Quantitative Strategies for Trading Volatility
Bruno Dupire Head of Quantitative Research Bloomberg L.P.
CBOE Dublin, September 7, 2012
Known Unknowns
I know one thing, that I know nothing. Plato
Known Unknowns
There are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns there are things we do not know we don't know. United States Secretary of Defense Donald Rumsfeld
Known Unknowns
I know very accurately how much I do not know. VIX market
OUTLINE
I. Frequency games II.Term structure games III. Historical/implied games
I. Frequency Games
SPX 2006 to 2011 Data
Historical volatility tends to depend on the sampling frequency
Data from 2006 to 2011 25 24.5 24
Historical Volatility
23.5 23 22.5 22 21.5 21 20.5 20 0 5 10 15
Periodicity
SPX 2006 Data
10
Data from 2006
9.5
Historical volatility
8.5
7.5
10
15
Periodicity
SPX 2007 Data
16.5 16 15.5
Data from 2007
Historical volatility
15 14.5 14 13.5 13 12.5 12
10
15
Periodicity
SPX 2008 Data
42
Data from 2008
40
Historical volatility
38
36
34
32
30
10
15
Periodicity
SPX 2009 Data
27.5
Data from 2009
27
Historical volatility
26.5
26
25.5
25
10
15
Periodicity
SPX 2010 Data
18.2
Data from 2010
18
Historical volatility
17.8
17.6
17.4
17.2
17
16.8
10
15
Periodicity
SPX 2011 Data
Data from 2006 to 2011 23.5 23 22.5
Historical Volatility
22 21.5 21 20.5 20 19.5 19 18.5 0 5 10 15
Periodicity
Historical Vol / Historical Vol Arbitrage
If weekly historical vol < daily historical vol : buy strip of T options, -hedge daily sell strip of T options, -hedge weekly Adding up : do not buy nor sell any option; play intra-week mean reversion until T; final P&L :
RVTdaily RVTweekly
RVTdaily RVTweekly
Bruno Dupire
14
Daily Vol / weekly Vol Arbitrage
-On each leg: always keep $ invested in the index and update every t -Resulting spot strategy: follow each week a mean reverting strategy -Keep each day the following exposure:
.(
where
1 1 ) S t i , j S t i ,1
ti , j is the j-th day of the i-th week
-It amounts to follow an intra-week mean reversion strategy
Bruno Dupire
15
Intra week mean reversion strategy
Data from 05/04/2009 to 05/18/2009 930 925 Short 920 915 Short
SP500
910 905 900 895 890 885 880 1 2 3 4 5 6 7 8 9 10 11 Long Long
II. Term Structure Games
VIX and VSTOXX
VIX and VSTOXX
SPX and VIX (MR portfolio)
SPX Index and VIX Index Data from 4/1/2010 to 4/1/2012 50 45 40 35 30 25 20 15 10 900
VIX Index
950
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450
SPX Index
SPX and VIX (MR portfolio)
SPX Index and VIX Index Data from 4/1/2010 to 4/1/2012 50 45 40 35 30 25 20 15 10 900
VIX Index
950
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450
SPX Index
SPX and VIX (MR portfolio)
SPX Index and VIX Index Data from 4/1/2010 to 4/1/2012 50 45 40 35 30 25 20 15 10 900
VIX Index
950
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450
SPX Index
SPX and VIX (MR portfolio)
Time series of portfolio P&L 5 5 Occupation time 4 4
-1
200 Time
400
600
-1
50
100
150
SPX and VIX (MR portfolio)
Time series of portfolio P&L 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 -0.5 04/01/10 09/06/10 02/11/11 Time 07/20/11 12/25/11 06/01/12
SPX and VIX
SPX Index and VIX Index Data from 4/1/2010 to 4/1/2012 50 45 40 35 30 25 20 15 10 900
VIX Index
950
1000 1050 1100 1150 1200 1250 1300 1350 1400 1450
SPX Index
SPX and VIX
SPX Index and VIX Index Data from 4/1/2010 to 4/1/2012 50 45 40 35
VIX Index
30 25 20 15 10 5 0 900 950 1000 1050 1100 1150 1200 1250 1300 1350 1400 1450
SPX Index
SPX and VIX
SPX and VIX Data from 4/1/2010 to 4/1/2012 50 45 40 35 30
VIX
25 20 15 10 5 0 950 1000 1050 1100 1150 1200 1250 1300 1350 1400
SPX
SPX and VIX
Time series of portfolio P&L 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 -0.5 0 200 Time 400 600 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 -0.5 0 50 100 150 Occupation time
SPX and VIX
Time series of portfolio P&L 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 -0.5 04/01/10 09/06/10 02/11/11 Time 07/20/11 12/25/11 06/01/12
VIX and V2X (MR portfolio)
VIX Index and V2X Index Data from 1/1/2009 to 9/10/2011 60 55 50 45
V2X Index
40 35 30 25 20 15 10
