2. Luc_Faucheux_2021
That deck
2
ยจ We now look at some concepts in Foreign Exchange markets (FX)
ยจ This is at the same time surprisingly easy and surprisingly difficult
ยจ When teaching my students at Fairfield University, the FX chapters are usually the ones that
surprisingly offer the most challenges
ยจ Everyone has travelled or has seen movies about exchanging one currency for another
ยจ Yet once you start introducing volatility for example you get things like the Siegel paradox, or
two-currency paradox
ยจ Once you start introducing the issue of the funding (in which currency is your PL actually
counted and matters), you get people confused on the what is the PL of a simple FX trade
for example
ยจ You can also get confused with the quanto effect when pricing some derivatives
ยจ Oh also I just got let go of my current job at Natixis, so that will give me some more time to
work on those presentations. But if you know of any job for me, am interested.
3. Luc_Faucheux_2021
That deck - II
ยจ In a way we needed all the tools of the Rates world (Bank Account numeraire, deflated
Zeros, change of measure, IOT Leibniz,..) to start really dealing with more than one currency
ยจ Also, as Godel rightly pointed out, notations sometimes can be a pain in the neck
ยจ (he never said that, but I would like to start that rumor)
ยจ In any case, a lot of FX textbooks have example where they use DEM or USD or GBP or YEN,
showing how old they are, and it does not help with the confusion
ยจ So I will try to be extra careful on the notation
ยจ In particular, even good textbooks like Hull on the Siegel paradox, essentially use the same
notation for two different processes, essentially negating the whole point that they were
trying to make
ยจ So, to quote Kurt Godel, โcareful on the notation, the notation is 95% of the workโ
ยจ (again, he did not say that, but starting that rumor is my pet peeve)
3
4. Luc_Faucheux_2021
That deck - III
ยจ Yours truly teaching FX this semester at the Fairfield University Dolan School of Business
4
6. Luc_Faucheux_2021
FX notations
ยจ We are going to try to be a little rigorous here.
ยจ Instead of using USD, DEM, EUR, JPY,โฆwe are going to index the currencies by integers
ยจ So we have currencies ๐, ๐, ๐ โฆ
ยจ We have to come up with a notation for the SPOT Foreign Exchange
ยจ Usual market convention is ๐
ยจ Again, am not sure I like that as usually this is reserved for a variable, but I do not want to
have 2 letters and use ๐น๐, so we will stick with ๐ for now
ยจ NOW, we will choose for convention:
ยจ ๐!,# = ๐!โ# is the value in currency (๐) of 1 unit of currency (๐)
ยจ ๐!,# = ๐!โ# translates into currency (๐) from currency (๐) the value of whatever you
multiply it by (cup of coffee, BigMac, barrel of oil, anything)
6
7. Luc_Faucheux_2021
FX notations - II
ยจ ๐!,# = ๐!โ# translates into currency (๐) from currency (๐) the value of whatever you
multiply it by (cup of coffee, BigMac, barrel of oil, anything)
ยจ So say a cup of coffee in London is 4 GBP (four quids for a cup of joe mate ?!)
ยจ And then the exchange rate is 1.3841
ยจ Which exchange rate ?
ยจ Well that is usually where a lot of the confusion occurs.
ยจ People do not like numbers smaller than 1, so historically, they have decided to quote the
exchange rate for that currency pair in terms of whatever order gives the higher number
ยจ For (YEN,USD) pair, it is around 108, and it is the number of the YEN that you get for 1 USD
ยจ For (GBP,USD) pair, is it around 1.4, and it is the number of USD that you get for 1 GBP
ยจ For (USD,CAD) pair, it is around 1.25, and it is the number of CAD that you get for 1 USD
7
9. Luc_Faucheux_2021
FX notations - IV
ยจ When I was working at FujiCap, I went for a week in London and I was sitting right next to
the FX desk there, and I guess โAn American Werewolf in Londonโ was still a big thing at the
time.
ยจ For some reason the guys on the FX desk kept yelling all day long
ยจ โWatch the crossโ
ยจ โWatch the moonโ
ยจ โStick to the roadsโ
ยจ โKeep off the moorsโ
ยจ Fun timesโฆ.
9
10. Luc_Faucheux_2021
FX notations - IV
ยจ The usual market convention for the (USD,EUR) currency pair exchange rate
ยจ People like to deal with numbers greater than 1
10
11. Luc_Faucheux_2021
FX notations - V
ยจ You can also get the inverse if you want, and start wondering what happens to a stochastic
process when you take the inverse of it (hint: convexity!)
11
12. Luc_Faucheux_2021
FX notations - VI
ยจ In Bloomberg, the convention is that:
ยจ EURUSD is the value in USD of 1 unit of EUR
ยจ USDEUR is the value in EUR of 1 unit of USD
ยจ GBPUSD is the value in USD of 1 unit of GBP
ยจ Back to that 4 quids cup of mocha
ยจ 1 cup = 4 GBP = GBP 4 = ยฃ4
ยจ 1 cup = (4GBP)*(GBPUSD)=(4)*(1.3841)=5.5364USD=USD 5.5364 = $5.5364
ยจ So for a lad living in London (William), 1 cup = ยฃ4
ยจ For a Yankee bloke living in NYC (Mike), that same cup is worth $5.5364
ยจ NOW of course, this is somewhat artificial, because there is no way that Mike could access
that same exact cup like William can, but it is nonetheless a true statement
12
13. Luc_Faucheux_2021
FX notations - VI
ยจ Going from the Bloomberg notation to ours:
ยจ (GBPUSD) = ๐%&',()* = ๐%&'โ()*
ยจ GBPUSD is the value in USD of 1 unit of GBP
ยจ GBPUSD is the value of 1 unit of GBP in USD
ยจ {coffee_cup}GPB is the value of 1 coffee cup in GBP
ยจ {coffee_cup}USD is the value of 1 coffee cup in USD
ยจ coffee_cup USD = coffee_cup GBP. (GBPUSD)
ยจ
+,--.._+01
%&'
=
+,--..!"#
()*
. GBPUSD =
+,--.._+01
()*
.
()*
%&'
ยจ coffee_cup USD = ๐%&'โ()*. coffee_cup GBP
ยจ coffee_cup USD = ๐%&',()*. coffee_cup GBP
13
15. Luc_Faucheux_2021
FX notations - VIII
ยจ GBPUSD=1.3841
ยจ GBPEUR=1.1617
ยจ EURUSD=1.1915
ยจ (GBPUSD)=(GBPEUR)*(EURUSD)=1.1617*1.1915=1.3841
15
16. Luc_Faucheux_2021
FX notations - IX
ยจ So the usual way it is quoted in the market is artificial just to make it easier
ยจ Remember it was the same for Eurodollar future, the quote price is not the market price
ยจ Letโs write Q(USD,GBP)=Q(BGP,USD) the usual way the exchange rate for the currency pair
(GBP,USD) is quoted
ยจ Q(GBP,USD) = Q(USD,GBP) = GBPUSD = ๐%&',()* = ๐%&'โ()*
ยจ Q(USD,YEN) = Q(YEN,USD) = USDYEN = ๐567,%&' = ๐567โ%&'
ยจ Q(CAD,USD) = Q(USD,CAD) = USDCAD= ๐89',%&' = ๐89'โ%&'
ยจ Q(USD,EUR) = Q(EUR,USD) = EURUSD = ๐%&',6%: = ๐%&'โ6%:
16
17. Luc_Faucheux_2021
FX notations - X
ยจ Just to make it a little more lively than ๐, ๐, ๐ and also because we already use those a lot for
the Zero Coupon Bonds, we will go with 1,2,3 most of the time
ยจ In the currency 1
ยจ We have the usual Bank Account numeraire associated to the Risk free measure
ยจ ๐โ2 ๐ก is the Brownian motion associated to the risk-free measure {โ1} which is
associated with the rolling numeraire ๐ต2 ๐ก = exp[โซ
<=>
<=?
๐ 2 ๐ , ๐ , ๐ . ๐๐ ]
ยจ We have in the currency 1 the Zero coupon bonds: ๐2 ๐ก, ๐ก!, ๐ก#
ยจ We will also have in the currency 1 assets (stocks for example, or cups of coffee),
DENOMINATED and tradeable in the currency 1, and we will note those ๐ด2 ๐ก
17
18. Luc_Faucheux_2021
FX notations - XI
ยจ As you might have guessed, once we have a process for the quantities ๐ด@ ๐ก and ๐@ ๐ก, ๐ก!, ๐ก#
in the (currency 2 world), we will crank the ITO Leibniz handle quite a lot in order to come
up with something useful to say for the processes:
ยจ [๐2โ@. ๐ด@ ๐ก ]
ยจ And:
ยจ [๐2โ@. ๐@ ๐ก, ๐ก!, ๐ก# ]
ยจ And yes, you guessed it right once again, the ever important deflated quantities:
ยจ
[3(โ).C) ?,?&,?$ ]
E( ?
and
[3(โ).F) ? ]
E( ?
