Originally Posted by Red Heeler
Do you know of a good source for a primer on IsoD/IsoP? I get the general idea, but I would like to know more about how it is calculated, the theory behind it, relevance, etc.
Thanks in advance.
Not sure if that primer exists, RH. But I'll help as best I can:
IsoD = OBP minus Batting Average (OBP-BA=IsoD)
IsoP = SLG minus Batting Average (SLG-BA=IsoP)
Secondary Average = (TB - H + BB + SB - CS) / AB
All three metrics speak to secondary base acquisition and work independent of Batting Average.
As for relevance, let's take a look at MLB 2005 from a behavior subset vantage point.
In 2005, Runs Created per 27 Outs produced a 0.919 correlation with Runs Scored per game on a team level. That's a scintilla better correlation than between Runs Created and Runs Scored. What we're looking for is that which correlates more strongly with RC/27 Outs within our secondary base acquisition subset.
IsoD to RC/27: 0.410
BB/K to RC/27: 0.415
IsoP to RC/27: 0.793
SECA to RC/27: 0.875
It's pretty clear that what we're looking for is high IsoP and, even better, high Secondary Average (which incorporates BB and SB which is why it's correlation runs higher than IsoP). It's interesting to note that, at this point, IsoD and BB/K are running neck in neck on our behavior chain. So let's take a look at which one correlates more strongly with high IsoP and SECA production.
IsoD to IsoP: 0.174
BB/K to IsoP: -0.011
IsoD to SECA: 0.610
BB/K to SECA: 0.364
The IsoD correlation to IsoP, while small, is still there. Not so with BB/K rate. It falls off the map. That's interesting because IsoD is BB/HBP related and doesn't incorporate a single base used to calculate SLG. Yet we find that IsoD is actually related to IsoP production while BB/K rate isn't.
The correlation of IsoD to SECA (our preferred secondary base acquisition metric) is much stronger because SECA incorporates BB already. Yet so does BB/K rate and IsoD holds a much stronger correlation with our preferred behavior.
Here's the kicker...
IsoD to Batting Average: -0.141
BB/K to Batting Average: 0.256
I have a feeling that folks using BB/K rather than IsoD are looking for BA-driver behavior. BB/K is somewhat useful in identifying behavior that may produce a slightly higher Batting Average over time. That being said...
BA to RC/27: 0.720
SECA to RC/27: 0.875
If BA doesn't correlate as highly with run production behavior versus better numbers, looking for a potential driver of that less productive behavior isn't the best way to go about it. Instead, what we'd rather do is search for the best driver/driver/driver chain. That's what we've done with IsoD and IsoP and we can apply those metrics at the individual hitter level to give us a pretty decent shot at realistically projecting performance over the long haul.