Kim to the pen, Barmes/Sullivan to the farm

[Heltonfan]

Is the whole "he gives the club a chance to win" a fallacy?

Yep.

And it's outside the model here, but perhaps the whole basis for the pro-Fogg-type bias is that managers generally overstate the systemic impact on their bullpens of carrying one BK-type starter in a 5-man rotation? (I suspect this is the real reason.)

If that's the case, then they're really nuts, since BK averaged more IP per start than Fogg did.

In this situation, I don't think we can rule out the possibility that they prefer Fogg to Kim because Fogg is the generic hard-working guy without much talent (kind of like a pitching version of David Eckstein), and they understand and respect that. Kim is the Oriental enigma, complete with apparent attitude problem. It's more about personalities than it is about performance.

[Roxpert]

Hey, Jackass, you've just GOT to grab yourself the 2007 Baseball Forecaster by Ron Shandler. He has a chapter devoted to the subject of pitching consistency, and shows PQS pitching logs for every starting pitcher.

What is PQS? It stands for "pure quality start", and is an advancement on the old idea of quality starts (6+ IP, 3 or fewere ER allowed). Each outing for a starter gets a PQS score, ranging from 0 to 5. Here's how it works:

To get 5 points.....

1. A pitcher must go a minimum of 6 innings. If he goes less than 5 innings, he gets a PQS score of 0 for that game, no matter the other stats.

2. He must have given up no more than the same number of hits as innings pitched.

3. His strikeouts must be no fewer than two less than innings pitched.

4. He must have struck out at least twice as many batters as he walked.

5. He must have allowed no more than one home run.

The number of PQS points for an outing depends on the number of the above five criteria that are met. Shandler then logs every start, and measures what percentage of outings the starter scored a 4 or 5, which is called "Domination Percent" (DOM%), and also measures what percentage of outings scored a 0 or 1, known as "Disaster Percent" (DIS%). This is really the measure of consistency that you would want to look at in comparing two pitchers.

How do Kim and Fogg look based on DOM%/DIS%?

Fogg
2004 - 19%/25%
2005 - 32%/21%
2006 - 23%/29%

FYI, that's not very good. By comparison, Jason Jennings had a ratio of 50%/9% in 2006. How about Kim?

Kim
2004 - N/A
2005 - 41%/27%
2006 - 44%/37%

By this measure, it's clear that Kim dominates more games than Fogg, but is also maddeningly inconsistent. Fogg's "qERA" (ERA derived from DOM%/DIS%) was 4.98, while Kim was 5.07. Their "xERAs" (derived from the component stats) were 4.18 and 3.57 respectively in 2006.

So, you can see in this illustration that consistency, or lack thereof, makes Kim look slightly less desirable than Fogg, while component stats make him look clearly superior.

How much do the Rockies value consistency, or penalize inconsistency? I wish I knew.

FYI, other Rockies starters' DOM%/DIS% in 2006:

Cook - 31%/13%
Francis - 34%/16%

And the DOM%/DIS% of baseball's Cy Young winners in 2006?

Johan Santana - 76%/3%
Brandon Webb - 64%/3%

[HoyaRoxFan]

Makes sense... the Rockies have been a remarkably "consistent" team throughout their entire history... the Marlins have not.

In 14 seasons in the league, the Rockies have posted better or identical records than the Marlins 8 times... yet the Marlins have racked up 14 more wins, resulting in 2 WS rings.

[Heltonfan]

Originally Posted by Roxpert

By this measure, it's clear that Kim dominates more games than Fogg, but is also maddeningly inconsistent. Fogg's "qERA" (ERA derived from DOM%/DIS%) was 4.98, while Kim was 5.07. Their "xERAs" (derived from the component stats) were 4.18 and 3.57 respectively in 2006.

So, you can see in this illustration that consistency, or lack thereof, makes Kim look slightly less desirable than Fogg, while component stats make him look clearly superior.

This is nonsense. PQS is a junk stat. And you certainly can't derive a meaningful expected ERA from DOM/DIS percentages, given that the system treats 4.2 brilliant innings as being no different than 4.2 horrid innings. That's just one error after another. The fact that these results show
Kim's inconsistency as a negative, when all other studies (and some basic logic) indicate that it's a positive, is an awfully good sign that the system is garbage; you'd be better off just looking at ERA (yes, I said ERA... and you know how much I despise that statistic) than being misled by this useless extra data. This stuff isn't sabermetrics, it's just a bunch of mildly amusing factoids for curious fantasy players.

