And, by the way, 0.83 is not equal to 1. I agree with Tom's point completely, if a stat says a 70 win team and 90 win team are roughly equivalent, it's missing something important. If it were just random flucutation, you'd expect a lot of 70 win teams to turn around into 90 win teams the next year and vice versa; in real life that just doesn't happen.

Really what it's saying is that if you put the context neutral stats in a sim and run it 10,000 times {WAG} then the teams records will wind up similar even if they diverge wildly in some of those 10,000 runs. Nevermind that they were also all in the same division along with the Yankees and with an unbalanced schedule all they would have to do is tank against the very good teams they play against disproportionately.

Are there people out there claiming that FIP+DefMetric=Actual Team Run Prevention?

To me that's a claim implicit in the use of FIP or DIPSera. FIP is fairly clearly an attempt to isolate the pitching component of run prevention.

Good question Monty. The DH should mean that more WAR is allocated to the AL than NL. A typical DH is an above average hitter, maybe +5 runs or so (though 1B usually outhit DH's). With the positional adjustment set anywhere from -15 to -20 runs for a DH, then the DH position, on average, adds 0.5 to 1.0 WAR per team. In my system, the average pitcher's hitting = replacement level, so pitchers as a group aren't adding anything. There are two solutions that I would find acceptable:

1. Consider this as an issue to be very careful with if trying to take 50+ total team WAR as a measure of expected team wins.

2. Ban the DH, now, forever, and retroactively.

To me that's a claim implicit in the use of FIP or DIPSera. FIP is fairly clearly an attempt to isolate the pitching component of run prevention.

But on an individual level. As I said, I agree with you/Tom that the individual level should scale up such that composite player stat values should be at least pretty close to team totals and any inability to do so should cast grave doubt over the utility of the stat, but I don't see where the use of FIP for individuals (or WAR) automatically implies what you think it does. I can't use FIP (even for individuals) without buying into the general framework that FIP+Fielding = Some permutation of runs allowed, but there's a ton of gray area there.

I *could* think that the implicit goal of FIP+Fielding is to approximate actual runs allowed as closely as possible. I could also think that what FIP+Fielding really would show is runs allowed minus statistical noise, which should be a better predictor than actual runs allowed. The latter philosophical starting point would mean it'd be much harder to refine the model the last few bits towards perfect, because I'd have to both determine and valuate the statistical noise portions instead of just being able to look at a known value (RA) as the goal. So anyone going that latter route wouldn't be making that claim of FIP+DefMetric=Actual Team Run Prevention. And anyone who goes for the former route has a long time before they have to stake out that claim. Given that I haven't seen anyone make the claim, it seems clear that it's understood that FIP and whatever defensive metric aren't yet perfect and that the process of combining them and in what proportions still needs to be thought through.

This came up as a benefit to doing the sanity checks, the ability to reject the claim of FIP+DM=ATRP, so all I'm really trying to say is that I don't see that as a benefit right now. Common sense and circumstance have already rejected that claim for the moment. And if you're making the *really* high-level point that it would be better if all this stuff was thought about on a macro level and how it fits in with the other "standard" advanced stat playing field, my agreement doesn't take away the fact that that ain't happening any time soon.