I think we are in for some major shifts in quantitative finance. Since 1987, there have been two “games” 1) is to use valuation of liquid derivatives to price illiquid derivatives and 2) is to use time series techniques to find mispricings and take advantage of them.
I think both games are over. Project 1) has basically involved trying to get more and more sophisticated models of the volatility surface. The problem with this is that it fails when derivative markets become large enough so that illiquidity is a factor and you can’t simply expolate prices. Project 2) I think is over. As more and more people enter into the statistical artibitrage fields, the amount of leverage you need to create any return has increased, and I think we are at the point that people with money are no longer willing to provide capital to hedge funds to do these highly leveraged transactions.
What’s interesting about project 1) is that the data that has come from the markets since mid-September just make no sense using the old models. Once things settle down I suspect that a lot of the old models and games just won’t work.