Discussion on Improving Trading Using Correlation Information at RMC

Three presenters worked together on a presentation that asks the question, Can We Improve Trading Using Correlation Information?  Kokou Agbo-Blou from Societe Generale, Neale Jackson of 36 South Capital Advisors, and Trung-Tu Nguyen of Capital Fund Management all divided duties to answer this question.

Each speaker took on various parts of the presentation relating to their individual expertise.  The presentation discussed the difference between long-term correlations and short-term correlations.  This theme was ever present over the three days of RMC, but discussed more often as current market dislocations.

With respect to dealing with highly correlated assets, it was suggested that traders look at four different asset classes for diversification purposes.  These classes are long stock index futures, short equity variance swaps, short volatility index futures, and buying credit default swaps.  For each class US and EU versions of returns were used and compared.

There was an analogy drawn to show a Nash Equilibrium proving necessity for a convex response.  This was possibly the most entertaining demonstration and definitely the first time I’ve seen a duel with (fake) pistols play out at an investment conference.  The point being made was hedging is not a linear decision and approaching a long volatility strategy in a portfolio can result in a dynamic correlation gap if managed improperly.  If you are up against a good shooter in a duel you are probably going to take large steps so you are a smaller target.  If you are up against a terrible shot, maybe you take shorter steps.  Either way, the length of your pace is determined by you assessment of your opponent.

Finally, a table was shown noting the time periods for SPX volatility to move from a peak to a trough.  We are all aware of how quickly VIX dropped back in the August / September period last year.  However, the typical cycles have run in the 30 to 50-day range since 2005 (for context the peak to trough last year was only 6 days).  Since this anomaly from last year, the cycles have run 51, 28, and 30 days – resuming was we had experienced in the past.