Considering that they were into statistical arbitrage (at least to my understanding) and considering that most of their positions would have then most likely been partially hedged (considering the typical stat arb model) with probably a pretty low VaR, I do find it quite surprising and scary that things collapsed that fast. A 15:1 leverage ratio seems much more reasonable than 45:1 considering that nature of an arb portfolio.
I don't know what could have been done differently by LTCM. Not sure how else LTCM could have measured the risk though - VaR with stress testing is probably the best method for the type of portfolio LTCM was running.
On a more personal note, I have always been wary of risk models that use correlations between securities to measure risk. I am not sure if LTCM was the first such instance I read about, but it does happen that correlations get disjointed over the short term. Reminds me of the following quote: "Markets can stay irrational for longer than you can stay solvent."
SMCJB - thanks for the link. Will give it a read once I get a chance.