20
30
40
50
60
70
VIX Index
VIX and V2X (MR portfolio)
VIX Index and V2X Index Data from 1/1/2009 to 9/10/2011 65 60 55 50
V2X Index
45 40 35 30 25 20 15 10 20 30 40 50 60 70
VIX Index
VIX and V2X (MR portfolio)
VIX Index and V2X Index Data from 1/1/2009 to 9/10/2011 65 60 55 50
V2X Index
45 40 35 30 25 20 15 10 20 30 40 50 60 70
VIX Index
VIX and V2X (MR portfolio)
Time series of portfolio P&L 7 6 5 4 3 2 1 0 -1 7 6 5 4 3 2 1 0 -1 Occupation time
200
400 Time
600
800
50
100
150
200
VIX and V2X (MR portfolio)
Time series of portfolio P&L 7 6 5 4 3 2 1 0 -1 -2 01/01/09
09/26/09
06/22/10 Time
03/17/11
12/11/11
09/05/12
VIX Futures
34 32 30 28 26 24 22 20 18
2-D representation
VIX futures
Jan
Feb
Mar
Apr
Maturity
VIX Futures and VStoxx Futures
34 32
2-D representation
VIX & VStoxx Futures
30 28 26 24 22 20 18
Jan
Feb
Mar
Apr
Maturity
VIX Futures and VStoxx Futures
45
2-D representation
40
VIX & VStoxx Futures
35
30
25
20
Jan
Feb
Mar
Apr
Maturity
III. Historical/Implied Games
Historical/implied moments
Volatility Skew: slope of implied volatility as a function of Strike Link with Skewness (asymmetry) of the Risk Neutral density function ? Moments 1 2 3 4
Bruno Dupire
Statistics Expectation Variance Skewness Kurtosis
39
Finance FWD price Level of implied vol Slope of implied vol Convexity of implied vol
S&P 500: Option Prices
Bruno Dupire
40
Non parametric fit of implied vols
40
SPX Implied Vols on 31-Jan-2o12 (1M)
35
Implied (%)
30
25
20
15
10 1000
1100
1200
1300
1400
1500
1600
1700
K
Bruno Dupire 41
Implied Volatilities
Bruno Dupire
42
Local Volatilities
Bruno Dupire
43
Risk Neutral Densities
0.01 0.009 0.008 0.007 0.006
Risk Neutral Density SPX on 31-Jan-2o12 (1M, 3M, 1Y)
PDF
0.005 0.004 0.003 0.002 0.001 0 500
750
1000
1250
1500
1750
2000
Bruno Dupire
SPOT 44
Theoretical Skew From Price History
Theoretical Skew from Prices
? =>
Problem : How to compute option prices on an underlying without options? For instance : compute 3 month 5% OTM Call from price history only. 1) Discounted average of the historical payoffs. Bad : depends on bull/bear, no call/put parity. 2) Generate paths by sampling 1 day return recentered histogram. Problem : CLT => converges quickly to same volatility for all strike/maturity; breaks autocorrelation and vol/spot dependency.
46
Bruno Dupire
Theoretical Skew from Prices (2)
3) Discounted average of the Intrinsic Value from recentered 3 month histogram. 4) -Hedging : compute the implied volatility which makes the hedging a fair game.
Bruno Dupire
47
Strike dependency
Fair or Break-Even volatility is an average of squared returns, weighted by the Gammas, which depend on the strike
Bruno Dupire
48
Strike dependency for multiple paths
Bruno Dupire
49
Theoretical Skew from historical prices
Bruno Dupire
50
S&P500 2006
SP500 1/3/2006 to 1/3/2009 BEVL surface 1/3/2006 Implied vol surface 1
0.8
0.6
0.4
0.2
0 80 90 100 110 120 1m 3m 2m maturity 1y 6m 18m 2y
Bruno Dupire
moneyness
51
S&P500 2008
SP500 10/1/2008 to 11/22/2010 BEVL surface 10/1/2008 Implied vol surface 0.65 0.6 0.55 0.5 0.45 0.4 0.35 0.3 0.25 0.2 80 90 100 110 120 1m 2m 3m 6m 1y 18m 2y
Bruno Dupire
moneyness
52
maturity
Conditional Realized Variance
Assume dSt = vt dWt , then E[vt ST = K ] E[vt S t' = E[ St ST = K ]] = vloc ( E[ S t ST = K ], t ) It follows that E[ RVT ST = K ] = E[ vt dt ST = K ] = E[vt ST = K ] dt
0 0 2 = vloc ( E[ S t ST = K ], t ) dt impl (K ,T )T 0 T T T
In summary
2 ( K , T ) T E[ RVT ST = K ] impl
We apply this approximation to historical realized variance
Bruno Dupire 53
S&P500 example
Realized Variance vs Final Return 0.12
0.1
0.08 Realized Variance
0.06
0.04
0.02
0 85
90
95
100 Final Return
105
110
S&P500 example
Realized Vol vs Final Return 0.4 0.35
0.3
Realized Vol
0.25
0.2
0.15
0.1
0.05 85
90
95
100 Final Return
105
110
Density extracted from previous skew
0.12
0.1
0.08
0.06
0.04
0.02
0 85
90
95
100
105
110
Summary
We can play with volatility even without options Term structure of volatility can be exploited Careful with implied/historical games More complex games with more dimensions
Bruno Dupire
57
