Player A and Player B

[RedEye]

Originally Posted by jojo

Here's a point that I don't think someone directly brought up though there was a comment earlier about streakiness. This of course is a generalization (i.e. it might simply be a unsupported belief). Guys like a Dunn, Howard, or Sexson tend to be very streaky. Whereas production like Medwick's tends to be a little more even keeled.

While both types of players get the same production ultimately, I wonder if one isn't more valuable if indeed it's a more consistent, steady production (i.e. spreading it out might impact more games than bursts of production might).

One of my all time favorite players is John Olerud. I would take him over Dunn any day of the week. Interestingly, another firstbaseman beat him out in the MVP voting in 1993 and it was a complete hose job IMHO owing to the power and influence of the long ball. Factor in the defense of Olerud vs Thomas and that year Olerud was clearly superior.

It was me who brought up the concept of "streakiness." It may be because I'm not up to date on all the recent state-of-the-art stat measurements, but I feel like it is a bit disingenuous to make arguments like "If you take out x game stretch, his stats aren't very good." IMO, almost ALL players put together a few games where they produce at a higher level. It seems to me that Dunn, because he walks alot, contributes to the lineup's fire power even when he isn't hitting. In that way he is "streak proof"--and that's why you look for players with high OBP rather than high BA alone.

Are there any metrics we currently have to address "streakiness"? If not, it might be an interesting angle to look at with more fine-grained analysis. IMO, it's useless to say that Dunn is of lower value because he is streaky. It's a meaningless word that can be cherry picked and applied to the advantage of an argument.

[SteelSD]

Originally Posted by RedEye

It was me who brought up the concept of "streakiness." It may be because I'm not up to date on all the recent state-of-the-art stat measurements, but I feel like it is a bit disingenuous to make arguments like "If you take out x game stretch, his stats aren't very good." IMO, almost ALL players put together a few games where they produce at a higher level. It seems to me that Dunn, because he walks alot, contributes to the lineup's fire power even when he isn't hitting. In that way he is "streak proof"--and that's why you look for players with high OBP rather than high BA alone.

Are there any metrics we currently have to address "streakiness"? If not, it might be an interesting angle to look at with more fine-grained analysis. IMO, it's useless to say that Dunn is of lower value because he is streaky. It's a meaningless word that can be cherry picked and applied to the advantage of an argument.

Baseball is a game of streaks and I've always considered high IsoD players like Dunn to be far more "slump-proof" than lower IsoD players. When the hits aren't falling, high IsoD players are able to provide more value than low IsoD players because they are better able to avoid Outs. Produce a team with a majority of low IsoD players and you've got a group that's primed to slump badly offensively when the hits aren't falling.

[Johnny Footstool]

Produce a team with a majority of low IsoD players and you've got a group that's primed to slump badly offensively when the hits aren't falling.

Ladies and gentlemen, your Los Angeles Angels of Anaheim.

[RedEye]

How is IsoD calculated?

[rdiersin]

Originally Posted by RedEye

How is IsoD calculated?

ISOD=OBP-AVG

[rdiersin]

Originally Posted by SteelSD

Baseball is a game of streaks and I've always considered high IsoD players like Dunn to be far more "slump-proof" than lower IsoD players. When the hits aren't falling, high IsoD players are able to provide more value than low IsoD players because they are better able to avoid Outs. Produce a team with a majority of low IsoD players and you've got a group that's primed to slump badly offensively when the hits aren't falling.

Is this true? I used to think so as well. Now, I am not as positive. Consider the game log data from 2000-2006 (thanks retrosheet). Here are the standard deviations of this data normalized to their means. (i.e. these are standard deviations about a mean of 1). I am not sure exactly what I make out of this yet, but I thought I would share it.

I think it is true to a point, as noticed by the obp. But at the extremes I am not as sure.

Normalized	Std
AVG		0.3031
OBP		0.2527
SLG		0.3735
ISOP		0.6533
ISOD		0.5855
OPS		0.3003
S/AB		0.3893

The reason I have been looking at this game log data however, is this weighting on OPS. IIRC Depodesta was mentioned in Moneyball as saying OBP was more important than SLG. I thought this might be interesting, from a variance standpoint. So I looked at a OPS_adjusted=b*OBP+SLG. It seems b=1.5 or so works well. Not that it matters much as it has nothing to do with this thread, but hey, what the heck I figured I would keep going.