Credit Suisse - Skew
Credit Suisse - Skew
Credit Suisse - Skew
1
Derivatives Strategy
16
20 14
12
15
10
8
10
6
5 4
0 0
02/03/00 13/03/01 25/03/02 07/04/03 20/04/04 29/04/05 11/05/06 23/05/07 04/06/08 02/03/00 13/03/01 25/03/02 07/04/03 20/04/04 29/04/05 11/05/06 23/05/07 04/06/08
Source: Credit Suisse Derivatives Strategy
2
Derivatives Strategy
Exhibit 5: SX5E 3M skew: clean calc. vs proxy Incidentally, there is acleaner way to calculate skew, still using
18 1600 fixed strike volatilities. For instance, using 90% and 100% strike
16 Clean calc.
Proxy (rhs)
1400
volatilities: Skew = (90% vol 100% vol) x 100% vol. Multiplying
by the level of volatility penalises the calculated volatility skew
14
1200
12
10
1000
when volatility is low, while inflating it when volatility is high,
8
800
therefore correcting the afore-mentioned volatility bias. As shown
6
600
on Exhibit 5, this gives results that are very close to the clean
4
400
calculation although at an entirely different scale but without the
2 200
hassle of having to calculate volatility skew in delta space.
0 0
Vol Skew 50
Debt/Equity 60.6% 78.2%
40
Debt/Capital 55.1% 83.7%
Skew
20
10
0
0 0.2 0.4 0.6 0.8 1 1.2
Debt to Capital
3
Derivatives Strategy
level of macro risk, in turn increasing index volatilities and the index
volatility skew.
Exhibit 8: SX5E and Total SA 3M volatility skews Exhibit 9: SX5E 3M Implied correlation skew
80 90
70 SX5E 80
TOTF.PA
60 70
50 60
40 50
30 40
20 30
60% 70% 80% 90% 100% 110% 120% 130% 140% 60% 70% 80% 90% 100% 110% 120% 130% 140%
30
10
PCA analysis
8
In order to provide an answer to this question, we look at the
20 6 dynamics of implied volatilities for different levels of delta but
4 identical, rolling maturity. Our objective is to find only a handful of
10
2 factors that will explain almost all variations of implied volatility of all
0 0
options with the same maturity and in particular explaining the
02/03/00 13/03/01 25/03/02 07/04/03 20/04/04 29/04/05 11/05/06 23/05/07 04/06/08 overall level of volatilities and the volatility skew.
Source: Credit Suisse Derivatives Strategy
For investigations involving a large number of observed variables, it
is often useful to simplify the analysis by considering a smaller
number of linear combinations of the original variables. Principal
Components Analysis, or PCA, finds a set of orthogonal (that is,
independent) linear combinations, deemed the Principal
Components, which together explain the variance of all of the
Exhibit 11: SX5E vols: factor 1 loadings original data.
0.45
0.25 99.4% of all variations in the implied volatility for all deltas.
0.2
As shown on Exhibit 11, the factors loadings (basically the
0.15 sensitivity of implied volatilities to factor 1) is positive for all deltas.
0.1 It also tends to decrease linearly for higher deltas. Overall factor 1
0.05 is therefore having a direct impact on both the general level of
0 volatility AND the volatility skew, with a high reading in factor 1
10d 20d 30d 40d 50d 60d 70d 80d 90d translating into high vols and higher skew. This is the reason for
Delta
Source: Credit Suisse Derivatives Strategy the high correlation we spotted just a few lines above.
4
Derivatives Strategy
-0.4
Our conclusion from PCA analysis is that volatility and skew tend
to be the same thing. Only on rare occasions does skew actually
-0.6 gains a life of its own and that specific trades could be put in place
to capture it.
-0.8
Delta
Source: Credit Suisse Derivatives Strategy
Trading the skew with vanilla options
Put options and option combinations
" A put option is a derivative offering its holder the right (not the obligation) to sell a predetermined asset at a predetermined strike price at
a predetermined date in the future. Payoff at expiration is therefore:
" Max ( 0, Strike Spot)
" A put option is said to be out-of-the-money if the spot trades over the strike price, at-the-money if it trades at the strike, or in the money
if it trades below the strike price (that is, the option would have positive pay off if we were at expiry).
" Given that the payoff of the put option is always positive, the trade can only be entered at a premium the seller being rewarded for
taking the risk to have to buy the stock at an unfavorable price in the future.
" By buying/selling put options with different strikes, an investor can get sophisticated exposure to the share price that are not available to
the long or short only investor, at a cheaper premium than a single put option, or even at a credit.
" Example of such combinations include put spreads (long one near-the-money put and short one out-of-the-money put), 1x2 put spreads
(long one near-the-money put and short two out-of-the-money puts), or vega-neutral put spreads where just enough out-of-the-money
puts are sold to create a zero sensitivity to moves in volatility.
