more or less a toy, but I want to answer some legitimate criticism of this column's newly invented pitching metric, pitcher run value, or PRV.
That's because I think PRV, which adjusts a National League pitcher's earned-run average to account for his contributions to the team as a hitter, is of some little bitty use.
As reader JTD pointed out in Tuesday's comments section, it might not make much difference with the best pitchers in the league, but look down the list at middle- or back-of-the-rotation starters, and it'd be nice to know if one of a pair of comparable hurlers can offset the damage he does on the mound a little with his bat.
Because of small sample sizes, PRV probably won't help you predict anything, as Nate Silver of Baseball Prospectus pointed out, but that's OK. Not every stat has to be predictive. ERA doesn't predict itself very well.
A reader with the signature "W" wrote, "Oy ... uncorrected ERA? No park effects? Ever hear of ERA-plus, or better yet DERA? Use a stat like DERA (Baseball Prospectus), and do the same with the batting stats, then you'll have something that will be interesting AND accurate."
My answer to that, W, is: You're worrying about baseball stats? That explains a lot.
My real answer is yes, I've heard of uncorrected ERA and park effects and ERA-plus and even DERA, which is ERA adjusted to be independent of defense. But has the average fan? Have most fans? I think not, except maybe for park effects.
PRV is intended to be a mainstream stat. It's an adjustment of the ERA. It's just a quick-and-dirty way to give you an idea of which pitchers are better hitters, and roughly how valuable they are. Yes, we could dress it up and make it more sophisticated, but that would be way too much work, and there are already sabermetricians doing similar things better.
But their results don't look like an adjusted ERA. The Davenport Translations at Baseball Prospectus, for example, can tell you that Mark Mulder was three batting runs above replacement this year.
That means that in his 36 plate appearances, he was responsible for creating three runs more than a replacement-level player would have -- that's a replacement-level player at any position, essentially a guy called up from Triple-A to play a position. That's damn good for a pitcher. Dontrelle Willis, a pretty good hitter, is six runs below replacement.
But then again: Huh? It took me two paragraphs to explain that.
I don't think the average fan in the stands is wise to the concept of the replacement-level player, which is the foundation for so much of sabermetric analysis. He or she could probably understand it pretty easily once it's explained, but that's a lot of explaining.
It's a lot quicker and easier to say about a pitcher with a 4.35 ERA that if you consider his hitting, you might want to think of him as having a 4.10. That's PRV. One sentence.
M. Turner writes, "Is there a reason to use ERA - RC/9IP rather than looking at the runs created by batters against the pitcher?
"So, it would look like: RC(Against)/9IP - RC(For)/9IP. Admittedly, it is more work, unless somewhere in cyberspace someone has calculated RC(Against) for every pitcher. But it seems odd to squeeze together two stats calculated by different means."
Everybody wants more work.
The reason is that runs created is an attempt to figure out how many runs each hitter is responsible for creating by assigning values to each offensive event, based on observation and calculation. That's because it's impossible to say that a double, say, is worth exactly a certain number of runs.
But for a pitcher, we can pretty accurately measure how many runs he's responsible for. We count the runs that are scored while he's pitching and add those scored by runners he put on base, though those guys aren't entirely his responsibility.
So I don't know why we would use runs created to try to calculate how many runs opposing batters had created against a pitcher when we can just look at the scoreboard and see exactly how many runs they created. On the team level, runs created is a superfluous stat. Runs created equals runs scored.
On Baseball Primer's thread discussing Tuesday's column, Rich Rifkin points out that Win Shares, Bill James' invention, "also takes into account a pitcher's hitting, as well as his pitching and fielding. I trust it more as a total measure of a pitcher's performance than PRV."
As well he should. I'm not a true believer in Win Shares, but again, it's a fairly sophisticated metric. And again, Jared and Katie Fan in Section 421 have never heard of it and are unlikely to want to do the semiheavy mental lifting that goes into understanding it.