ยจ As you can also guess, we are going to say that some of those quantities are going to be
martingale (driftless process) under the appropriate measure, and then crank back up the
Ito Leibniz handle back up to the original processes for ๐ด@ ๐ก and ๐@ ๐ก, ๐ก!, ๐ก#
ยจ But first, some cool geometric tricks on correlation and volatilities
18
20. Luc_Faucheux_2021
The FX correlation triangle
ยจ Essentially it is:
ยจ ๐!,# = ๐!,4 โ ๐4,#
ยจ ๐2,G = ๐2,@ โ ๐@,G
ยจ ๐2,@ = ๐2,G/๐@,G
ยจ Suppose that I know something about the stochastic process for ๐2,G and ๐@,G
ยจ Is there something useful that I can say about the stochastic process for ๐2,@ ?
ยจ In particular about the variance of that process (leaving out the complicated issue of the
drift aside)
20
21. Luc_Faucheux_2021
The FX correlation triangle - II
ยจ First some results on ๐ and (
2
3
)
ยจ Letโs pick some ๐ = ๐!,# and
2
3
=
2
3&,$
= ๐#,!
ยจ The usual market convention for FX is to work with GBM (geometric Brownian Motion).
ยจ Again there is an assumption that Spot FX cannot become negative
ยจ This assumption sounds a little more grounded that the one in rates (which turned out to be
wrong, you can have negative rates). It is a little harder to imagine a negative spot FX
ยจ So usually the process is the following:
ยจ
H3&,$
3&,$
= ๐ ๐!,# . ๐๐ก + ๐ ๐!,# . [ . ๐๐3&,$(๐ก)
ยจ
H3
3
= ๐ ๐ . ๐๐ก + ๐ ๐ . [ . ๐๐3(๐ก)
21
22. Luc_Faucheux_2021
The FX correlation triangle - III
ยจ
H3&,$
3&,$
= ๐ ๐!,# . ๐๐ก + ๐ ๐!,# . [ . ๐๐3&,$(๐ก)
ยจ
H3&โ$
3&โ$
= ๐ ๐!โ# . ๐๐ก + ๐ ๐!โ# . [ . ๐๐3&โ$(๐ก)
ยจ
H3&,$
3&,$
= ๐ ๐!,# . ๐๐ก + ๐ ๐!,# . [ . ๐๐3&,$(๐ก)
ยจ
H3$,&
3$,&
= ๐ ๐#,! . ๐๐ก + ๐ ๐#,! . [ . ๐๐3$,&(๐ก)
ยจ In that case we can use the very useful relation that is :
ยจ ๐ ๐!,# = ๐ ๐#,!
ยจ ๐ ๐!,# = ๐ 1/๐!,# = ๐
2
3&,$
22
24. Luc_Faucheux_2021
The FX correlation triangle - IV
ยจ We can use our good old friend the ITO lemma on:
ยจ ๐ ๐ = 1/๐ Stochastic Variable
ยจ ๐ ๐ฅ = 1/๐ฅ Regular โNewtonianโ variable with well defined partial derivatives
ยจ ๐ฟ๐ =
IJ
IK
. ๐ฟ๐ +
2
@
.
I)J
IK) . (๐ฟ๐)@
ยจ
IJ
IK
=
IJ
IK
|K=3 ?
ยจ
I)J
IK) =
I)J
IK) |K=3 ?
24
29. Luc_Faucheux_2021
The FX correlation triangle - IX
ยจ
H
(
+
(
+
= โ ๐ ๐ โ ๐ ๐ @ . ๐๐ก โ ๐ ๐ . [ . ๐๐3(๐ก)
ยจ
H
(
+
(
+
= ๐
2
3
. ๐๐ก + ๐
2
3
. [ . ๐๐
(
+ (๐ก)
ยจ Note that at this point, all that we can say is that:
ยจ (๐๐3(๐ก))@= ๐๐ก
ยจ (๐๐
(
+ (๐ก) )@= ๐๐ก
ยจ And that is it, but that is enough for now
29
31. Luc_Faucheux_2021
The FX correlation triangle - XI
ยจ [
H
(
+
(
+
]@= [๐
2
3
]@. [
2
(
+
]@= [
2
(
+
]@. [
2
3
. ๐ ๐ . [ . ๐๐3(๐ก)]@= [๐ ๐ ]@. ๐๐ก
ยจ [
H
(
+
(
+
]@= [๐
2
3
]@. [
2
(
+
]@= [
2
(
+
]@. [
2
3
. ๐
2
3
. [ . ๐๐
(
+ (๐ก) ]@= [๐
2
3
]@. ๐๐ก
ยจ [๐ ๐ ]@. ๐๐ก = [๐
2
3
]@. ๐๐ก
ยจ If we also make the assumption, which seems quite reasonable, that we are dealing with
positive volatilities,
ยจ ๐ ๐ = ๐
2
3
31
32. Luc_Faucheux_2021
The FX correlation triangle - XII
ยจ ๐ ๐ = ๐
2
3
ยจ ๐ ๐#,! = ๐
2
3$,&
= ๐ ๐!,#
ยจ ๐ ๐#,! = ๐ ๐!,#
ยจ ๐ ๐#โ! = ๐ ๐!โ#
ยจ So that is pretty cool, and will be useful when we derive the correlation triangle
ยจ Note that this is ONLY true when we define a GBM:
ยจ
H3&,$
3&,$
= ๐ ๐!,# . ๐๐ก + ๐ ๐!,# . [ . ๐๐3&,$(๐ก)
32
33. Luc_Faucheux_2021
The FX correlation triangle - XIII
ยจ As always, if you do not trust me, you can always resort to Excel
ยจ I have a feeling that if Excel, and cheap CPU had come out say maybe 100 years ago, we
would have just reduced everything to a massive GoalSeek, and not spend the time to
derive equations and such.
ยจ Hey, that is kind of what we are doing now with the big data /AI/ML/DL/cloud stuff.
33
34. Luc_Faucheux_2021
The FX correlation triangle - XIV
ยจ Calculating the Expected Value, Variance and all that good stuff for ๐!,# in a discrete binomial
example
ยจ
H3&,$
3&,$
= ๐ ๐!,# . ๐๐ก + ๐ ๐!,# . [ . ๐๐3&,$(๐ก)
ยจ ๐ ๐!,# = ๐ธ ๐!,# =< ๐!,# > = โ ๐๐๐๐๐ ๐ . ๐!,#(๐ ) where ๐ index of the outcomes
ยจ ๐ ๐!,# = ๐ธ (๐!,# โ ๐[๐!,#])@ =< (๐!,# โ < ๐!,# >)@> = โ ๐๐๐๐๐ ๐ . (๐!,#(๐ ) โ ๐[๐!,#])@
ยจ ๐ ๐!,# = (๐ ๐!,# . ๐!,#)@
ยจ ๐ ๐!,# =
2
3&,$
. ๐ ๐!,#
34
38. Luc_Faucheux_2021
The FX correlation triangle - XVII
ยจ Ok, so we have not even started talking about correlationsโฆ
ยจ So letโs get to it, and we will stick to 1,2,3 instead of ๐, ๐, ๐ for ease of notation
ยจ ๐2,@ = ๐2,G โ ๐G,@
ยจ We use our good old friend the ITO Leibniz
ยจ ๐ฟ๐ ๐, ๐ =
IJ
I3
. ๐ฟ๐ +
IJ
IM
. ๐ฟ๐ +
2
@
.
I)J
I3) . ๐ฟ๐@ +
2
@
.
I)J
IM) . ๐ฟ๐@ +
I)J
I3IM
. ๐ฟ๐. ๐ฟ๐
ยจ ๐ ๐, ๐ is really noted ๐ ๐ฅ = ๐(๐ก), ๐ฆ = ๐(๐ก) and all the partial derivatives are for example:
ยจ
I)J
I3IM
=
I)J
IKIN
|K=3 ? ,N=M(?)