5
Derivatives Strategy
Exhibit 13: SX5E 3M 80%/120% risk reversal Exhibit 14: SX5E 3M 80%/ATM put spread
900,000 1500 1750 2000 2250 2500 2750 3000 3250 3500 3750 4000
Option premium (EUR)
0
800,000
700,000 -200
Delta = +250 futures
Hedge = -250 futures)
Option delta (shares)
600,000
-400
500,000
New Delta = +390 futures
400,000
-600 New Hedge = -390 futures)
300,000
Convexity (gamma) P&L
200,000 -800
Delta = -250 futures
100,000
-1000
0
1500 1750 2000 2250 2500 2750 3000 3250 3500 3750 4000
SX5E spot price
-1200
6
Derivatives Strategy
At the same time the hedge consisting of 250 futures will drop in
value by roughly EUR46,000 (or the change in the futures value
times number of futures: (2685 2500) * 250), leaving the trader
with an overall P&L of EUR11,000.
Exhibit 17: SX5E Dec08 put gamma summary Assume now that you re-hedge at the new delta of 390 shares,
Current SX5E at 2,500 SX5E back to 2,685 Total P&L and that the SX5E goes back to its starting point. Loss on the
Option premium 77,000 134,000 77,000 option position will therefore be EUR57,000, cancelling previous
Option P&L 57,000 -57,000 0 options gain. However, the gain on the hedge will more than make
Hedge (nbr futures) 250 (old)/390 (new) 390
up for the options loss: (2,685 2500) * 390 = EUR72,000,
Hedge P&L -46,000 72,000 26,000
Total P&L 11,000 15,000 leaving the trader with a net P&L on the SX5E rebound of
Source: Credit Suisse Derivatives Strategy
EUR15,000. Total net gain on the SX5E fall and rebound is
EUR26,000.
Overall, if the underlying moves enough between re-hedgings, the
cumulated P&Ls will more than compensate the initial premium
paid for the option and the net P&L will be positive. This happens
when subsequent realized volatility is higher than the implied
volatility at the time of pricing.
Conversely, a short gamma position (for instance here selling the
put option and hedging it by going short futures) would then have
a negative net P&L.
200,000 -400
0 -500
1000 1500 2000 2500 3000 3500 4000
-200,000 -600
-800,000 -900
-1,000,000 -1000
Source: Credit Suisse Derivatives Strategy
7
Derivatives Strategy
However we can also see that below 2,250, the risk reversal
Exhibit 20: SX5E change in 1M vol vs pct change in index actually regains some convexity. Should the SX5E fall down to
15 1,500, the new delta would be -719 futures, meaning we would
y = 0.107 - 1.009 * x
have to go long 572 futures to maintain our long hedge.
Change in fixed strike vol (vol points
10
Until 1st Sept
After 1st Sept
What if volatility is remarked?
5 However, the previous analysis disregards an important fact: when
the underlying plunges, volatility tends to be remarked. As shown
-10 -5
0
0 5 10 15
on Exhibit 20, on a fixed strike basis, SX5E 1-month implied
volatilities have tended to increase by as many vol points as
-5
percentage point the index fell. Although the relationship is
relatively loose (r-squared of the linear regression is only 18%), it
-10
Pct change in the index
cannot be ignored, in particular for extreme, down moves in the
Source: Credit Suisse Derivatives Strategy
index.
On Exhibit 21, we show how the delta on the risk reversal behaves
Exhibit 21: SX5E Delta: current vols vs vols up 10 vol pts when the overall level of implied volatility is remarked up by 10%.
0
1000 1500 2000 2500 3000 3500 4000
-100 Delta = -110 futures (no remark) Delta = -147 futures Reminder: at current prices the calculated delta on the risk reversal
-200
Hedge = +110 futures) Hedge = +147 futures) is -147, meaning the vol trader has to go long 147 futures to
-300
Current Vols hedge the options.
True Delta = -719 futures Vols up 10pts
-400 Re-Hedge = +719 futures) If the SX5E falls to 2,500 (same scenario than before when new
-500 delta was -110 futures), but if this time the vol trader remarks
-600
volatilities up by 10 volatility points, the actual new delta will be -
182. A delta of a larger scale had to be expected, since the put
-700
option would now have a larger probability of ending in-the-money
-800 according to the new implied volatility.
-900
In order to maintain his hedge, the trader will have to buy another
-1000
Source: Credit Suisse Derivatives Strategy
35 futures where as if vols were not remarked, he would have had
to sell 37 futures a trade that is typical of long gamma, not short
Exhibit 23: Expected P&L from different vol scenarios gamma positions.