It's better than PRV for no other reason that it takes fielding into account. Fielding is the next frontier for PRV! But listen: With his hitting, that 4.35 ERA is more like 4.10.
Jared and Katie can get that one easy.
- - - - - - - - - - - -
Neifi Perez doing his thing in, er, to Detroit [PERMALINK]
Speaking of dumb stats invented by this column, reader Eric Schleder suggests that perhaps it's time to revive the Neifi Index because its namesake, Neifi Perez, is really shining this year, Neifi-wise.
You'll recall that the Neifi Index measures the difference in winning percentage for a player's team when he plays and when he doesn't play. It was invented during Perez's time with the San Francisco Giants, when they pretty much never won when he played and never lost when he sat.
Schleder, a Cubs fan, made the suggestion in mid-August, before Perez was traded from Chicago to the Detroit Tigers, forcing him to leave a manager with an inexplicable love for him, Dusty Baker, but happily landing him in the arms of the other one, Jim Leyland.
But Schleder wrote back the other day to point out that just as Perez had been doing his Neifi thing with the Cubs, he's Neifi-ized the Tigers.
Since Perez joined Detroit on Aug. 21, the Tigers have gone 8-14, continuing a slide that had begun two weeks earlier and has brought the Tigers back to the rest of the A.L. Central. As of Thursday morning they led the Minnesota Twins by a game and a half.
When Perez, whom Leyland often bats first or second, has been in the lineup, the Tigers have gone 4-11, a winning percentage of .267, which is hauntingly similar to Perez's anemic on-base percentage, .258.
When he stays glued to the bench where he belongs -- even though the Tigers acquired him to fill in for the injured Placido Polanco at second base -- the Tigers are 4-3, .571. Subtracting .267 from .571 gives Perez a Neifi Index of .304, which is a Neifi Award kind of number.
Perez had the same effect on the Cubs during the first four and a half months of the season. Of course he did. He's Neifi Perez.
When they traded Perez, the Cubs were 53-70. But they were 36-51, .414, when he played, and 17-19, .472, when he sat. If they'd just left him on the bench, played with 24 guys, they'd have been 58-65 overall, a mere five and a half games out in the wild-card race, rather than 10 and a half out.
But wait, it gets better. This is hard to remember now, but the Cubs started off pretty well this year. They were 14-10 on May 2 -- 8-8 when Perez played, 6-2 when he didn't -- when they started an eight-game losing streak. They never recovered. They went 7-22 in May, 9-18 in June.
There was a lot of talk about the injured pitchers and Derrek Lee's wrist but o-ho, dear readers, you and I know what was really up. Neifi was up. He was up about 60 times in May and about 50 times in June. That's about 110 times too many.
In May, the Cubs were 2-18 when Perez played, but they were 5-4 -- they were a winning team! -- when he stayed in the dugout. That is a colossal Neifi Index of .446. To give you some context, the difference between the winning percentages of the best and worst teams in baseball, the New York Mets and Kansas City Royals, is .237. And Neifi Perez, one guy, made almost twice that difference for the Cubs!
And don't talk to me about small sample size. You don't need large sample sizes to know about the Neifi Effect.
In June the Cubs were 5-13 with Neifi, 4-5 without him, for a Neifi Index of .166, which is nothing to write home about, but it is hauntingly similar to Perez's slugging percentage since joining the Tigers, .167.
But you know, life is complicated. Neifi Perez did not, in fact, win the Neifi Award in 2003, the only year it was awarded. And now that the Cubs have traded Perez, you'd think they'd become the roughly .500 team they were when he was in uniform but not on the field. But it hasn't worked out that way.
The Cubs are 5-18 in the post-Neifi era, a winning percentage of .217. That doesn't make a lot of sense to me, but it is hauntingly similar to Perez's on-base percentage since joining the Tigers, .216.
The magic of Neifi is a great and mysterious thing.
Previous column: PRV leaders, "MNF" announcers
- - - - - - - - - - - -