ยจ Where ๐ ๐ฅ, ๐ฆ is a nice function operating in the usual Newtonian calculus, where taking
partial derivatives is well defined and makes sense (remember that a stochastic process is
NOT differentiable)
38
41. Luc_Faucheux_2021
The FX correlation triangle - XX
ยจ ๐ ๐2,G โ ๐G,@ = ๐G,@. ๐2,G. ๐ ๐2,G . ๐๐ก + ๐ ๐2,G . [ . ๐๐3(,* ๐ก + ๐2,G. ๐G,@. `
a
๐ ๐G,@ . ๐๐ก +
๐ ๐G,@ . [ . ๐๐3*,) ๐ก + ๐G,@. ๐2,G. ๐ ๐2,G . ๐๐ก + ๐ ๐2,G . [ . ๐๐3(,* ๐ก . `
a
๐ ๐G,@ . ๐๐ก +
๐ ๐G,@ . [ . ๐๐3*,) ๐ก
ยจ This is quite cumbersome
ยจ Luckily for us, we are after computing the variance, so we can neglect all the terms that are not
in first order of the Brownian motion
ยจ ๐ ๐,,- โ ๐-,. = ๐-,.. ๐,,-. ๐ ๐,,- . [ . ๐๐1!,# ๐ก + ๐,,-. ๐-,.. ๐ ๐-,. . [ . ๐๐1#,$ ๐ก + ๐๐๐๐๐กโ๐๐๐
ยจ ๐๐๐๐๐กโ๐๐๐ = ๐๐๐๐๐ก๐๐๐๐ ๐๐. ๐๐ก + ๐ ๐๐๐๐๐กโ๐๐๐ก๐๐๐๐ ๐๐ . ๐๐ก. ๐๐
41
42. Luc_Faucheux_2021
The FX correlation triangle - XXI
ยจ ๐ ๐,,- โ ๐-,. = ๐-,.. ๐,,-. ๐ ๐,,- . [ . ๐๐1!,# ๐ก + ๐,,-. ๐-,.. ๐ ๐-,. . [ . ๐๐1#,$ ๐ก + ๐๐๐๐๐กโ๐๐๐
ยจ We also have:
ยจ (๐๐3&,$(๐ก) )@= ๐๐ก
ยจ ๐๐3&,$ ๐ก . ๐๐32,3 ๐ก = ๐ ๐!,#; ๐Q,R . ๐๐ก
ยจ Of course:
ยจ ๐ ๐!,#; ๐Q,R = ๐ ๐Q,R; ๐!,#
ยจ ๐ ๐!,#; ๐!,# = 1
ยจ Question: is it that obvious that ๐ < 1 ?
ยจ Will leave that for another note at some point later in time
42
44. Luc_Faucheux_2021
The FX correlation triangle - XXIII
ยจ (๐ ๐2,G โ ๐G,@ )@= (๐G,@. ๐2,G)@. {๐ ๐2,G
@
. ๐๐ก + ๐ ๐G,@
@
. ๐๐ก + 2. ๐ ๐2,G; ๐G,@ ๐ ๐2,G . ๐ ๐G,@ . ๐๐ก}
ยจ (๐ ๐2,G โ ๐G,@ )@= (๐๐2,@)@
ยจ (๐ ๐2,G โ ๐G,@ )@= (๐๐2,@)@= (๐2,@)@. ๐ ๐2,@
@
. ๐๐ก
ยจ ๐ ๐2,@
@
= ๐ ๐2,G
@
+ ๐ ๐G,@
@
+ 2. ๐ ๐2,G; ๐G,@ . ๐ ๐2,G . ๐ ๐G,@
ยจ OK, that is a good start, we should be able to start doing a lot with that relation.
44
45. Luc_Faucheux_2021
The FX correlation triangle - XXIV
ยจ ๐ ๐2,@
@
= ๐ ๐2,G
@
+ ๐ ๐G,@
@
+ 2. ๐ ๐2,G; ๐G,@ . ๐ ๐2,G . ๐ ๐G,@
ยจ We also know that we can change the order in the volatility:
ยจ ๐ ๐!,# = ๐ ๐#,!
ยจ The real question is inside the correlation exponent:
ยจ What can we say about for example ๐ ๐2,G; ๐G,@ if we know ๐ ๐2,G; ๐@,G ?
ยจ Letโs derive the relation in another way to illustrate the difference:
45
54. Luc_Faucheux_2021
The FX correlation triangle - XXXII
ยจ ๐ ๐2,@
@
= ๐ ๐2,G
@
+ ๐ ๐G,@
@
โ 2. ๐ ๐2,G; ๐@,G . ๐ ๐2,G . ๐ ๐G,@
ยจ ๐ ๐2,@
@
= ๐ ๐2,G
@
+ ๐ ๐G,@
@
+ 2. ๐ ๐2,G; ๐G,@ . ๐ ๐2,G . ๐ ๐G,@
ยจ So we get:
ยจ ๐ ๐2,G; ๐G,@ = โ๐ ๐2,G; ๐@,G
ยจ ๐ ๐2,G; ๐G,@ = โ๐ ๐2,G;
2
3*,)
ยจ With more general stochastic variables ๐ and ๐
ยจ ๐ ๐; ๐ = โ๐ ๐;
2
3
ยจ Which is general, we did not need the specific derivation above to get to that result, we
could have plugged directly once we got the first equation, but it is worth doing it just to
have the pleasure of doing ITO Leibniz, truly a marvel of stochastic calculus.
54
55. Luc_Faucheux_2021
The FX correlation triangle - XXXIII
ยจ Why go through all that trouble ?
ยจ First of all it is fun
ยจ Second of all it provides constraints on the FX currency pairs volatilities, so if in the market
you observe volatilities that deviate from the equation, you might be thinking about putting
a trade on (careful that this is not a hard arbitrage per se, as most likely you will be putting
an option trade against another option trade)
ยจ ๐ ๐2,@
@
= ๐ ๐2,G
@
+ ๐ ๐G,@
@
โ 2. ๐ ๐2,G; ๐@,G . ๐ ๐2,G . ๐ ๐G,@
ยจ ๐ ๐!,#
@
= ๐ ๐!,4
@
+ ๐ ๐4,#
@
โ 2. ๐ ๐!,#; ๐#,4 . ๐ ๐!,# . ๐ ๐4,#
ยจ ๐ ๐!,# = ๐ ๐#,!
ยจ ๐ ๐!,4; ๐4,# = โ๐ ๐!,4; ๐#,4
55
56. Luc_Faucheux_2021
The FX correlation triangle - XXXIV
ยจ And thirdly, it is usually botched up in a number of textbooks, because they donโt pay
attention to either the notation (Uncle Godel will not be happy), or even worse in some
cases, they derive the relation by using regular calculus and completely sweeping under the
rug the fact that the exchange rates are stochastic, which is ironic, because they are talking
about variance, and usually also right after they go gently into talking about โFX options
modeling).
ยจ I will not name any textbooks/publication, but sometimes you encounter something like the
next couples of slides, which do get indeed the right result at the end, but going completely
willy-nilly about differentiating ratios of stochastic process as if we were in regular
Newtonian calculus and like we never heard about ITO lemma or ITO Leibniz
ยจ Turns out that the end result is correct as we are concerning ourselves with the variance
which means that we are not concerning ourselves with the higher order terms, but still a
little rigor never hurt anyone
56
59. Luc_Faucheux_2021
The FX correlation triangle - XXXVII
ยจ All right Robin, enough equations, letโs draw some nice graphs, and use the power of
analogies.
59
61. Luc_Faucheux_2021
The triangle โ the power of analogies - I
ยจ So we are getting the formula:
ยจ ๐ ๐2,@
@
= ๐ ๐2,G
@
+ ๐ ๐G,@
@
โ 2. ๐ ๐2,G; ๐@,G . ๐ ๐2,G . ๐ ๐G,@
ยจ There is a beautiful way to express this in a geometric manner
ยจ We used to be really good at geometry, when we were building houses and using the stars
to navigate the seas
ยจ Not so much anymore, now I guess we are getting good at coding (and maybe hopefully
stochastic calculus).
61
62. Luc_Faucheux_2021
The triangle โ the power of analogies - II
ยจ ๐ ๐2,@
@
= ๐ ๐2,G
@
+ ๐ ๐G,@
@
โ 2. ๐ ๐2,G; ๐@,G . ๐ ๐2,G . ๐ ๐G,@
ยจ ๐ ๐2,G; ๐@,G =
S 3(,)
)
L S 3(,*
)
L S 3*,)
)
@.S 3(,* .S 3*,)
ยจ Again, not so obvious that ๐ ๐2,G; ๐@,G has to be smaller than 1 in absolute value, but we
will check that in later notes
ยจ Not super obvious.
ยจ So not super obvious that there is a name for it, the Cauchy-Schwartz inequality
ยจ Back to the triangle for now
62
63. Luc_Faucheux_2021
The triangle โ the power of analogies - III
ยจ Letโs look at the following triangle and letโs note 1,2,3 the corners, and the length of the
sides by ๐ฟ(๐, ๐)
ยจ We can easily verify the first rule of triangles: it is impossible to draw a triangle that does not
look like a special case (right, isosceles,..)