[Roxpert]

I wouldn't dismiss PQS and DOM%/DIS% entirely, though. They aren't good stats for predicting much, but they are "indicators" that tell us more than the old "Quality starts" or "Quality start percentage". And while you're right that 4.2 innings with 0 ER gets the same zero points as 4.2 IP and 8 ER, how often does a pitcher really come out in less than 5 innings throwing a shutout? Unless it's due to injury, that's quite rare.

In short, I agree that PQS as a literal "stat" is a junk stat. As an INDICATOR of consistency and dominance, it's not totally useless. It's a metric that perhaps more than fantasy junkies ought to look at, because it does tell you something about a pitcher.

[Heltonfan]

As an INDICATOR of consistency and dominance, it's not totally useless.

Fair enough, although that's really just semantics. It's useless to me, because I don't need "indicators" - that is, if I was interested in measuring those things, I know exactly how I'd do it. Calculate WAR (wins above replacement) for each start, and take the standard deviation of the WAR numbers as the measure of consistency. Simple enough, and it keeps everything in meaningful units (which is the downfall of the PQS system, and why I called it a "junk stat"). The key point is that you can't use the PQS numbers, or anything derived from them, as a measure of value.

[BigRapidsJackass]

Just flipped on the TV after spending a perfect early afternoon outside ... Plan B, a/k/a Brian Lawrence, doing exactly what we expect him to -- generally moving the ball in and out, up and down, getting by nicely except for a couple gopher balls to Eric Byrnes.

I'm thinking he'll be up pretty soon, but not in place of Fogg. I think Rodrigo's already digging himself a big hole, and I'm really beginning to wonder why they sunk $4.5 million into him.

EDIT: Purplerow says Rodrigo's starting the "B" game this afternoon. Doesn't that strike you as weird for a guy who will probably just get one more tuneup start before he goes in Game 3 of the regular season, while meanwhile Lawrence -- ostensibly headed for extended spring training -- gets serious work as if in anticipation of the regular season? Don't be so sure Rodrigo doesn't get moved again before the season starts. Seems to me he's rubbing Apodaca and Hurdle the wrong way (which is exactly the way he rubbed Mazzone in Baltimore).

BRJ One-Note Act Alert: Fogg + Rodrigo + BK = $10.5 million = Old Reliable "Grossly Overpaid" League Average Innings Eater Like Jeff Suppan.

Once again, I think we'll learn the hard way the quality (even very generously defined to include a guy like Suppan) beats quantity.

[Heltonfan]

210 innings of Suppan = 21 RAR.
180 innings of Kim = 19 RAR.
I'll take the 2-run dropoff and pocket the $8 million any day. We really need to ditch this idea that Suppan is actually league-average.

[BigRapidsJackass]

HF, again -- in a perfect rational world -- you are correct.

But on the real planet earth, in which Nobel Prize winning economists are not running baseball clubs like the Rockies, the $2.5 million investment in BK is likely to be a total loss. The Red Sox ate $6.5 million of BK bucks a couple years ago, and now it's our turn.

What I can't understand is why nobody else seems to want him. I'd say the obvious candidates are:

-- the Nationals. He would immediately be at least their #2 starter, and they must have something in their system to give us in exchange.

-- the A's. I'd swap him straight up for Joe Kennedy in the kind of "we'll exchange each other's problems" trade. If you could convince the A's to take Tom Martin's and 1/3 of his salary too, it'd be a great deal.

-- the Mets. Yeah, the assumption is he'd melt under the NY media pressure. Or maybe his role -- disgruntled Korean who feels the need to start or sulk -- is being filled by Chan Ho Park already.

-- the DRays, who have a tremendous surge of young hitting talent, who have a happy and well-adjusted Korean influence in Jae Seo, and who really, really need pitching if they have any hope of getting off to a decent start.

-- the Twins, who apparently have to rely on Carlos Silva, and who have a fine park for Kim

-- the Giants, who have a Benitez problem in the pen, a good park for BK, and may actually think of using him as a closer.

-- the Padres, who have a Wells-sized question mark in the rotation and the best possible park for BK.

C'mon, none of these clubs could use him? Exactly how obnoxious can a guy be?

[BigRapidsJackass]

FWIW, the FSN announcers say BK looked sharp in the B game this morning, and the Denver Post reports on similar success for fellow problem spring guy Rodrigo Lopez:

DenverPost.com - Rox Rundown: Lopez finds rhythm

[TheIncredibleRox]

So far, two of my predictions have reached a conclusion (or come very close);

--I predicted we would find some way to screw Ianetta in favor of vets; I was wrong, and we jettisoned Lopez. Very happy to be wrong here.

--I said we'd be hurt by liking Rodrigo Lopez way more than we should, and hating BK Kim way more than we should. The fruits of that decision haven't been borne out yet, but the stars are lined up just right. I think Lopez--or anyone who allows that H rate and HR rate-- is a can of gas. Coors will be the blowtorch.

[BigRapidsJackass]

I think the official story today is that Rodrigo pitched the "B" game because the Rox didn't want the DBacks to get a good look at him (since Rodrigo is scheduled to pitch in the opening series of against the Dbacks).

IR, Camden Chat (a great Orioles blog) ran a hilarious "a picture paints a thousand words" photo montage after the trade of what we'll be seeing on the Coors field pitching mound from Rodrigo starting next week:

Camden Chat :: Let's us remember the Rodrigo

[Dante Bichette]

This isn't just about Fogg vs. Kim, since Lawrence and possibly even Jimenez factor into this decision as well. Kim goes to the pen, but given his salary and his constant complaining I bet he gets moved sooner rather than later, and then if Fogg or Lopez struggle or we have an injury Lawrence immediately steps into the rotation. Jimenez works in the Springs for now, but if he succeeds there he will be close to coming up.

If the Rockies are smart they will shop Kim, Lopez, and Fogg and jump all over the best offer they get for any of the three.

[Heltonfan]

If I was interested in measuring those things, I know exactly how I'd do it. Calculate WAR (wins above replacement) for each start, and take the standard deviation of the WAR numbers as the measure of consistency.

I had a little time to kill this morning, and baseballreference's game logs make this sort of thing a snap, so I went ahead and ran these numbers.

The results:
Fogg: 1.71 WAR, .211 standard deviation
Kim: 1.88 WAR, .214 standard deviation

For further reference, their WARs using the normal method (looking at the full season's stat line, rather than start by start) are 1.65 for Kim and 1.49 for Fogg. So the gap between the two is the same, regardless of which method we use. The reason that the numbers are higher with the new method is that a) inconsistency is a good thing and b) looking at the full season's stat line effectively presumes 0 variability from start to start, i.e. a perfectly consistent performance.

Of course, the obvious thing to notice here is that Kim wasn't actually any less consistent than Fogg. This is a nice illustration of how something like the PQS system, which is built on arbitrary cutoffs between different levels of performance and which isn't expressed in meaningful units, can lead to completely wrong results. Garbage in, garbage out. And this took me about 10 minutes, which really makes one wonder why Ron Shandler can't do better.

I should also add that the amount of difference that (in)consistency can make, within the real population of MLB pitchers, is very small indeed. The most and least consistent pitchers in the game last year were Johan Santana (.171 stdev) and C.C. Sabathia (.269 stdev). Sabathia had 4.82 WAR using the normal method and 5.42 going game-by-game; Santana had 7.54 with the old method and 7.82 with the new. Which means that there are no pitchers who are more than 2 runs above or below average in this department; in other words, even if consistency had predictive value (which it doesn't), it would still be virtually irrelevant.

[BigRapidsJackass]

Interesting, HF.

I'm sold. BK's got a bad rap. The only tangible things Fogg has going for him in the battle for the 5th starter job -- "consistency" and "ability to pitch more innings per start" -- turn out to be illusory.

[OK, I haven't looked crunched the numbers, but how about this one -- and I'm really digging here for some rational basis for preferring Fogg -- "Fogg makes it through at least 5 innings more frequently than Kim." Any truth to that?]

So I think we're left with the dreaded "intangibles." The "Unlike Kim, Fogg is a battler and isn't discouraged by a bad start or by pitching at Coors" stuff. But again, if this were true, it would show up in their actual performances; i.e, the discouraged BK would tend to follow up a bad performance with another bad performance (and so on), while Fogg wouldn't. And the numbers just don't show that.

Well, maybe they want to keep Fogg because they have more invested in him (a million bucks more this year in salary)? Well, that doesn't make sense since they'd owe Fogg a million bucks for waiving him, which means it would cost $3.5 million to keep BK and waive Fogg and $3.5 million to keep Fogg and waive BK. So that ain't it.

So I'm left with 3 possible explanations for this perverse Foggophilia:

-- the "Fogg makes it through 5 or 6 IP significantly more frequently than BK" thing (and I doubt the stats support this.)

-- the "Fogg will pitch through pain and BK won't" thing, so that even if they average the same IP per start, Fogg will tend to give you more starts per year. There may actually be something to this one.

-- the "We prefer the American guy" thing. And I think there's something to this when it comes to Kim. I can't prove it, but I think there's some bias here.

-- the "BK is a major league jackass" thing. And I think about 28 GMs are on board on this one.

Still, I'll take the jerk who gives you a better chance to win anyday.