I wouldn't put a whole lot of weight...well, actually, I wouldn't put any weight at all on Allan Lichtman's prediction that Barack Obama will be re-elected, based on his "keys" system. What he has is a combination of things that are generally causes (such as the economy) of incumbent party success, things that are effects of incumbent party success (such as the incumbent winning renomination uncontested and third party challenges), and things that are arbitrary and dubious (such as whether the candidates have "charisma").
It's not surprising that you can "predict" the winner with that batch of stuff. After all, while Lichtman's system has worked since he debuted it for the 1984 cycle, a much simpler system that predicts the incumbent party wins barring an election-year recession also successfully calls the winners from 1984 through 2008, at least if you count 2000 for Gore (as Lichtman does). That doesn't make Lichtman wrong as much as it just means his system isn't telling us much that we don't already know otherwise.
(Updated w/typo fixed)