You are on page 1of 6

# Topics in Asset Pricing

Homework 3 - Fixed

Students:
José Ignacio Carter
Fabrizio Monroy
Raúl Pefaur
Moisés Valdebenito

Professor:
Pablo Castañeda

Date:
October 28, 2013
2
Introduction

We changed to the market model “Smile Adjustment with Sticky
Moneyness”, from the paper of Alexander & Kaeck. Given that we need to
simulate 10,000 samples of the P&L of a rolling short position, we created
the following scenario:

We are short in a Call, and five days later, we want to close that position.
At the first day, we also build the hedging portfolio, consisting in an opposite
position on another derivative (which has to be different, so the results are not
trivial: it may change on its maturity, and/or its strike price,but remain a call
option, to preserve pricing and mark to market ease), a position in forward
underlying asset and on a money market account. Plus, each day, until we
close the position, we will recalibrate the hedging portfolio.

Given that we want 10,000 simulations for the P&L of out hedge, we will
use the contracts with maturities 1Y, 6M and 3M, with each of the 7 different
pillars of the smile curve for each maturity, giving that 21 positions to hedge
per day, during 592 days (because we have information for 593 days, but
market information is needed for one additional day to simulate position
closure).

Hedging Strategy

The hedging portfolio will be given by:
( )
0 0 0 0 1
0 0
F F
X C C t t t t
O O
= = ÷ ÷ O + + O
And, in t=t:
( ) ( )
0 0 0 1 0 1
rt rT t T t T
t F
X C C e e F F t t t
÷ ÷ ÷
O O
~ = ÷ O + ÷ + O

We see that at t=0, the short possition we are trying to hedge has the same
value as the portfolio, but not necessarily the same at t=t. The portfolio
changes its value over time: the money market account, the forward contract,
and the option we are using to hedge change their values.

There comes a problem: we want to close the position at day 1, we need
to see the value of the option we are trying to hedge at that time; this means,
3
we have to get the value of an option that is missing, for example 59 days (if
it was originally a 60 day vanilla). This also happens to the discount factors,
the implied volatility, and the forward price; this is information we do not
have, so we will have to interpolate it.

Model Description

Given that we are working in FX trading, the standard in the
aforementioned market is to use Forward prices rather than Spot Prices,
everything in the pricing and hedging equations we are going to use is in terms
of F. Plus, the ‘Sticky Moneyness” model consists in using a variable called
“Moneyness”, which is:
( ) log / K F
m
t
=
we use this term given that the Implied Volatility Surface is a function of
the moneyness. This also means that we will no longer work with Strikes, but
with Moneyness values.

To adjust the smile, we need:
( ) ( ) ( ) ( ) | | | |
adj BSM BSM F
m F m F m F IV m F o o v = +
and
( ) ( ) ( ) ( ) | | | |
adj BSM BSM FF
m F m F m F IV m F ¸ ¸ v = +
( ) ( ) ( ) ( )
2
2 | | | |
BSM F BSM F
m F IV m F m F IV m F q k + +

For simplicity, we will calculate this derivatives as suggested in
Alexander & Kaeck (2011): by its numerical method: A difference between
two deltas or gammas, given an up and a down value for F:

( ) ( )
( ) ( )
, , , , , , , ,
up up down down
SM
up down
BSM F K IV m BSM F K IV m
F F
t o t o
o
÷
=
÷

But, as we explained before, we will not be using K, but its value in
moneyness. This means, the previous equation for delta sticky moneyness
will be:
4
( ) ( )
( ) ( )
, , , , , , , ,
up up up down down down
SM
up down
BSM F m IV m BSM F m IV m
F F
t o t o
o
÷
=
÷

For the adjusted gamma sticky moneyness, we do the same procedure but
as a difference of up and down in two values of delta.

And to compute

IV
m
(moneyness adjusted implied volatility), we
parameterize the implied volatility:

( )
2 3
0 1 2 3
, IV m m m m o o o o o = + + +

The problem with out hedging strategy is that the Greeks have to be
calculated again, given that now we are using moneyness instead of K, and
Forward Price instead of Spot.

Results

First, we needed to calibrate the alphas of the cubic function of
moneyness to have the values of each of them, for every day.
1

The idea is to compare our results with the ones that will be obtained
if using the normal Black Scholes Merton model. The P&L, for 10,000
different hedgings, are as follows:

1
Code “Tarea 3.m”, Local Calibration
5

We can appreciate that our adjusted model has more values around 0 than
the traditional BSM. The main difference is that in our model, we see some
very extreme values; up to 1000% of profit (where BSM maximum is around
350%). A similar scenario occurs to the losses.

Mean 0.0961 0.1736
Stdev 0.6469 0.4327

As we may have suspected, the thin is that our model has better results in
hedging if we focus on the mean: we are closest to zero in all 10,000
simulations than BSM, but with a higher standard deviation; our model’s
results are more widely spread than BSM, but we have a better hedge than it.
The focus is to be near zero, this is a special case where there are profits, but
in another context they may be losses.

There has to be an explaination for this extraordinary scenarios with
extreme high profits/losses. We searched in our database for the days the
highest profits occurs in each model (day 157 for the adjusted, 158 for BSM).
As we may have suspected, the crisis is the explanation; these dates
-6 -4 -2 0 2 4 6 8 10 12
0
2000
4000
6000
8000
P&L histogram, adjusted model
-3 -2 -1 0 1 2 3 4
0
1000
2000
3000
4000
P&L histogram, BSM model
6
(15/09/2011 and 16/09/2011) are when the underlying spot’s volatility rose
almost 100% in just 4 days.

The main conclusion about using the Sticky Moneyness or traditional
BSM is that the Sticky Moneyness hedges better, but in scenarios of shocks
in markets, the hedging for some days may result in extreme values.
Nevertheless, it is a better model, on average, than BSM