Manager Writes: Volatility Arbitrage

Time for a new approach?

Originally published in the December 2006/January 2007 issue

The trading of volatility as an asset class within investment banks’ market-making and proprietary trading desks has been around for many years. However, until relatively recently there have been very few funds with a standalone volatility strategy.

Historically, when investors wanted to achieve volatility exposure, they would invest in convertible arbitrage managers. While convertible strategies certainly do contain an element of volatility, this is usually longer dated exposure, and valuations and performance are also very dependent on other factors such as credit and directional stock moves. As the convertible market tends to be longer dated, it is also less liquid and the negative effects of this liquidity trap were witnessed by the sharp contraction of valuations and associated fund NAVs in 2004-05.

Historically the barriers to entry for volatility strategies have been very high, although the significant reduction in transaction costs over recent years and the development of ‘pure’ volatility products, such as variance swaps, have resulted in a burgeoning hedge fund sub-industry based on volatility, either as standalone funds or as part of a multi strategy portfolio.

So what is volatility?

To complicate things there are two concepts of volatility that we must consider. Realised (or historic) volatility is a simple mathematical calculation based on the standard deviation of the log returns of closing prices over a period of time multiplied by an annualisation factor which is the square root of the number of business days in the trading calendar. Realised volatility is a backward looking concept and for any given length of observation period there is only one easily calculated discrete value.

Let us consider the concept of implied volatility, which is a forward looking concept and for which there is no known answer. Instead, implied volatility can be thought of as a measure of the perceived amount of uncertainty of the price action of an underlying security. It is the risk premium associated with the expected price movements looking forward in time.

Furthermore, as there are different levels of perceived risk over different periods of time and different ranges of price movement, there exists not one discrete value but a whole range of implied volatilities for options and other derivatives with different strike prices and maturities. It is these implied volatility levels that make one option relatively more or less expensive than the next and which are subject to market forces and the laws of supply and demand.

Trading volatility

Historically, reliable data on options prices or implied volatility has been scarce and of dubious quality. In contrast, realised volatility is easy to calculate from the closing prices of the underlying shares. This has meant that the main indicator that traders have used to forecast future volatility has been historic volatility, i.e. they have looked at previous price action to give an indicator of forward price action.The main method for judging whether to buy or sell options has been to measure the spread between implied and realised volatilities. There are various models, such as GARCH (Generalised Autoregressive Conditional Heteroskedasticity), which give different weightings to different periods of realised volatility, but in essence the technique is the same.

This works well in a constant regime where volatility is seen to mean revert through cycles of high and low volatility. An example of this is the period from 1996-2002 when every year there was a wide range of observed implied volatility as 1-correlation moves in the market resulted from systemic shocks such as the Russian bond default crisis, the collapse of LTCM, the 911 terrorist activities and the 2002 accountancy scandals. During this period the technique worked well and traders bought relatively cheap options and sold them when they spiked up by relying on this spread between implied and historic volatility.

However, this technique broke down in 2002 when there was a sharp decline in implied volatility, caused by the introduction of low interest rates to stabilise corporate balance sheets and massive supply of convertible bond and structured volatility products. Traders were buying options at levels of implied volatility that looked cheap compared to historic volatility and as volatility ground lower, they bought more and more. Volatility failed to spike, there was no ‘mean reversion’ and banks and hedge funds accumulated ever-increasing inventories of volatility that collapsed in value. The banks switched their volatility focus away from proprietary trading to market-making, but hedge fund managers who had been used to the single recipe of buying cheap and waiting for the spike haemorrhaged cash. This is why volatility funds quickly lost favour.

Funds of funds and other investors wanted the long volatility bias, to provide insurance from adversely sharp moves in the market, but were overwhelmed by the degree of losses from their volatility managers who relied on the old tried and tested method of using the spread between implied and historic.

A new approach

At Old Mutual we now believe that a different approach is necessary. Instead of relying on the standard spread between implied and historic volatility, hoping for volatility levels to pick up and then taking profits on mean reversion, we have developed a model for the option prices or implied volatilities on their own for a large universe of stocks globally. This isolated model for implied volatilities is used to drive a relative value strategy consisting of long volatility and short volatility positions on a diverse list of names across a broad range of sectors and geographical areas.

We believe that we are unlikely to see a constant linear increase in volatility, but rather an increase in the frequency of volatility spikes and therefore aim to generate a consistent and un-correlated alpha from the convergence of over- and under-priced options rather than betting on a global macro pick- up in volatility itself.

Volatility is currently close to historic lows and we believe that the dynamics of the equity derivative and credit markets are such that there is no clear reason for it to trade at significantly higher levels. Instead, with increased M&A speculation the market will become longer or shorter volatility depending on the trading range of the under- lying stockmarkets.

Therefore, rather than speculating on a trend with so many degrees of freedom, the approach of observing price action in equities and credits to drive a quantitative relative value approach is the smart way to generate returns from volatility and the evidence is that investors are showing a lot of interest in this new approach.

Steve Kelso and Paul Jones are managers of the Old Mutual Global Volatility Fund, Old Mutual Asset Managers (UK)