I'll give you this much, you're selling me on Cain's career being better than I had heretofore considered it. I'm still going to say Horlen is the best starter with a losing record, but Cain is in the conversation.

Right, and I guess that's my point. You put out the parameters and as I've said a bunch of times, I really was curious to see who was out there that might have the 200 starts, be a good starting pitcher, and have a losing record. I was actually curious.

You're a huge baseball fan. I thought you'd be curious too. I certainly didn't think I'd get into a two day pissing match over whether Matt Cain was a good pitcher because (1) he clearly was and (2) that is something nobody wants to do.

Honestly JORR I really expected you to say, something like "Good catch on Cain, did you find any others?" Then Matlack, and Gubicza (one nice thing about this exchange is I'll never have to look up how to spell Gubicza again) who I think we both agree are good pitchers. I mean there are other candidates that we could logically discuss.

Mike Moore, 161-176 career record, 1600+ K's, one Cy Young top 3 finish, key contributor to that Oakland team in the late 1980s early 1990s, 28 career WAR for those who want that, 95 ERA+ over a long career, 4.39 actual ERA. Telling stat though, when he played on the Mariners, some of those teams just pure garbage his record was 66-96, once he got to Oakland 66-46. Christ, same number of wins, and FIFTY fewer losses.

Interesting case if he's "good" or not.

Also I guess Matt Cain owes me a few beers or something because if JORR is ever in a bar and some dude says that Cain sucked, JORR will straighten that dude out.

And actually, the r for that top chart is 0.36, indicating a medium level of correlation, bordering on a stronger level of correlation.

If we express "run support" as "runs scored per start" (or simply, RS/GS), we get a better correlation coefficient at .401:



Interestingly, a polynomial line of best fit increases r-squared to above 2, though I'm not sure if that is indicative of anything, because a polynomial relationship between runs scored per start and winning percentage is nonsensical to me.

That graph is exactly what I disdain about modern SABRmetrics. It takes something obvious and simple and needlessly makes it complex, apparently for no other purpose than to make the less educated feel stupid. It's pseudo-intellectualism at its worst. The graph is nothing more than the geometric equivalent of Tim McCarver saying, "The team that scores the most runs usually wins", a statement that would undoubtedly be mocked by the maker of said graph.

It is charting the predictive ability of a pitcher's "run support" as it pertains to that pitcher's rate of "winning" games. Specifically, it shows that a certain percentage of a pitcher's W% can be explained simply by the amount of runs his team scores while he is on the mound, a facet of the game on which MANY pitchers have absolutely zero influence.

Bear in mind, this is a thread (one of many) in which you have stated that the problem of Jose Quintana not accruing more "wins" than "losses" decidedly isn't "run support", and this is undeniable evidence that at least 16% of said problem is explained by his offense not scoring more for him while on the mound.

I understand what the graph is charting. Did someone really have to plot points to figure out that if a team scores more runs the likelihood of that team, and thus the starting pitcher for said team, winning is greater?

"Run support" is nothing more than the runs allowed- or not allowed- by an opposing pitcher. Again, this becomes somewhat of a philosophical argument. Did the Cubs lose last night because Hendricks "lacked run support" or because James Shields pitched better than he did? They're really the same thing but what language do you feel describes what occurred more accurately?

The Cubs lost because James Shields had more run support.

If you're allowing more "run support" than the other guy, doesn't that speak to how you are competing?

But thanks, JORR, you helped me get re-acquainted with that data-set, and while looking through it I noticed some odd outliers (guys with RS/GS below 1, "winning" pitchers with RS/GS less than 2, etc.), and I found that there were some pitchers that were credited with Run Support they earned while in relief, or with Games Started and no concurrent Run Support metrics, inflating or deflating their RS/GS arbitrarily. After removing these from the set, I was left with this:



Nearly 25% of a pitcher's W% can be explained by their Run Support; Not how they "competed", not who was better that day, but by how many runs their team gives them. 1 out of every 4 points in a pitcher's W% is attributable to his batting lineup, not to him. What a wonderful statistic.

That also says that 75% is in their control though.

That also says that 75% is in their control though.

You are making the case that run support makes up 25% of the statistic. To act like run support makes the statistic meaningless it should be much higher than 25%.

I'm not sure teams can control 75% of the game either(and this is the common message when people say the playoffs are largely random) so why even worry about team W-L record? I'm sure we can find some sort of statistic that determines who actually won a game instead of the score.

I think it is called WTCPM. Which Team Crossed the Plate More.

To act like run support makes the statistic meaningless it should be much higher than 25%.

Why? I'm told W% for starters measures how they "compete". If 25% of the statistic that measures how a pitcher "competes" is completely and unquestionably out of their control...how is it still a viable measure of competitiveness?

In a sample size of 1? How can you differentiate the competing from the noise?

Things like the weather affect baseball games too.

I guess the answer is that we can't know anything about anything.

Enough starts.

No, the answer is that we know 25% of W% can be explained by in-game happenings over which the pitcher in question has no control. The question then becomes why is it still used to evaluate competitiveness?

You control 75% of the game. Not many positions in any sport give you more than 75% control of the in-game happenings.

But James Shields has only "competed" against Kyle Hendricks by pitching to the Cubs lineup while Hendricks pitches to the Sox lineup once this season...how do we differentiate the competition from the noise so that we can definitively say that James Shields "competed" better than Kyle Hendricks?

Are you suggesting the Sox lineup is suddenly somehow better than the Cubs'?

But James Shields has only "competed" against Kyle Hendricks by pitching to the Cubs lineup while Hendricks pitches to the Sox lineup once this season...how do we differentiate the competition from the noise so that we can definitively say that James Shields "competed" better than Kyle Hendricks?