63
1
2
3
๐ฟ(1,2)
๐ฟ(1,3)
๐ฟ(3,2)
๐(3)
64. Luc_Faucheux_2021
The triangle โ the power of analogies - IV
ยจ So first of all, we have obviously
ยจ ๐ฟ ๐, ๐ = ๐ฟ(๐, ๐)
ยจ Which kinds of reminds us of:
ยจ ๐ ๐!,# = ๐ ๐#,!
ยจ So we are kind of thinking that maybe if there is an analogy there, the length of the sides
will be the volatilities.
ยจ Letโs try to derive a relation between the length of the sides in the triangle
64
65. Luc_Faucheux_2021
The triangle โ the power of analogies - V
ยจ Letโs compute ๐ฟ 1,2
ยจ Since we are not that great in geometry (and neither have we as a specie for a thousand
years or so), let me use the power of calculus by computing the coordinates of corner 3.
ยจ I choose corner 1 to be the origin
65
1
2
3
๐ฟ(1,2)
๐ฟ(1,3)
๐ฟ(3,2)
๐(3)
๐ฆ
๐ฅ
66. Luc_Faucheux_2021
The triangle โ the power of analogies - VI
ยจ The coordinates are then:
ยจ ๐ถ 1 = (0,0)
ยจ ๐ถ 3 = (๐ฟ(1,3), 0)
ยจ ๐ถ 2 = (๐ฅ@, ๐ฆ@)
ยจ We have :
ยจ ๐ฅ@ = ๐ฅG โ cos ๐ 3 โ ๐ฟ(3,2)
ยจ ๐ฆ@ = ๐ฆG + sin ๐ 3 โ ๐ฟ(3,2)
ยจ With
ยจ ๐ฅG = ๐ฟ 1,3
ยจ ๐ฆG = 0
66
67. Luc_Faucheux_2021
The triangle โ the power of analogies - VII
ยจ ๐ฅ@ = ๐ฟ(1,3) โ cos ๐ 3 โ ๐ฟ(3,2)
ยจ ๐ฆ@ = sin ๐ 3 โ ๐ฟ(3,2)
ยจ And we have:
ยจ ๐ฟ(1,2)@ = ๐ฅ@
@ + ๐ฆ@
@
ยจ ๐ฟ(1,2)@ = (๐ฟ(1,3) โ cos ๐ 3 โ ๐ฟ(3,2))@+(sin ๐ 3 โ ๐ฟ(3,2) )@
ยจ ๐ฟ(1,2).
= ๐ฟ(1,3).
โ 2. ๐ฟ 1,3 . cos ๐ 3 . ๐ฟ 3,2 + ๐ฟ 3,2 .
. cos ๐ 3
.
+ (sin ๐ 3 โ ๐ฟ(3,2) ).
ยจ ๐ฟ(1,2)@ = ๐ฟ(1,3)@ โ 2. ๐ฟ 1,3 . cos ๐ 3 . ๐ฟ 3,2 + ๐ฟ 3,2 @. {cos ๐ 3
@
+ sin ๐ 3
@
}
ยจ ๐ฟ(1,2)@ = ๐ฟ(1,3)@ โ 2. ๐ฟ 1,3 . cos ๐ 3 . ๐ฟ 3,2 + ๐ฟ 3,2 @. {1}
ยจ ๐ฟ(1,2)@ = ๐ฟ(1,3)@ + ๐ฟ 3,2 @ โ 2. ๐ฟ 1,3 . cos ๐ 3 . ๐ฟ 3,2
ยจ Now we getting somewhere
67
68. Luc_Faucheux_2021
The triangle โ the power of analogies - VIII
ยจ ๐ฟ(1,2)@ = ๐ฟ(1,3)@ + ๐ฟ 3,2 @ โ 2. ๐ฟ 1,3 . cos ๐ 3 . ๐ฟ 3,2
ยจ ๐ฟ(1,2)@ = ๐ฟ(1,3)@ + ๐ฟ 3,2 @ โ 2. cos ๐ 3 . ๐ฟ 1,3 . ๐ฟ 3,2
ยจ ๐ ๐2,@
@
= ๐ ๐2,G
@
+ ๐ ๐G,@
@
โ 2. ๐ ๐2,G; ๐@,G . ๐ ๐2,G . ๐ ๐G,@
ยจ So we have the analogy, or the mapping between the two problems:
ยจ ๐ฟ ๐, ๐ โ ๐ ๐!,#
ยจ cos ๐ ๐ โ ๐ ๐!,4; ๐#,4
68
69. Luc_Faucheux_2021
The triangle โ the power of analogies - IX
ยจ cos ๐ 3 = ๐ ๐2,G; ๐@,G
ยจ This is also another strong indication that: โ1 โค ๐ ๐2,G; ๐@,G โค (1)
69
1
2
3
๐ฟ 3,2 = ๐(3,2)
๐(3)
๐ฟ 1,2 = ๐(1,2)
๐ฟ 1,3 = ๐(1,3)
70. Luc_Faucheux_2021
The triangle โ the power of analogies - X
ยจ This is one of the beautiful analogies that one can find in Maths or Physics or science in
general.
ยจ Here is a little random thread that keeps popping into my mind every time I look at the
correlation triangle in FX
ยจ As you know I really like the Godelโs theorem of Incompleteness
ยจ Every couple of years or so, to prove to myself that my brain is still semi-functioning, I tried
to go through the simplified derivation that you could find in a couple of great books (I never
attempted going through the actual derivation from the original Godelโs paper yet, maybe
one day when I get really bored).
ยจ One of those great books is the one by Nagel and Newman:
70
72. Luc_Faucheux_2021
The triangle โ the power of analogies - XII
ยจ In that book, as an example so that we can familiarize ourselves with the formal mapping
that Godel achieved in his famous paper, they use the triangle example (page 15 and 16):
72
73. Luc_Faucheux_2021
The triangle โ the power of analogies - XIII
ยจ Later in the book they revisit the concept of mapping on a triangle and the crucial concept
of duality with the theorem of Pappus:
73
74. Luc_Faucheux_2021
The triangle โ the power of analogies - XIV
ยจ Now, chance would have it that the preface to this beautiful little book is written by no other
than Douglas Hofstadter, who went on to write the very successful and seminal โGEBโ on
Godel, Escher and Bach, centered around the Godelโs theorem, but using analogies with
music and drawings, biology and a lot of other stuff.
74
75. Luc_Faucheux_2021
The triangle โ the power of analogies - XV
ยจ He also wrote a more recent one about the immense power of analogies in thinking:
75
76. Luc_Faucheux_2021
The triangle โ the power of analogies - XVI
ยจ So this is quite funny and circular at the same time.
ยจ A couple of additional notes:
ยจ Another great book (which was partially rewritten during the Covid pandemic) is the one by
Peter Smith:
76
77. Luc_Faucheux_2021
The triangle โ the power of analogies - XVII
ยจ Also in Physics, conformal mapping is a very powerful tool.
ยจ Finally the triangle is quite powerful. As we know now from Dark, the triangle is essential
for time travel:
77
78. Luc_Faucheux_2021
The triangle โ the power of analogies - XVIII
ยจ So that was my little โHeart of darknessโ random stream of consciousness on triangles and
the power of analogies
ยจ Now onto some more ITO Leibniz to look at changing measures when looking at assets
ยจ But before, a blast from the past..
78
80. Luc_Faucheux_2021
A blast from the past
ยจ I was not lying to you when I was telling you that I am doing those decks mostly for myself
ยจ I have all those notes flying around the house on legal pads, and it is time to put them in a
more formal format and in the cloud
ยจ What is sad is that I am struggling now to do in an hour what I could do in like 5 minutes
then. Maybe back then I was not really understanding deeply what I was doing, and now I
am much wiser?
ยจ I doubt it
ยจ There is a study that says that after 30 years old, on average for every year that passes by,
the speed of your tennis serve goes down by 1mph
ยจ I used to clock those at 110mph, not anymore, so I would say that study sounds right
ยจ Maybe it is the same for the brain, and the ability to do ITO Leibniz, every year it takes 1
more minute to do it
ยจ Check out the date on those notes !
80
85. Luc_Faucheux_2021
Measures in the foreign and domestic world
ยจ Essentially it is putting in Powerpoint and many many slides the handwritten notes on
yellow legal pad from more than 25 years ago.
85
86. Luc_Faucheux_2021
Measures in the foreign and domestic world - II
ยจ Letโs assume that currency 1 is the domestic currency
ยจ We will note currency 2 to be the foreign currency
ยจ ๐2,@ = ๐2โ@ is the value in currency (1) of 1 unit of currency (2)
ยจ ๐ต2 ๐ก = exp[โซ
<=>
<=?
๐ 2 ๐ , ๐ , ๐ . ๐๐ ] is the Bank Account Numeraire in the domestic world of
currency (1).
ยจ ๐ต2 ๐ก = exp[โซ
<=>
<=?
