ESPN's Christina Kahrl has a tip when examining early run differential stats: Don't. And she uses the 2010 Astros as the reason why.
They finished the year 76-86, with a run differential of minus-118. By the Pythagorean theorem of expected wins and losses, that translated to a record eight games worse. Voilà, a conclusion: the 2010 Astros overperformed because they should have won 68 games...
Except the Astros didn't overperform, and that's because of the way their season broke down early. In the first two months, rookie skipper Brad Mills' ballclub went 17-34, with a run differential of minus-101, which gives you an expected winning percentage of .285, worse than the 1887 Cleveland Spiders. The 2010 Astros weren't just initially bad, they were supposed to be even worse.
Click the link for the whole thing, because no one can be worse than the 1887 Cleveland Spiders.