SX5E at 2,500/ SX5E back to 2,685/
Vols marked up Vols marked down Long vol of vol, short spot/vol correlation
Current 10 pts 10pts Total P&L
Option premium 1,180 22,300 1,180
Coming to P&L: assuming the SX5E falls to 2,500 and that
Option P&L 21,100 -21,100 0 volatilities are marked up 10 volatility points, the new price of the
Hedge (nbr futures) 14747 (old)/182 (new) 182 option position will be EUR22,300, roughly landing a P&L of
Hedge P&L -27,200 33,700 6,500
Total P&L -6,100 12,600 EUR21,100 on the options position. At the same time, the loss on
Source: Credit Suisse Derivatives Strategy the futures long hedge will be 147 * (2685 2500) = 27,200 so
net loss will be EUR6,100.
Exhibit 24: Expected P&L from different vol scenarios The position is re-hedged at the new delta of -182. If the SX5E
20,000
index returns to its starting point then net loss on the options
position will be EUR21,100 (cancelling the previous options gain),
Expected P&L from rehedging delta (EUR)
15,000
while P&L on the futures will be EUR33,700, landing a net gain
10,000 on the rebound in the SX5E of EUR12,600.
5,000
In total, the trader would be able to lock in a P&L of EUR6,500 on
the fall and rebound in the SX5E.
0
-5 0 5 10 15 20 Actually, the real driver of P&L in the above example is the scale
-5,000 of the remark in implied volatilities. Had the trader not remarked
volatilities before re-hedging, then he would not have been able to
-10,000
show any profit in the risk reversal. Exhibit 24 shows the P&L
-15,000
expected from a fall in the SX5E down to 2,500 followed by a
Change in fixed strike volatility (vol points)
Source: Credit Suisse Derivatives Strategy
8
Derivatives Strategy
10
Until 1st Sept index over the last two months range from -7.9% to 11% while
5
After 1st Sept
changes in volatility (fixed strike) range from -8.3 to 10.3 volatility
points.
-10 -5
0
0 5 10 15
Exhibit 26 plots the SX5E vol of vol, defined as the volatility of
percentage changes in the fixed strike volatility, since 2000. SX5E
-5
vol of vol has now reached decade highs of over 75% after two
weeks of active index volatility remarking, higher than the 50%
-10
Pct change in the index peak reached during the 2002 credit crunch. This comes at a time
Source: Credit Suisse Derivatives Strategy
when spot/volatility correlation is also at a low level (Exhibit 27).
Exhibit 26: SX5E 1-month vol of vol (%) Exhibit 27: SX5E Dec08 risk reversal delta (shares)
90 0.8
80 0.6
70 0.4
60 0.2
50
0
40 04/02/00 02/02/01 01/02/02 31/01/03 30/01/04 28/01/05 27/01/06 26/01/07 25/01/08
-0.2
30
-0.4
20
-0.6
10
-0.8
0
04/02/00 02/02/01 01/02/02 31/01/03 30/01/04 28/01/05 27/01/06 26/01/07 25/01/08 -1
9
Derivatives Strategy
Exhibit 28: SX5E PCA factor 2 Pure skew at decade highs short skew?
0.1
On Exhibit 28, we back-calculated the unobserved factor 2 (which
0.08
we defined as pure skew factor) based on factor loadings and
0.06
past implied volatilities. Factor 2 is now at decade lows of -0.08,
0.04
or 4.7 standard deviations below its average.
0.02
0 In our PCA analysis of the volatility skew on page 4, we mentioned
02/03/00 29/03/01 09/04/02 22/04/03 04/05/04 13/05/05 24/05/06 06/06/07 17/06/08
-0.02 that skew tends to be closely correlated to the overall level of
-0.04 volatility. Only in rare occasions does skew actually gain a life of its
-0.06 own and that specific trades could be put in place to capture it.
-0.08 This is apparently one of those moments. In our last volatility
-0.1 outlook Now its the Economy, dated 3rd November, we suggested
-0.12
Source: Credit Suisse Derivatives Strategy
buying Dec08 put spreads as a hold-to-maturity investment,
making the point that current inflated levels of the volatility skew
were still pricing in a high probability of a market collapse although
the financial crisis was being addressed and we had moved to a
more familiar recessionary environment.
With this new publication we showed that in the current
environment, there was mutual interest from trading desks to buy
the skew in order to trade the high vol of vol, and from more
traditional investors to sell it to them, potentially under the form of
put spreads, in order to benefit from better pricing stemming from
record-high skews.
10
Derivatives Strategy
Asia/Japan
Dan Wheeler +44 20 7888 4105 [email protected]
Structured/Hedge Funds
Matt Shelton +44 20 7888 7032 [email protected]
Stephanie Hoefling +44 20 7888 4248 [email protected]
11