๐ 2 ๐ , ๐ , ๐ . ๐๐ ] is associated to the Risk free measure in the domestic
currency (1) that we will note ๐โ2 ๐ก
ยจ ๐ต@ ๐ก = exp[โซ
<=>
<=?
๐ @ ๐ , ๐ , ๐ . ๐๐ ] is the Bank Account Numeraire in the foreign world of
currency (2).
ยจ ๐ต@ ๐ก = exp[โซ
<=>
<=?
๐ @ ๐ , ๐ , ๐ . ๐๐ ] is associated to the Risk free measure in the foreign
currency (2) that we will note ๐โ@ ๐ก
86
87. Luc_Faucheux_2021
Measures in the foreign and domestic world - III
ยจ The Zero Coupon bonds in the domestic world of currency (1) are noted ๐2 ๐ก, ๐ก!, ๐ก#
ยจ The Zero Coupon bonds in the foreign world of currency (2) are noted ๐@ ๐ก, ๐ก!, ๐ก#
ยจ ๐ด2 ๐ก is a tradeable asset in the domestic world of currency (1) (like a stock)
ยจ ๐ด@ ๐ก is a tradeable asset in the foreign world of currency (2) (like a stock)
ยจ We are going to assume the following processes:
ยจ
HF(
F(
= ๐ ๐ด2 . ๐๐ก + ๐ ๐ด2 . [ . ๐๐F((๐ก)
ยจ
HF)
F)
= ๐ ๐ด@ . ๐๐ก + ๐ ๐ด@ . [ . ๐๐F)(๐ก)
ยจ
H3(,)
3(,)
= ๐ ๐2,@ . ๐๐ก + ๐ ๐2,@ . [ . ๐๐3(,)(๐ก)
87
88. Luc_Faucheux_2021
Measures in the foreign and domestic world - IV
ยจ The trick is going to be essentially:
ยจ Create in the domestic world the quantities:
ยจ [๐2โ@. ๐ด@ ๐ก ]
ยจ And:
ยจ [๐2โ@. ๐@ ๐ก, ๐ก!, ๐ก# ]
ยจ And yes, you guessed it right once again, the ever important deflated quantities:
ยจ
[3(โ).C) ?,?&,?$ ]
E( ?
and
[3(โ).F) ? ]
E( ?
ยจ As you can also guess, we are going to say that some of those quantities are going to be
martingale (driftless process) under the appropriate measure, and then crank back up the
Ito Leibniz handle back up to the original processes for ๐ด@ ๐ก and ๐@ ๐ก, ๐ก!, ๐ก#
ยจ Hopefully we can discover something interesting, because it is going to be rather tedious, I
warn you, but here we goโฆ.
88
89. Luc_Faucheux_2021
Measures in the foreign and domestic world - V
ยจ Just a quick note before we start
ยจ ๐ด@ ๐ก is a tradeable asset in the foreign world of currency (2) (like a stock)
ยจ [๐2โ@. ๐ด@ ๐ก ] is a tradeable asset in the domestic world of currency (1)
ยจ That does NOT mean that:
ยจ [๐2โ@. ๐ด@ ๐ก ] = ๐ด2 ๐ก
ยจ Rather obvious but it is sometimes easy to get confused
89
91. Luc_Faucheux_2021
Measures in the foreign and domestic world - VI
ยจ All right here we go, letโs do ITO Leibniz on [๐2โ@. ๐ด@ ๐ก ]
ยจ ๐ฟ๐ ๐, ๐ =
IJ
I3
. ๐ฟ๐ +
IJ
IM
. ๐ฟ๐ +
2
@
.
I)J
I3) . ๐ฟ๐@ +
2
@
.
I)J
IM) . ๐ฟ๐@ +
I)J
I3IM
. ๐ฟ๐. ๐ฟ๐
ยจ ๐ ๐, ๐ is really noted ๐ ๐ฅ = ๐(๐ก), ๐ฆ = ๐(๐ก) and all the partial derivatives are for example:
ยจ
I)J
I3IM
=
I)J
IKIN
|K=3 ? ,N=M(?)
ยจ Where ๐ ๐ฅ, ๐ฆ is a nice function operating in the usual Newtonian calculus, where taking
partial derivatives is well defined and makes sense (remember that a stochastic process is
NOT differentiable)
91
100. Luc_Faucheux_2021
Measures in the foreign and domestic world - XIV
ยจ ๐ ๐2,@. ๐ต@ = ๐ต@. ๐๐2,@ + ๐2,@. ๐๐ต@
ยจ ๐๐ต@ ๐ก = ๐ต@ ๐ก . ๐ @ ๐ก, ๐ก, ๐ก . ๐๐ก
ยจ
H3(,)
3(,)
= ๐ ๐2,@ . ๐๐ก + ๐ ๐2,@ . [ . ๐๐3(,)(๐ก)
ยจ
H(3(,).E))
3(,).E)
= ๐ ๐2,@ + ๐ @ ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐2,@ . [ . ๐๐3(,)(๐ก)
ยจ All right, we like a third of the way there.
ยจ We are now going to look at the usual deflated quantities:
ยจ
[3(โ).E) ? ]
E( ?
= r
[๐2โ@. ๐ต@ ๐ก ] and
[3(โ).F) ? ]
E( ?
= r
[๐2,@. ๐ด@ ๐ก ]
ยจ Again, letโs crank the ITO Leibniz handle on those bad boys
100
109. Luc_Faucheux_2021
Measures in the foreign and domestic world - XXI
ยจ ๐ ๐2,@. ๐ต@, ๐ต2 ๐ก =
3(,).E)
E(
=
3(,).E)
E(
ยจ We can just replace in the previous slides ๐ด@ by ๐ต@
ยจ ๐๐ ๐2,@. ๐ด@, ๐ต2 = ๐
3(,).F)
E(
=
2
E(
. ๐(๐2,@. ๐ด@) + (
L 3(,).F)
E(
) ). ๐๐ต2
ยจ Becomes:
ยจ ๐๐ ๐2,@. ๐ต@, ๐ต2 = ๐
3(,).E)
E(
=
2
E(
. ๐(๐2,@. ๐ต@) + (
L 3(,).E)
E(
) ). ๐๐ต2
ยจ If you do not believe me, you can also redo the ITO Leibniz from scratch
109
114. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXV
ยจ So we are picking another Brownian motion such that the variance of the process is not
affected, only the expected value.
ยจ We are going to define:
ยจ ๐๐
โ
3(,)
๐ก = ๐๐3(,) ๐ก + ๐ ๐2,@ . ๐๐ก
ยจ ๐๐
โ
F)
๐ก = ๐๐F) ๐ก + ๐ ๐ด@ . ๐๐ก
ยจ The reason why we are denoting those with โ like the risk free measure will become soon
apparent, but for now we can just say that we are just picking another Brownian motion
which is defined in relationship to the initial one with an additive drift.
ยจ We will actually solve for that drift so that the deflated processes are martingales under the
new measures.
114
115. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXVI
ยจ
H d
(3(,).E))
d
(3(,).E))
= ๐ ๐2,@ + ๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐2,@ . [ . ๐๐3(,) ๐ก
ยจ ๐๐
โ
3(,)
๐ก = ๐๐3(,) ๐ก + ๐ ๐2,@ . ๐๐ก
ยจ So using the new Brownian motion we get:
ยจ
6 @
(1!,$.=$)
@
(1!,$.=$)
= ๐ ๐,,. + ๐ . ๐ก, ๐ก, ๐ก โ ๐ , ๐ก, ๐ก, ๐ก โ ๐ ๐,,. . ๐ ๐,,. . ๐๐ก + ๐ ๐,,. . [ . ๐๐
โ
1!,$
๐ก
ยจ We would want the risk measure associated to ๐
โ
3(,)
to be the risk free measure, meaning
that tradeable instruments deflated by the Bank Account numeraire are martingale in this
measure, meaning they are driftless, or:
ยจ ๐ ๐2,@ + ๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก โ ๐ ๐2,@ . ๐ ๐2,@ = 0
115
116. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXVII
ยจ ๐ ๐2,@ + ๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก โ ๐ ๐2,@ . ๐ ๐2,@ = 0
ยจ ๐ ๐2,@ =
e 3(,) fg) ?,?,? Lg( ?,?,?
S 3(,)
ยจ ๐ ๐2,@ =
e 3(,) L(g( ?,?,? Lg) ?,?,? )
S 3(,)
ยจ This expression should be familiar, and reminds us of the market price of risk that we looked
at in the single currency case
ยจ The difference here is that the rate ๐ ๐ก, ๐ก, ๐ก is actually the differential of the the domestic
and the foreign rate:
ยจ ๐ ๐ก, ๐ก, ๐ก = ๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก
ยจ ๐ ๐2,@ =
e 3(,) Lg ?,?,?
