Some points on volatility by CSFB.
Key Points
- Quarter after quarter, the volatility of volatility is apparently rising and we are progressively losing confidence in short volatility positions. The balance of risks has been deteriorating quickly during the last few days. A moderate volatility environment now seems likely to prevail.
- We suggest orderly unwinding the risk-on trades opened during the first semester across volatility markets, still with profits. Across carry trades, only the sale of equity index implied correlation still seems promising.
- During the first semester, implied volatility and skew levels eased. Implied volatility is now at a relatively low absolute value and the risk-premium has been substantially melting. Despite some easing, skew however remains rather elevated across indices and can be exploited, directionally (call ladders) as well as for volatility trading (smile-locks).
- We see some remaining pockets of profitability for short variance positions. For instance, the ratio of implied to realized is still comfortably high on the S&P 500. Across emerging markets, the Bovespa screens well for selling volatility and variance, especially in long/short versus the RDXUSD.
In other words, even if we are likely to be still in a low volatility environment, do not take the risk of selling implied volatility on these compressed levels, with realized volatility progressively rising.
Overall, implied volatilities eased over the first half of the year. Despite a succession of shocks at the end of each quarter, none of these have been of a scale significant enough to trigger a durable change of paradigm. Realized volatility stayed low and options sellers have been confident enough to drag down the implied. Focusing specifically on the latest episode of risk aversion, the most interesting conclusion is probably to note how implied volatility eased faster than it rose. Post the Greece parliament voting, at-the- money mark obviously eased due to the rally of equity prices. However, even at constant strike, the easing was significant: by 1 point on the EURO STOXX 50 Dec11 2800 strike for instance.
Across the main benchmarks (Fig10), the S&P 500’’s 1Y implied volatility remains the cheapest, close to 18%, while, around 21%, the EURO STOXX 50’’s is the most expensive. However, as we will see, when compared to the realized, the S&P 500’’s volatility is the most attractive to sell, as it was at the end of Q1. In Asia, the Kospi 200’’s briefly rose above the Nikkei 225 before easing with the substantial bounce of the index and the 1Y implied volatility stand close to 19% for both indices while the Hang-Seng is an inch higher, at 20%. In Europe, the cheapest 1Y implied volatility remains the SMI (15%) followed, by the AEX and the FTSE 100 (approximately 17%). The most expensive implied volatilities are IBEX and FTSE MIB (23%/24%) followed by the EURO STOXX 50 (22%). At 20%, the DAX stands in the middle of the range.
It is less of an attractive purchase than it was at the end of Q1, however it is good to keep its relative cheapness in mind and avoid short gamma/variance/volatility on the German index.
Despite the recent renewed tensions around the periphery of Europe, there has not been much divergence between the lowest and the highest volatilities in the area. At the difference of Q1 end, we are more keen to bank on widening, seeing ahead, than further tightening.
Skew still rich but fading some (climb of worry still?)
Over the last three months, and at some exception (AEX, DAX), the skew levels slightly faded across indices (Tab4). The most significant easing have been seen on the Bovespa (-0.7 standard deviation over the quarter) and the S&P 500 (-0.6 standard deviation). Even if they eased, the skew levels however still remain significantly above their long-term average as shows Tab4. This richness is more blatant for more distant tenors. Fig18 and 19 chart the 5Y and 10Y skew on the EURO STOXX 50 and the S&P 500 – the metrics are at record level. In our opinion, the multiplication of tail risk strategies and the freshness of the 2007/08 events in investors’’ mind lead to sustain interest for small delta options. Specifically on the S&P 500, the expensive skew level reinforces our view that selling delta-hedged downside options and variance are trades to consider with a good degree of confidence for the coming quarter.
With Lehman still in fresh memory a recap of that Fat Tail by CSFB,
What is tail risk?
Tail risk qualifies an extreme outcome that is in the ““tail”” of the distribution of possible outcomes. We commonly use
assumption of a ““normal distribution”” as it is convenient as it leads to mathematically pleasant results. However,
empirically, equity market returns are not normally distributed: equity markets tend to gap down and trend up, making
shocks on the downside more likely and pronounced than on the upside.
Quantifying Tails
Using the EURO STOXX 50 index daily returns as a case study, we find:
- Since 1987, 1% of all daily variations of the EURO STOXX 50 have exceeded 5% in absolute value, to be
compared with a probability < 0.01% for the corresponding normal distribution.
- Although almost half of these moves were on the
Distribution daily log returns of EURO STOXX 50
1987 to H1 2010
upside, investors fear the downside moves.
- Historically, there are more returns on the downside,
(102), contained in the 5% lowest range compared to the
upside, where there are 26 contained in the 5% highest range.
- Although magnitudes are roughly equal, negative
returns –– even of smaller amplitude –– are more likely to occur.
But not all investors are sensitive to extreme daily moves. Some investors are primarily concerned more about the declines
in the their assets over the course of one year.
0 returns represents 0.84% of the returns distribution).
The 1 year returns distribution is impacted by overlapping periods, resulting in non ndependent returns and a biased distribution. Despite this, we can still use this istribution to measure the magnitude of annual tail events and the relative
Using EURO STOXX 50 index annual returns*, we find:
Distribution of annual log returns of
EURO STOXX 50, 1997 to H1 2010
–– Historically, the distribution is deeply skewed to th
downside.
–– The 50 largest moves = 50 largest declines (5
frequency of up- vs. downside annual tail events.
We conclude that:
␣ Extreme moves are of larger magnitude on the downside than on the upside
␣ Extreme negative returns are more likely to occur (than extreme positive returns)
␣ Downside moves occur more frequently than implied by the theoretical distribution
No comments:
Post a Comment