Well, of course he won the 2004 Iowa Caucus. He was the incumbent president and de facto nominee. I'm just wondering, though, what the track record of the Iowa Caucus looks like before the year 2000.
Does your theory hold up before then? I haven't googled to find the answer.
Let’s look at how all the recent nominees did in Iowa and New Hampshire, not counting incumbent Presidents Winner in () if the nominee didn’t win
2008
Dem: Obama
Iowa 1st
NH 2nd (Hillary)
GOP: McCain
Iowa 4th (Huckabee)
NH 1st
2004
Dem: Kerry
Iowa 1st
NH 1st
2000
Dem: Gore
Iowa 1st
NH 1st
GOP: Bush
Iowa 1st
NH 2nd (McCain)
1996
GOP: Dole
Iowa 1st
NH 2nd (Pat Buchanan)
1992
Dem: Clinton
Iowa 4th (very! low turnout, no one contested it cause Tom Harkin ran, “uncommitted” came in second)
NH 2nd (Tsongas)
1988
Dem: Dukakis
Iowa 3rd (Dick Gephardt)
NH 1st
GOP: Bush
Iowa 3rd (Bob Dole)
NH 1st
1984
Dem: Mondale
Iowa 1st
NH 2nd (Gary Hart)
1980
GOP: Reagan
Iowa 2nd (Bush)
NH 1st
1976
Dem: Carter
Iowa 2nd (”Uncommitted” won)
NH 1st
1972
Dem: McGovern
Iowa 3rd (Uncommitted won followed by Muskie)
NH 2nd (Muskie)
Before 1972 the Iowa caucuses and most primary elections were irrelevant.
So Iowa usually does not choose the ultimate winner in open seat races.
And since Romney “won” I hope it once again doesn’t! In contested GOP fights (not counting Ford/Reagan, Ford won or Bush/Buchanan 1992, Bush won) it chose the GOP winner only in 1996 and 2000 both, Dole and Bush were strong favorites for the nomination. NH chose the GOP winner in 2008, 1988 and 1980.