S 3(,)
116
119. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXX
ยจ So the conditions on the drift for the processes to be martingales in the new measures
associated with the new Brownian motions are:
ยจ ๐ ๐2,@ . ๐ ๐2,@ = ๐ ๐2,@ + ๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก
ยจ ๐ ๐ด@ . ๐ ๐ด@ = ๐ ๐ด@ + ๐ ๐ด@ . ๐ ๐2,@ . ๐ ๐2,@; ๐ด@ โ ๐ @ ๐ก, ๐ก, ๐ก
ยจ The processes with those conditions now read:
ยจ
H d
(3(,).E))
d
(3(,).E))
= ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ
H d
(3(,).F))
d
(3(,).F))
= ๐ ๐2,@ . ๐๐
โ
3(,)
+ ๐ ๐ด@ . ๐๐
โ
F)
ยจ I know that in some other decks I had the โ on top and not at the bottom, but I would have
needed to start with ๐3(,)
๐ก . Next time I rewrite this deck, will try to make it look nicer
and more consistent with the previous decks, sorry for that
119
120. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXI
ยจ
H d
(3(,).E))
d
(3(,).E))
= ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ
H d
(3(,).F))
d
(3(,).F))
= ๐ ๐2,@ . ๐๐
โ
3(,)
+ ๐ ๐ด@ . ๐๐
โ
F)
ยจ All right, now we need to unfold back the deflated processes back to the original processes
that we started with: ๐2,@ and ๐ด@
ยจ Time to crank the ITO Leibniz handle again
120
121. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXII
ยจ
H d
(3(,).E))
d
(3(,).E))
= ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ
3(,).E)
E(
= r
(๐2,@. ๐ต@)
ยจ ๐2,@. ๐ต@ = ๐ต2. r
(๐2,@. ๐ต@)
ยจ ๐๐ต2 ๐ก = ๐ต2 ๐ก . ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก
ยจ We will do Ito Leibniz on ๐ ๐ต2, r
๐2,@. ๐ต@ = ๐ต2. r
(๐2,@. ๐ต@)
121
122. Luc_Faucheux_2021
Going back up one level from the dream
ITO Leibniz on:
๐ต!. ,
๐!,#. ๐ต# = ๐ต!.
%!,$.,$
,!
= ๐!,#. ๐ต#
122
123. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXIII
ยจ ๐ ๐ต2, r
๐2,@. ๐ต@ = ๐ต2. r
(๐2,@. ๐ต@)
ยจ ๐๐ ๐ต(, @
๐(,). ๐ต) =
*+
* ,
-!,#..#
. ๐ @
๐(,). ๐ต) +
*+
*.!
. ๐๐ต( +
(
)
.
*#+
* ,
-!,#..#
# . ๐ @
๐(,). ๐ต)
)
+
(
)
.
*#+
*.!
# . ๐๐ต(
)
+
*#+
* ,
-!,#..# *.!
. ๐ @
๐(,). ๐ต) . ๐๐ต(
ยจ
IJ
I d
3(,).E)
= ๐ต2
ยจ
I)J
I d
3(,).E)
) = 0
ยจ
IJ
IE(
= r
(๐2,@. ๐ต@)
ยจ
I)J
IE(
) = 0
ยจ
I)J
I d
3(,).E) IE(
= 1
123
124. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXIV
ยจ ๐๐ ๐ต!, d
๐!,#. ๐ต# =
$%
$ -
&!,#.)#
. ๐ d
๐!,#. ๐ต# +
$%
$)!
. ๐๐ต! +
!
#
.
$#%
$ -
&!,#.)#
# . ๐ d
๐!,#. ๐ต#
#
+
!
#
.
$#%
$)!
# . ๐๐ต!
#
+
$#%
$ -
&!,#.)# $)!
. ๐ d
๐!,#. ๐ต# . ๐๐ต!
ยจ ๐๐ ๐ต!, d
๐!,#. ๐ต# = ๐ต!. ๐ d
๐!,#. ๐ต# + d
(๐!,#. ๐ต#). ๐๐ต! +
!
#
. 0. ๐ d
๐!,#. ๐ต#
#
+
!
#
. 0. ๐๐ต!
#
+ 1. ๐ d
๐!,#. ๐ต# . ๐๐ต!
ยจ ๐๐ ๐ต2, r
๐2,@. ๐ต@ = ๐ต2. ๐ r
๐2,@. ๐ต@ + r
(๐2,@. ๐ต@). ๐๐ต2 + 1. ๐ r
๐2,@. ๐ต@ . ๐๐ต2
ยจ And we have:
ยจ ๐๐ต2 ๐ก = ๐ต2 ๐ก . ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก
ยจ
H d
(3(,).E))
d
(3(,).E))
= ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ So we have ๐ r
๐2,@. ๐ต@ . ๐๐ต2 = 0
ยจ ๐๐ ๐ต2, r
๐2,@. ๐ต@ = ๐ต2. ๐ r
๐2,@. ๐ต@ + r
(๐2,@. ๐ต@). ๐๐ต2
124
125. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXV
ยจ ๐๐ ๐ต2, r
๐2,@. ๐ต@ = ๐ ๐ต2. r
๐2,@. ๐ต@ = ๐ต2. ๐ r
๐2,@. ๐ต@ + r
(๐2,@. ๐ต@). ๐๐ต2
ยจ ๐๐ต2 ๐ก = ๐ต2 ๐ก . ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก
ยจ
H d
(3(,).E))
d
(3(,).E))
= ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ ๐ ๐ต2. r
๐2,@. ๐ต@ = ๐ต2. r
๐2,@. ๐ต@ . ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก + r
(๐2,@. ๐ต@). ๐ต2 ๐ก . ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก
ยจ And since:
ยจ
3(,).E)
E(
= r
(๐2,@. ๐ต@)
ยจ ๐ต2. r
๐2,@. ๐ต@ = ๐ต2.
3(,).E)
E(
= ๐2,@. ๐ต@
ยจ ๐ ๐2,@. ๐ต@ = ๐2,@. ๐ต@. ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก + ๐2,@. ๐ต@. ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก
125
126. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXVI
ยจ
H d
(3(,).E))
d
(3(,).E))
= ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ
H(3(,).E))
(3(,).E))
= ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ We should be accustomed to this by now, going between the deflated process and the
process in the risk free measure amounts to just a drift ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก
ยจ Letโs now go one step further unfolding our processes:
ยจ
H(3(,).E))
(3(,).E))
= ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ ๐๐ต@ ๐ก = ๐ต@ ๐ก . ๐ @ ๐ก, ๐ก, ๐ก . ๐๐ก
ยจ We will do Ito Leibniz on: ๐ ๐ต@, (๐2,@. ๐ต@) =
2
E)
. (๐2,@. ๐ต@)
126
131. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXX
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ Letโs stop here for a moment
ยจ We started with:
ยจ
H3(,)
3(,)
= ๐ ๐2,@ . ๐๐ก + ๐ ๐2,@ . [ . ๐๐3(,)(๐ก)
ยจ Through change of measure and martingale / driftless process, we showed that using the
Brownian motion associated with the risk free measure in the domestic world:
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ This seems like an awful lot of slides and math to recover something that actually makes a
lot of sense
131
132. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXXI
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ A lot of time, it is super helpful to check our results against our intuition, especially using the
deterministic case of zero volatility
ยจ ๐ ๐2,@ = 0
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก
ยจ This is actually what I teach in the undergraduate class on Money, Banking and Financial
Markets
ยจ Oh also just learnt that I was let go of Natixis this morning
ยจ So if any of you know of any interesting job around NYC, Chicago, Miami or San Diego, drop
me a line .
132
133. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXXII
ยจ In any case, here is the textbook that we use in class. It is really good!
133
134. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXXIII
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก
ยจ This actually makes sense using the concept of Purchasing Power Parity (p.246 in Cecchetti),
also saying that the real exchange rate should be equal to 1
ยจ If ๐2 is the price in currency (1) of a basket of goods (coffee, burger, computer, basis goods
that are somewhat transportable or comparable), and ๐@ the price in currency (2) of that
same (or very similar) basket of goods, and if people in both the foreign and domestic world
essentially value equally that basket of goods (which is a big assumption), then the nominal
exchange rate is given by:
ยจ ๐2,@ = ๐2โ@ =
m(
m)
ยจ ๐2 = ๐2โ@. ๐@
134
135. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXXIV
ยจ ๐2 = ๐2โ@. ๐@
ยจ That also makes sense when I use the โplane travelโ example when I teach that class.
ยจ You live in the foreign country with foreign currency (2)
ยจ You have a basket of goods ๐@
ยจ You sell it to get an amount of currency (2)
ยจ You come to the domestic country with currency (1) (either walking or flying or swimming,
letโs assume that there is no cost to travel, no restrictions due to pandemic or political
differences between the countries,โฆ), with your handful bills or coins of currency (2), go to
exchange them for another handful of bills or coins of currency (1), and then proceed to buy
an equivalent basket of goods ๐2 so that you can settle and live in that new country of yours.
ยจ If things are sort of the same, then ๐2 = ๐2โ@. ๐@
ยจ Note that before I made sure to note that ๐ด2 โ ๐2โ@. ๐ด@ and ๐ต2 โ ๐2โ@. ๐ต@
135
136. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXXV
ยจ ๐2 = ๐2โ@. ๐@
ยจ ๐2,@ = ๐2โ@ =
m(
m)
ยจ In the classroom, we then usually equate inflation to the return of the basket of goods (to
some approximation), and say that over the long run, the return on the nominal exchange
rate is equal to the return on the basket of goods ๐2 minus the return on the basket of
goods ๐@
ยจ In the deterministic world, we are on the safe and firm ground of Newtonian calculus
ยจ ๐2,@ = ๐2โ@ =
m(
m)
ยจ ๐๐2,@ =
2
m)
. ๐๐2 โ
m(
m)
) . ๐๐@
136
137. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXXVI
ยจ ๐2,@ = ๐2โ@ =
m(
m)
ยจ ๐๐2,@ =
2
m)
. ๐๐2 โ
m(
m)
) . ๐๐@
ยจ If we assume that the inflation in each country is noted ๐ 2 and ๐ @, the inflation is then the
rate of return of each basket of goods (again, that makes sense)
ยจ ๐๐2 = ๐ 2. ๐2. ๐๐ก
ยจ ๐๐@ = ๐ @. ๐@. ๐๐ก
ยจ ๐๐2,@ =
2
m)
. ๐๐2 โ
m(
m)
) . ๐๐@
ยจ ๐๐2,@ =
2
m)
. ๐ 2. ๐2. ๐๐ก โ
m(
m)
) . ๐ @. ๐@. ๐๐ก =
m(
m)
. ๐ 2 โ ๐ @ . ๐๐ก = ๐2,@. ๐ 2 โ ๐ @ . ๐๐ก
137
138. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXXVII
ยจ ๐๐2,@ = ๐2,@. ๐ 2 โ ๐ @ . ๐๐ก
ยจ
H3(,)
3(,)
= (๐ 2โ๐ @). ๐๐ก
ยจ To be compared to what we had obtained after a couple thousand slides of stochastic
calculus:
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก
138
139. Luc_Faucheux_2021
Measures in the foreign and domestic world โ XXXXVIII
ยจ This is what Cecchetti and Schoenholtz explains on p.246, making sur to point out that such
arguments tend to be true over the long run, and that short term behaviors of rates and FX
are much more random and sometimes violent in nature, and driven by supply and demand.
139
140. Luc_Faucheux_2021
Measures in the foreign and domestic world - IL
ยจ Ahhh !! I got you !
ยจ The Roman numeral for 49 is NOT IL, because I can only be subtracted from V and X
ยจ Wanted to see if you guys were following.
ยจ Do not trust me ? To the Google Robin !
ยจ But if you follow superbowl and such, you know all about Roman numerals
140
141. Luc_Faucheux_2021
Measures in the foreign and domestic world - L
ยจ OK, so now here is an amazing graph (although to be fully transparent I think that they
should have plotted it on a logarithmic scale, not doing it is a tad ethnocentric if you ask
me).
ยจ But in any case, they looks at returns over 1980 to 2010 on the USD exchange rate for a
number of countries (that is easy to do)
ยจ Then they looked at the difference in annualized inflation rate between that country and the
US (that is a little harder to do, because how you measure inflation is not super easy, is that
CPI, is that PPI, does the basket change over time, like computer percentage in the basket, or
rentals, or gas,..). But anyways they did their homework and produced a graph that is quite
impressive !
ยจ Over the long run, and with some common sense assumptions, we do have indeed:
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
141
142. Luc_Faucheux_2021
Measures in the foreign and domestic world - LI
142
๐๐!,#
๐!,#
= {๐ ! ๐ก, ๐ก, ๐ก โ ๐ # ๐ก, ๐ก, ๐ก }. ๐๐ก
IT WORKS !!!!
143. Luc_Faucheux_2021
Measures in the foreign and domestic world - LII
ยจ When the real exchange rate deviates from 1, then things get a little weirder. Letโs see if you
could answer on of the questions I ask the class every week:
143
145. Luc_Faucheux_2021
Going from X to (1/X)
ยจ Before we finish the deck, letโs go over some quick notes on going from X to (1/X)
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก
ยจ Now of course by symmetry we have:
ยจ
H3),(
3),(
= {๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐@,2 . [ . ๐๐
โ
3),(
๐ก
ยจ We already know that:
ยจ ๐@,2 =
2
3(,)
ยจ And that:
ยจ ๐ ๐2,@ = ๐ ๐@,2
145
146. Luc_Faucheux_2021
Going from X to (1/X) - II
ยจ Letโs use good old friend ITO lemma on ๐ ๐2,@ = (
2
3(,)
)
ยจ ๐ฟ๐ =
IJ
IK
. ๐ฟ๐ +
2
@
.
I)J
IK) . (๐ฟ๐)@
ยจ
IJ
I3(,)
=
L2
3(,)
)
ยจ
I)J
I3(,)
) =
@
3(,)
*
ยจ ๐๐@,2 = ๐
2
3(,)
= ๐๐(๐2,@) =
IJ
I3(,)
. ๐๐2,@ +
2
@
.
I)J
I3(,)
) . (๐๐2,@)@
ยจ Note that again, we really are dealing with a regular function ๐ ๐ฅ = 1/๐ฅ
ยจ Which is nicely differentiable and where the rules of regular (Newtonian) calculus do apply
146
147. Luc_Faucheux_2021
Going from X to (1/X) - III
ยจ So really we should be writing to be rigorous:
ยจ ๐๐(๐) =
IJ
IK
|K=3 ? . ๐๐ +
2
@
.
I)J
IK) |K=3 ? . (๐๐)@
ยจ Which is usually abbreviated for sake of simplicity to:
ยจ ๐๐(๐) =
IJ
I3
. ๐๐ +
2
@
.
I)J
I3) . (๐๐)@
ยจ But bear in mind that this is the whole point of stochastic processes, is that they are not
differentiable.
ยจ So writing something like
IJ
I3
is fraught with peril
147
151. Luc_Faucheux_2021
Going from X to (1/X) - VII
ยจ
H3),(
3),(
= ๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก + ๐ ๐2,@
@
. ๐๐ก โ ๐ ๐@,2 . [ . ๐๐
โ
3(,)
๐ก
ยจ
H3),(
3),(
= {๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐@,2 . [ . ๐๐
โ
3),(
๐ก
ยจ So we have:
ยจ ๐๐
โ
3),(
๐ก = โ๐๐
โ
3(,)
๐ก โ ๐ ๐@,2 . ๐๐ก
ยจ This if you want explain the Siegel paradox for 2 currencies.
ยจ We saw in the first part of the deck (even using Excel) that ๐ ๐@,2 = ๐ ๐2,@ but at the
time we did not look in details at the drift, and in particular the difference in drift between
an exchange rate and its inverse
151
153. Luc_Faucheux_2021
Going back up one level from the dream
ITO Leibniz on:
๐ต!. ,
๐!,#. ๐ด# = ๐ต!.
%!,$.)$
,!
= ๐!,#. ๐ด#
153
154. Luc_Faucheux_2021
The arrival of correlation in the drift
ยจ All right, we almost there, we have one more couple of ITO Leibniz to do to go back from the
deflated to the original process
ยจ
H d
(3(,).F))
d
(3(,).F))
= ๐ ๐2,@ . ๐๐
โ
3(,)
+ ๐ ๐ด@ . ๐๐
โ
F)
ยจ r
(๐2,@. ๐ด@) =
3(,).F)
E(
ยจ ๐๐ต2 ๐ก = ๐ต2 ๐ก . ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก
ยจ By now we should be super familiar with going from deflated to reflated, you can do ITO
Leibniz or just trust me by now
ยจ
H d
(3(,).F))
d
(3(,).F))
= ๐ ๐2,@ . ๐๐
โ
3(,)
+ ๐ ๐ด@ . ๐๐
โ
F)
ยจ
H(3(,).F))
(3(,).F))
= ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐2,@ . ๐๐
โ
3(,)
+ ๐ ๐ด@ . ๐๐
โ
F)
154
155. Luc_Faucheux_2021
The arrival of correlation in the drift - II
ยจ
H(3(,).F))
(3(,).F))
= ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐2,@ . ๐๐
โ
3(,)
+ ๐ ๐ด@ . ๐๐
โ
F)
ยจ ๐ด@ =
3(,).F)
3(,)
ยจ All right so by now we are quite familiar with the ITO Leibniz handle cranking
ยจ It is a little different than from we unfolded ๐2,@ because then we were dealing with:
ยจ ๐ ๐ต@, (๐2,@. ๐ต@) =
2
E)
. (๐2,@. ๐ต@)
ยจ Here we are going to be dealing with:
ยจ ๐ ๐2,@, (๐2,@. ๐ด@) =
2
3(,)
. (๐2,@. ๐ด@)
ยจ So instead of just an ๐ @ ๐ก, ๐ก, ๐ก term popping in the drift, we will have some non zero crosses
between the driver of ๐2,@ and the driver of (๐2,@. ๐ด@). All right, letโs do it, the finish line is
almost there
155
162. Luc_Faucheux_2021
The arrival of correlation in the drift - VIII
ยจ
HF)
F)
= ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐2,@ . ๐๐
โ
3(,)
+ ๐ ๐ด@ . ๐๐
โ
F)
โ v
w
๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก . ๐๐ก +
๐ ๐2,@ . [ . ๐๐
โ
3(,)
๐ก + ๐ ๐2,@
@
. ๐๐ก โ `
a
๐ ๐2,@ . ๐ ๐2,@ . ๐๐ก +
๐ ๐2,@ . ๐ ๐ด@ . ๐โ ๐2,@, ๐ด@ . ๐๐ก }
ยจ There are now a number of terms that โmagicallyโ disappear (it is not magic, it is the fact
that we are unfolding back the ITO Leibniz
ยจ That reminds me of the turtle in GEB, will put that in the next slide, hopefully we did not
mess upby being off one level going back down and back up in the different levels of the
dreamโฆ
162
163. Luc_Faucheux_2021
The arrival of correlation in the drift โ VIII-a
ยจ Every time I go down and back up those ITO Leibniz derivations, I am thinking about the
tortoise and Achilles going back down and up in the book Godel Escher Bachโฆand am always
scared that I missed one level, or that I am not back at the right levelโฆ
163
164. Luc_Faucheux_2021
The arrival of correlation in the drift โ VIII-b
ยจ Or it is also like that Chris Nolan movie, where I am going back up from the levels of the
dream, but not quite sure where I end up, and I might have gotten lost on the
wayโฆterrifying stuff..
164
166. Luc_Faucheux_2021
The arrival of correlation in the drift - X
ยจ
HF)
F)
= ๐ @ ๐ก, ๐ก, ๐ก โ ๐ ๐2,@ . ๐ ๐ด@ . ๐โ ๐2,@, ๐ด@ . ๐๐ก + ๐ ๐ด@ . ๐๐
โ
F)
ยจ We are all the way back in the first level of the dream (or are we?)
ยจ We started with:
ยจ
HF)
F)
= ๐ ๐ด@ . ๐๐ก + ๐ ๐ด@ . [ . ๐๐F)(๐ก)
ยจ We ended up with:
ยจ
HF)
F)
= ๐ @ ๐ก, ๐ก, ๐ก โ ๐ ๐2,@ . ๐ ๐ด@ . ๐โ ๐2,@, ๐ด@ . ๐๐ก + ๐ ๐ด@ . ๐๐
โ
F)
ยจ So the first term in ๐ @ ๐ก, ๐ก, ๐ก should not surprise us, when we go into the risk free measure,
the drift from the physical measure ๐ ๐ด@ is replaced by the drift ๐ @ ๐ก, ๐ก, ๐ก from the
Numeraire associated to the risk free measure, meaning the bank account
ยจ The second term in the drift โ๐ ๐2,@ . ๐ ๐ด@ . ๐โ ๐2,@, ๐ด@ is new to us
166
167. Luc_Faucheux_2021
The arrival of correlation in the drift - XI
ยจ As usual we note that the variance of the process is not affected by the change of measure
that we performed.
ยจ Changing the measure changes the drift, it does not change the variance
ยจ Let us ponder a little that new adjustment to the drift that we have encountered
ยจ
HF)
F)
= ๐ @ ๐ก, ๐ก, ๐ก โ ๐ ๐2,@ . ๐ ๐ด@ . ๐โ ๐2,@, ๐ด@ . ๐๐ก + ๐ ๐ด@ . ๐๐
โ
F)
ยจ Note that if we were only in the domestic world of the currency (2), and if we had done the
usual change of measure from the physical to the risk free for an asset ๐ด@, we would have
obtained:
ยจ
HF)
F)
= ๐ @ ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐ด@ . ๐๐
โ
F)
ยจ So what is wrong ?
167
169. Luc_Faucheux_2021
The arrival of correlation in the drift - XII
ยจ There is nothing wrong, we just need to be a little careful with the notation
ยจ Remember, all we did up to here was to bring the foreign asset (currency 2) back into the
domestic world (currency 1), deflate it by the Bank Account:
ยจ ๐ต2 ๐ก = exp[โซ
<=>
<=?
๐ 2 ๐ , ๐ , ๐ . ๐๐ ]
ยจ Express it as martingale under a new Brownian motion associated with that asset โIN THE
DOMESTIC WORLD 1โ
ยจ So really (we did not do it for sake of notation, but now is the time to do it), when we were
writing โ it should really have been โ2
169
170. Luc_Faucheux_2021
The arrival of correlation in the drift - XIII
ยจ To be rigorous on the Exchange Rate:
ยจ
H3(,)
3(,)
= {๐ 2 ๐ก, ๐ก, ๐ก โ ๐ @ ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐2,@ . [ . ๐๐
โ(
3(,)
๐ก
ยจ Now of course by symmetry we have:
ยจ
H3),(
3),(
= {๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐@,2 . [ . ๐๐
โ)
3),(
๐ก
ยจ And doing ITO lemma on ๐ ๐ฅ = (
2
K
) in the first equation led to :
ยจ
H3),(
3),(
= ๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก + ๐ ๐2,@
@
. ๐๐ก โ ๐ ๐@,2 . [ . ๐๐
โ(
3(,)
๐ก
ยจ
H3),(
3),(
= {๐ @ ๐ก, ๐ก, ๐ก โ ๐ 2 ๐ก, ๐ก, ๐ก }. ๐๐ก + ๐ ๐@,2 . [ . ๐๐
โ)
3),(
๐ก
170
171. Luc_Faucheux_2021
The arrival of correlation in the drift - XIV
ยจ So we have:
ยจ ๐๐
โ)
3),(
๐ก = โ๐๐
โ(
3(,)
๐ก โ ๐ ๐@,2 . ๐๐ก
ยจ ๐
โ)
3),(
๐ก = โ๐
โ(
3(,)
๐ก โ ๐ ๐@,2 . ๐ก
171
172. Luc_Faucheux_2021
The arrival of correlation in the drift - XV
ยจ Similar for the asset:
ยจ
HF)
F)
= ๐ @ ๐ก, ๐ก, ๐ก โ ๐ ๐2,@ . ๐ ๐ด@ . ๐โ ๐2,@, ๐ด@ . ๐๐ก + ๐ ๐ด@ . ๐๐
โ
F)
ยจ IS really:
ยจ
HF)
F)
= ๐ @ ๐ก, ๐ก, ๐ก โ ๐ ๐2,@ . ๐ ๐ด@ . ๐โ(
๐2,@, ๐ด@ . ๐๐ก + ๐ ๐ด@ . ๐๐
โ(
F)
ยจ By symmetry of course we will have:
ยจ
HF(
F(
= ๐ 2 ๐ก, ๐ก, ๐ก โ ๐ ๐@,2 . ๐ ๐ด2 . ๐โ)
๐@,2, ๐ด2 . ๐๐ก + ๐ ๐ด2 . ๐๐
โ)
F(
ยจ And if we were to only stick to the domestic world:
ยจ
HF(
F(
= ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐ด2 . ๐๐
โ(
F(
172
173. Luc_Faucheux_2021
The arrival of correlation in the drift โ XV-a
ยจ
HF(
F(
= ๐ 2 ๐ก, ๐ก, ๐ก . ๐๐ก + ๐ ๐ด2 . ๐๐
โ(
F(
ยจ
Hn
F(
n
F(
= ๐ ๐ด2 . ๐๐
โ(
F(
ยจ x
๐ด2 =
F(
E(
173
177. Luc_Faucheux_2021
Things to still do in FX
ยจ Expand on the quanto adjustment
ยจ Redo the quanto adjustment using the Radon Nikodym derivative
ยจ Draw more figures and examples on the correlation triangle
ยจ Explain how FX options are traded in practice
ยจ Some more slides on the correlations
ยจ Build some examples rom the quanto drift
ยจ Link quanto to bi-curve valuations in the swap world
177