There has been a significant amount of research on the value of managed futures when there is a crisis, but most of the work has been empirical. The results show that there is a strong negative correlation between managed futures and large stock declines. This conditional correlation has been referred to as crisis alpha. Crisis alpha is measurable, but there has been little discussion on why markets dislocate and why some strategies such as managed futures will do well during these special times.
Nevertheless, a recent paper in the American Economic Journal: Microeconomics provides some insight on why divergent trading may be successful. It does not discuss managed futures or investment strategies but focuses on the bigger issue of why markets may dislocate. The paper is called "Falling Dominos: A Theory of Rare Events and Crisis Contagion" and was written by Heng Chen and Wing Suen The paper's premise is game theory approach with Bayesian learning. Rare events can lead to crisis contagion when market participants start to rethink how the world operates and change their perceptions of the environment.
Think of the world as falling into two states, tranquil and frantic. You are not certain which state you are in. More importantly, if you have some rare event, it may shake your guess on the state you are in. If there is a rare trigger event, belief revision can lead to contagion because you will think the "attack" which caused a crisis is more likely to be successful. The next attack could be stronger and the defender may be less able to protect against this attack.
A rare financial crisis may be a good example of a trigger that can cause investors to change expectations. When this change occurs, investors will likely start to reprice assets by thinking a crisis is more likely somewhere else. In the case of the Asian crisis, a rare event in one country started to spill-over to others as investors thought a crisis would be more likely elsewhere. A large bank failure could easily trigger a view that other failures are possible. A credit failure may cause investors to think that other firms are also weaker. The model is not about herding but a rational updating of expectations.
Once the revisions of expectations start to spread across market or countries, there will be price divergences and dislocations. The sum of dislocations, correlated divergences, will mean those that trading the dislocations or trends will be profitable. The diversification that may normally exist may hamper the size of trading gains, but a correlation or contagion of expectation changes will mean those strategies that have positive convexity or crisis alpha will be more profitable. This is a simple explanation but one that should suggest that positive convexity strategies have value when rare events occur.