Sunday, November 27, 2011

Redefining volatility

Volatility should be redefined to actually reflect the evaluation by clients. Just because something has higher variance does not mean that it is at once worse than something else which is steadier. Also there is no directly observable relationship between variance of returns and expectations of returns by clients.
Real clients in real world look at risk very differently. A measure like standard deviation hence does not quite do the job well. Also sharpe ratio based on this definition of volatility is thus too flawed.

I dont know the right answer just yet. Nevertheless, in a bid to get there, let us start with some simple questions. What do investors really look at while defining risk?
- loss of capital (there is special value loss in losing part of capital; 110 going to 105 is not the same as 100 going to 95); maybe drawdown captures this in a generic sense
- horizon is relevant: investors look at their investments every so often - monthly or daily or quarterly or annually or on every publication of nav. It is the variance in this horizon that is most important. High daily variance coupled with steady monthly returns hence might not be so bad for investors reviewing their portfolio every month.
- downside variance is more important than upside variance. Seeing value fluctuate to go down is worse. If assets appreciate it matters less whether they appreciate with high variance or low variance.

Some other observations can also be added to this. Volatility as a representative of risk then needs to be redefined to include the impact of above parameters. The idea is to match what we call risky in a mathematical sense and what investors perceive as being risky. That then might be useful in creating really client centric products.

No comments: