Practical Relative-Value

Volatility Trading
Stephen Blyth,
Managing Director, Head of European Arbitrage Trading
21 January 2005

Volatility modelling

Construct a consistent framework to identify and extract value from

interest-rate options markets
Use this framework to inform decisions concerning inception and
management of proprietary options trading positions

Modelling paradigm
Insist on consistent pricing and hedging framework: only one US
dollar (or euro) libor yield curve, therefore only one process
random or otherwise driving this curve
Avoid pragmatic use of simplest model per product can result in
inconsistent dynamics
Make sensible choices about objective features to incorporate into
modelling formulation: e.g. multifactor, skew dynamics
Require advanced analytics to develop tractable pricing and risk
management tools often an obstacle to successful implementation

Modelling paradigm (contd)

Pricing inconsistencies between market and model can indicate:
market features
modelling errors
trading opportunities

Use market experience and judgment to identify the latter

Warning: Models have Limitations!
It is wrong to think that the task of physics is to find out how nature is.
Physics concerns what we can say about nature Niels Bohr
Nature

Financial markets

Physics

Quantitative modelling

A trader armed only with a clever model is soon removed

from his capital
4

Modelling paradigm (contd)

Significant advances in quantitative financial modelling over past ten
years. Need judgment to harness these advances
Tukey: judgment based on
mathematical knowledge of the particular techniques
experience of the particular field of subject matter
experience of how these techniques have worked out in practice

Forward volatility surfaces

Translate universe of option prices into how each point on yield
curve (e.g. each Euribor future) oscillates over time
More precisely: strip swaptions, caps and other option products into
a forward volatility surface, (t,T)
For fixed T, (t,T) represents the volatility of a particular Eurodollar
(or Euribor) contract over its life. These forward volatilities are
observable quantities about which can make subjective judgments

Forward volatility surfaces (contd)

Framework adopts multifactor BGM Model
n

df(t, T) = a(t,T) dt + (t, T) i (t, T) dWi (t)

i=1

f (t , T) 3m-forward rate from T at time t

Drift a(t, T) determined from volatility (t, T)

Forward volatility surfaces (contd)

Implement sparsely-parametrized surface. Use market knowledge to
impose reasonable functional forms (subjective judgment about
objective features)
We employ parametric surfaces of form:
1+

exp ( -

)(

1-

: forward, calendar or relative time

Discretize for fully non-parametric surface
Can impose smoothing or linking on nonparametric surface
For further details see Blyth (2004)
8

Parametric forward volatility surface for US dollar

USD Forward Volatility Surface

1.80%
1.60%
1.40%
1.20%
1.00%
0.80%
0.60%
0.40%
0.20%
0
tenor T

0.00%
5

10

15

20

25

30
30

25

20

15

10

time t

Map of a forward volatility surface

Time t

5
Instantaneous volatility
of futures strip
in two years time

10

Tenor T
2

Instantaneous
volatility
of futures strip

Volatility of Dec 2005

future over its life

Volatility of Dec 2007

future over its life

US dollar relative-value indicators

USD residuals
0.60%
0.40%

Rich

0.20%
0.00%
Cheap

-0.20%
-0.40%

3y
4y
5y
Expiration
7y
10y

-0.60%
1y

11

2y

5y
Tenor

7y

10y

20y

Exact fit to US dollar market prices

USD Forward Volatility Surface

1.80%
1.60%
1.40%
1.20%
1.00%
0.80%
0.60%
0.40%
0.20%
0.00%
0

tenor T

12

10

15

20

25

30
30

25

20

15

10

time t

Using forward volatility surfaces in practice

Implement both parametric smooth surface and exact-fit surface
Trading decisions informed by:
Residuals between market and smooth-fit prices
Shape of exact-fit forward volatility surface

Hedge trades under consistent dynamic

Hold to maturity or close out strategies when volatility levels revert to
fair value. Holding periods can vary from days to years
Approach predicated on power of consistent dynamic to offset
inevitable modelling shortcomings

13

Further examples
1. Hedge 10yr CMS liabilities with 5yr-tailed swaptions
10yr-tailed swaptions are rich due to hedging of 10yr CMS product
Portfolio of 5yr-tail options with spectrum of expirations captures
similar volatility, but at cheaper levels
10yr-5yr and 15yr-5yr payers are better value than 10yr-10yr payer

14

Euro relative-value indicators

EUR residuals

0.60%
0.40%

Rich

0.20%
0.00%

Cheap

-0.20%
-0.40%

3y
5y

-0.60%
1y

2y
Tenor

15

5y

7y

10y
10y

20y

Expiration

2. Caps versus swaptions, 1998-9

Unwind of proprietary desks short positions in caps versus long
positions in swaptions widened volatility spread between products
In forward volatility space, this resulted in large spikes in forward
volatility of the front contracts
Volatility dynamic implied by these prices absurd

16

Euro forward volatility surface, Aug 98-June 99:

The approach of the storm

June 1999

August 1998

3.00%

2.50%

2.50%

2.00%

2.00%

1.50%

1.50%

1.00%

1.00%

10.00

time

2.00

8.00
10.00

2.00

6.00
4.00

tenor

time

0.00

0.00%

4.00

6.00

10.00

4.00

8.00

6.00

6.00

8.00

tenor

2.00

4.00

0.00

0.00%

8.00

2.00

0.50%

0.00

10.00

0.50%

0.00

17

3.00%

Euro forward volatility surface, Aug 99:

Inundation

August 1999

7 year column

3.00%
2.50%
3.00%
2.50%
2.00%

2.00%
1.50%

1.50%
1.00%

1.00%

0.50%

2.00

10.00

10.00

2.00

4.00

8.00

6.00

6.00

8.00

tenor

time

0.00

0.00%

4.00

0.50%
0.00%
0.
00
0.
50
1.
00
1.
50
2.
00
2.
50
3.
00
3.
50
4.
00
4.
50
5.
00
5.
50
6.
00
6.
50
7.
00

0.00

Aug-98

18

Jun-99

Aug-99

3. Euro swaption relative value: comparing indicators

Consistent modelling framework can identify better trading strategies
to simpler historical implied analysis - although often indicators
concur.
Consider performance of relative-value trades in euro swaption
market.

19

Euro swaption relative value (contd)

Relative Value Indicators
12%

10%

Percentage Rich/Cheap

8%

6%

Buy 1y5y
Sell 1y20y

4%

2%

0%
Jan-02
-2%

-4%

20

Buy 2y5y
Sell 2y20y

Jul-02

Jan-03

Jul-03

Jan-04

Jul-04

Euro swaption relative value (contd)

Ratio of Implied Normalised Volatilities
1.40

1.35

1.30

2y-5y/2y-20y
1.25

1y-5y/1y-10y
1.20

1.15

1.10

1.05
Jan-02

21

Jul-02

Jan-03

Jul-03

Jan-04

Jul-04

Additional dimensions
Skew Modelling: demand consistency between pricing of skew and
dynamics of implied volatility under market movements
Term Structure of Skew: skew structure for short-term rates coupled
with specified dynamics may not be consistent with certain skew
structure for longer-dated rates
Stochastic Volatility: demand consistency between stochastic
volatility overlays used to price option smile and prices of compound
options

22

Practical Relative-Value
Volatility Trading
Stephen Blyth,
Managing Director, Head of European Arbitrage Trading
21 January 2005