Monday, March 19, 2018

Quants and uncertainty - Post-February thinking


As far as the laws of mathematics refer to reality, they are not certain; and as far as they are certain, they do not refer to reality.  -Albert Einstein


Many quant firms took a return hit with February performance. It was not just for trend-following but also for many risk premium programs. Short-term traders who were able to actively trade the volatility shock were an exception. 

For most firms, these poor returns were unexpected. First, the programs may not have anticipated the price patterns associated with this volatility shock because they just do not occur frequently. This is a sampling and testing problem. Second, models often make risk management assumptions that are not grounded in reality. Whether liquidity shocks, fund failures, odd behavior by market participants or price moves after the close, models are often not built for all of the structural events that may occur. Models are based on simplifying assumptions and not formed to account for all of the realities of trading. 

Given the slippage between modeling and reality, firms have three options for program protection and it is important to find out how managers approach the model reality error problem. 

One option is the scenario approach to reality. Managers should stress a portfolio based on events that are possible but not actually found in the empirical data. Be ready for what has not yet occurred. 

Second, managers should use a precautionary principle to portfolio construction. Build a portfolio that can survive any extreme event. A precautionary principal can take the form of a barbell between low risk and high risk portfolios. The high risk portfolio can potentially take a devastating hit, but still leave the overall portfolio able to survive to some predetermined loss. 

Finally, a core principle can be to never lever a portfolio beyond a set level especially in a low volatility world. A key problem was with the target volatility crowd is that they continued to add positions to maintain volatility targets in a low volatility environment. A max leverage principle would have capped the leverage regard of target volatility levels. Volatility may stay below the target level but leverage caps are more important.

Working to account for extremes that may never be seem may seem cautious and clearly leave money on the table if the risk event never occurs, but it is the only way to protect against the slippage between models and reality that may occur in a February shock.

No comments: