SlideShare a Scribd company logo
1 of 28
Download to read offline
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Quantile estimation
and optimal portfolios
Arthur Charpentier & Abder Oulidi
ENSAI-ENSAE-CREST & IMA Angers
Journée SFdS, Mai 2007
1
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Portfolio management and optimal allocations
Idea: allocating capital among a set of assets to maximize return and minimize
risk.
If diversification effects were intuited early, and Markowitz (1952) proposed a
mathematical model.
• return is measured by the expected value of the portfolio return,
• risk is quantified by the variance of this return.
Agenda
1. statistical issue in the mean-variance framework
2. portfolio optimization with general risk measures
2
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Portfolio optimization (parametric framework)
Consider a risk measure R (variance or Value-at-Risk). Solve
ω∗
=



argmin{R(ωt
X)},
u.c. ωt
1 = 1 and E(ωt
X) ≥ η,
where X ∼ L(θ), θ unknown.
θ is unknown but can be estimated using a sample {X1, . . . , Xn}.
“The parameters governing the central tendency and dispersion of returns are
usually not known, however; and are often estimated or guessed at using
observed returns and other available data. In empirical applications, the
estimated parameters are used as if they were the true value” (Coles
& Loewenstein (1988)).
If ω∗
= ψ(θ) (e.g. mean-variance)
ω∗
= ψ(θ).
3
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Optimization of standard deviation (or variance)
Allocation
in
the
first asset
Allocation in the second asset
Standard deviation of the portfolio
−200 −100 0 100 200
−200−1000100200 Allocation in the first asset
Allocationinthesecondasset
Figure 1: Portfolio variance optimization problem.
4
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Portfolio optimization (parametric framework)
In the case of no explicit expression of the optimum, solve (numerically)
ω∗
=



argminR(ωt
X),
u.c.ω ∈ {(ωk)k∈{1,...,m}}
where X ∼ L(θ).
The idea is to generate samples Xi’s,
• either from a parametric distribution L(θ),
• or from a nonparametric distribution (bootstrap approach).
5
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Optimization of Value-at-Risk
VaR of the portfolio
−4 −2 0 2 4 6
−2−1012345
−4
−2
0
2
4
6
VaR of the portfolio
−4 −2 0 2 4 6
012345
−4
−2
0
2
4
6
Figure 2: Optimization in the mean-VaR framework.
6
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Classical mean-variance allocation problem
Consider d risky assets, with weekly returns X = (X1, . . . , Xd). Denote
µ = E(X) and Σ = var(X).
Let ω = (ω1, . . . , ωd) ∈ Rd
denote the weights in all risky assets.
• the expected return of the portfolio is E(ωt
X) = ωt
µ,
• the variance of the portfolio is var(ωt
X) = ωt
Σω.



ω∗
∈ argmin{ωt
Σω}
u.c. ωt
µ ≥ η and ωt
1 = 1
convex
⇐⇒



ω∗
∈ argmax{ωt
µ}
u.c. ωt
Σω ≤ η and ωt
1 = 1
7
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Classical mean-variance allocation problem
The solution can be given explicitly (see Markowitz (1952)) as
ω∗
= ψ(µ, Σ) = p + ηq
where µ = E(X), Σ = var(X),
p =
1
d
bΣ−1
1 − aΣ−1
µ and q =
1
d
cΣ−1
µ − aΣ−1
1 ,
and a = 1t
Σ−1
µ, b = µt
Σ−1
µ, c = 1t
Σ−1
1, d = bc − a2
.
Note that p is an allocation, and q indicates how the original portfolio should
be modified.
8
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Efficient frontier
first asset
−0.2 0.0 0.2 0.4 −0.4 0.0 0.4
−0.20.00.2
−0.20.00.20.4
second asset
third asset
−0.20.20.6
−0.2 0.0 0.2
−0.40.00.4
−0.2 0.2 0.6
fourth asset
Portfolio with 4assets
0.010 0.015 0.020 0.025 0.030
0.0000.0010.0020.0030.0040.0050.006
Efficient Frontier
Standard deviation
Expectedvalue
Figure 3: Solving a variance optimization problem.
9
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Inference issues
In practice, µ = [µi] and Σ = [Σi,j] are unknown, and should be estimated.
A natural idea is to define
µi =
1
n
n
t=1
Xi,t et Σi,j =
1
n − 1
n
t=1
(Xi,t − µi)(Xj,t − µj).
Given n observed observed returns,
µ|Σ ∼ N µ,
Σ
n
and nΣ|Σ ∼ W (n − 1, Σ) .
where the two random variables µ and Σ are independent, given Σ.
10
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
0.05 0.10 0.15 0.20 0.25
0.000.050.100.150.200.25
Efficient Frontier, with 250 past observations
Standard deviation
Expectedvalue
0.05 0.10 0.15 0.20 0.25
0.000.050.100.150.200.25
Efficient Frontier, with 1000 past observations
Standard deviation
Expectedvalue
Figure 4: Efficient frontiers and estimation.
11
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Parametric bootstrap
Assume that X ∼ L(θ). estimate θ by θn. The procedure is the following
1. generate n returns X1, . . . , Xn from L(θn);
2. estimate µ and Σ, i.e. µn and Σn,
3. solve the minimization problem, i.e.
ω∗
=
1
d
bΣ
−1
1 − aΣ
−1
µ + η
1
d
cΣ
−1
µ − aΣ
−1
1 ,
Using several simulations, the distribution of the ω∗
k and vark(ω∗
k
t
X) can be
obtained.
12
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Nonparametric bootstrap
A nonparametric procedure can also be considered. Consider a n sample
{X1, . . . , Xn}
1. generate a bootstrap sample from {X1, . . . , Xn}
2. estimate µ and Σ, i.e. µn and Σn,
3. solve the minimization problem, i.e.
ω∗
=
1
d
bΣ
−1
1 − aΣ
−1
µ + η
1
d
cΣ
−1
µ − aΣ
−1
1 ,
Using several simulations, the distribution of the ω∗
k and vark(ω∗
k
t
X) can be
obtained.
13
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
0.4 0.6 0.8 1.0
012345
Allocation in the first asset
Allocation weight
Density
0.0 0.2 0.4
012345
Allocation in the second asset
Allocation weight
Density
−0.3 −0.1 0.0 0.1
02468
Allocation in the third asset
Allocation weight
Density
0.05 0.15 0.25
0246810
Allocation in the fourth asset
Allocation weightDensity
Figure 5: Distributions of optimal allocations ω∗
k’s.
14
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinsecondasset
Joint distribution of optimal allocations (1−2)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinthirdasset
Joint distribution of optimal allocations (1−3)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinfourthasset
Joint distribution of optimal allocations (1−4)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the second asset
Allocationinfourthasset
Joint distribution of optimal allocations (2−4)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the second asset
Allocationinthirdasset
Joint distribution of optimal allocations (2−3)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the third asset
Allocationinfourthasset
Joint distribution of optimal allocations (3−4)
Figure 6: Joint distributions of optimal allocations ω∗
k’s.
15
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
0.05 0.06 0.07 0.08 0.09 0.10 0.11
020406080100
Density of estimated optimal standard deviation
Optimal standard deviation
Density
Figure 7: Distribution of vark(ω∗
k
t
X)
16
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Value-at-Risk minimization
With V aR(X, p) = F−1
(p) = sup{x, F(x) < p}, the program is



ω∗
∈ argmin{VaR(ωt
X, α)}
u.c. E(ωt
X) ≥ η,ωt
1 = 1
nonconvex



ω∗
∈ argmax{E(ωt
X)}
u.c. {VaR(ωt
X, α)} ≤ η ,ωt
1 = 1
In the previous framework (mean-variance), it could be done easily since
• there are only a few estimates of the variance
• there exists an analytical expression of the optimal allocation,
In the case of Value-at- Risk minimization,
• there are several estimators of quantiles (see Charpentier & Oulidi
(2007)),
• numerical optimization should be considered.
17
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Quantile estimation
• raw estimator of the quantile, X[pn]:n = Xi:n = F−1
n (i/n) such that
i ≤ pn < i + 1.
• weighted average of F−1
n (p), e.g. αXi:n + (1 − α)Xi+1:n,
• weighted average of F−1
n (p), e.g.
n
i=1
αiXi:n =
1
0
αuF−1
n (u)du,
• smoothed version of the cdf, F−1
K (p) where FK(x) =
1
nh
n
i=1
K
x − Xi
h
• semiparametric approach, based on Hill’s estimator, Xn−k:n
n
k
(1 − p)
−ξk
,
where ξk =
1
k
k
i=1
log Xn+1−i:n − log Xn−k:n (if ξ > 0),
• fully parametric approach, Xn + u1−pvar(X) (if X ∼ N(µ, σ2
))
18
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
0.0 0.5 1.0 1.5 2.0
0.00.20.40.60.81.0 Empirical quantile estimation
Value
Probability
0.0 0.5 1.0 1.5 2.0
0.00.20.40.60.81.0
Empirical quantile estimation
Value
Probability
Figure 8: Classical estimation of the quantile, based on F−1
(·).
19
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
0.0 0.5 1.0 1.5 2.0
0.00.51.01.5
Smoothed empirical quantile estimation
Value
Smootheddensity
0.0 0.5 1.0 1.5 2.0
0.00.20.40.60.81.0
Smoothed empirical quantile estimation
ValueProbability
Figure 9: Smoothed estimation of the quantile, based on F−1
K (·).
20
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
A short extention to general risk measures
In a much more general setting, spectral risk measures can be considered, i.e.
R(X) =
1
0
φ(p)F−1
X (p)dp,
for some distortion function φ : [0, 1] → [0, 1].
21
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Parametric bootstrap
Assume that X ∼ L(θ). The procedure is the following
1. generate n returns X1, . . . , Xn from L(θ);
2. estimate for all ω on a finite grid, estimate VaR(ωt
X),
3. solve the minimization problem on the grid to get numerically ω∗
n.
Using several simulations, the distribution of ω∗
n and var(ω∗
nX) can be
obtained.
22
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinsecondasset
Joint distribution of optimal allocations (1−2)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinthirdasset
Joint distribution of optimal allocations (1−3)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinfourthasset
Joint distribution of optimal allocations (1−4)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the second asset
Allocationinfourthasset
Joint distribution of optimal allocations (2−4)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the third asset
Allocationinfourthasset
Joint distribution of optimal allocations (3−4)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the second asset
Allocationinthirdasset
Joint distribution of optimal allocations (2−3)
Figure 10: Joint distributions of optimal allocations ω∗
k’s, smoothed quantile
estimator.
23
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
0.16 0.18 0.20 0.22 0.24
05101520
Density of estimated optimal 99% quantile
Optimal Value−at−Risk
Density
Figure 11: Distribution of VaRk(ω∗
k
t
X, 95%).
24
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinsecondasset
Joint distribution of optimal allocations (1−2)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinthirdasset
Joint distribution of optimal allocations (1−3)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the first asset
Allocationinfourthasset
Joint distribution of optimal allocations (1−4)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the second asset
Allocationinfourthasset
Joint distribution of optimal allocations (2−4)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the third asset
Allocationinfourthasset
Joint distribution of optimal allocations (3−4)
−0.2 0.0 0.2 0.4 0.6 0.8 1.0
−0.20.00.20.40.60.81.0
Allocation in the second asset
Allocationinthirdasset
Joint distribution of optimal allocations (2−3)
Figure 12: Joint distributions of optimal allocations ω∗
k’s, raw quantile estima-
tor.
25
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
0.16 0.18 0.20 0.22 0.24
010203040
Density of estimated optimal 95% quantile
Optimal Value−at−Risk
Density
Figure 13: Distribution of VaRk(ω∗
k
t
X, 95%).
26
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Conclusion
Dealing with only 4 assets, it is difficult to get robust optimal allocation, only
because of statistical uncertainty of classical estimators. Remark: this was
mentioned in Liu (2003) on high frequency data (every 5 minutes, i.e.
n = 10, 000) with 100 assets.
27
Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios
Some references
Coles, J.L. & Loewenstein, U. (1988). Equilibrium pricing and portfolio composition in the presence of uncertain
parameters. Journal of Financial Economics, 22, 279-303.
Dowd, K. & Blake, D.. (2006). After VaR: the theory, estimation, and insurance applications of quantile-based risk
measures. Journal of Risk & Insurance, 73, 193-229.
Duarte, A. (1999). Fast computation of efficient portfolios. Journal of Risk, 1, 71-94.
Duffie, D. & Pan, J. (1997). An overview of Value at Risk. Journal of Derivatives, 4, 7-49.
Gaivoronski, A.A. & Pflug, G. (2000). Value-at-Risk in portfolio optimization: properties and computational
approach. Working Paper 00-2, Norwegian University of Sciences & Technology.
Jorion, P. (1997). Value at Risk: the new benchmark for controlling market risk. McGraw-Hill.
Kast, R., Luciano, E. & Peccati, L. (1998). VaR and optimization: 2nd international workshop on preferences and
decisions. Trento, July 1998.
Klein, R.W. & Bawa, V.S. (1976). The effect of estimation risk on optimal portfolio choice. Journal of Financial
Economics, 3, 215-231.
Larsen, N., Mausser, H. & Uryasev, S. (2002). Algorithms for optimization of Value at Risk. in Financial
engineering, e-commerce and supply-chain, Pardalos and Tsitsiringos eds., Kluwer Academic Publichers, 129-157.
Litterman, R. (1997). Hot spots and edges II. Risk, 10, 38-42.
Rockafellar, R.T. & Uryasev, S. (2000). Optimization of Conditional Value-at-Risk. Journal of Risk, 2, 21-41.
28

More Related Content

What's hot (20)

Slides toulouse
Slides toulouseSlides toulouse
Slides toulouse
 
Slides amsterdam-2013
Slides amsterdam-2013Slides amsterdam-2013
Slides amsterdam-2013
 
Slides astin
Slides astinSlides astin
Slides astin
 
Slides ensae 8
Slides ensae 8Slides ensae 8
Slides ensae 8
 
Slides erasmus
Slides erasmusSlides erasmus
Slides erasmus
 
Slides euria-1
Slides euria-1Slides euria-1
Slides euria-1
 
Slides dauphine
Slides dauphineSlides dauphine
Slides dauphine
 
Slides ACTINFO 2016
Slides ACTINFO 2016Slides ACTINFO 2016
Slides ACTINFO 2016
 
Slides ensae 9
Slides ensae 9Slides ensae 9
Slides ensae 9
 
Slides edf-2
Slides edf-2Slides edf-2
Slides edf-2
 
Slides risk-rennes
Slides risk-rennesSlides risk-rennes
Slides risk-rennes
 
Berlin
BerlinBerlin
Berlin
 
Slides université Laval, Actuariat, Avril 2011
Slides université Laval, Actuariat, Avril 2011Slides université Laval, Actuariat, Avril 2011
Slides université Laval, Actuariat, Avril 2011
 
Slides erm-cea-ia
Slides erm-cea-iaSlides erm-cea-ia
Slides erm-cea-ia
 
Graduate Econometrics Course, part 4, 2017
Graduate Econometrics Course, part 4, 2017Graduate Econometrics Course, part 4, 2017
Graduate Econometrics Course, part 4, 2017
 
Slides ineq-4
Slides ineq-4Slides ineq-4
Slides ineq-4
 
Lundi 16h15-copules-charpentier
Lundi 16h15-copules-charpentierLundi 16h15-copules-charpentier
Lundi 16h15-copules-charpentier
 
Slides barcelona Machine Learning
Slides barcelona Machine LearningSlides barcelona Machine Learning
Slides barcelona Machine Learning
 
Inequality, slides #2
Inequality, slides #2Inequality, slides #2
Inequality, slides #2
 
Slides ensae-2016-8
Slides ensae-2016-8Slides ensae-2016-8
Slides ensae-2016-8
 

Viewers also liked (20)

Slides sales-forecasting-session1-web
Slides sales-forecasting-session1-webSlides sales-forecasting-session1-web
Slides sales-forecasting-session1-web
 
Slides edf-1
Slides edf-1Slides edf-1
Slides edf-1
 
Slides axa
Slides axaSlides axa
Slides axa
 
Slides econometrics-2017-graduate-2
Slides econometrics-2017-graduate-2Slides econometrics-2017-graduate-2
Slides econometrics-2017-graduate-2
 
Econometrics 2017-graduate-3
Econometrics 2017-graduate-3Econometrics 2017-graduate-3
Econometrics 2017-graduate-3
 
Econometrics, PhD Course, #1 Nonlinearities
Econometrics, PhD Course, #1 NonlinearitiesEconometrics, PhD Course, #1 Nonlinearities
Econometrics, PhD Course, #1 Nonlinearities
 
Slides picard-6
Slides picard-6Slides picard-6
Slides picard-6
 
Slides saopaulo-catastrophe (1)
Slides saopaulo-catastrophe (1)Slides saopaulo-catastrophe (1)
Slides saopaulo-catastrophe (1)
 
Cours add-r1-part5
Cours add-r1-part5Cours add-r1-part5
Cours add-r1-part5
 
Lg ph d_slides_vfinal
Lg ph d_slides_vfinalLg ph d_slides_vfinal
Lg ph d_slides_vfinal
 
Soutenance julie viard_partie_1
Soutenance julie viard_partie_1Soutenance julie viard_partie_1
Soutenance julie viard_partie_1
 
HdR
HdRHdR
HdR
 
Slides ensae-2016-2
Slides ensae-2016-2Slides ensae-2016-2
Slides ensae-2016-2
 
Slides ensae-2016-1
Slides ensae-2016-1Slides ensae-2016-1
Slides ensae-2016-1
 
Slides ensae-2016-6
Slides ensae-2016-6Slides ensae-2016-6
Slides ensae-2016-6
 
Slides ensae-2016-9
Slides ensae-2016-9Slides ensae-2016-9
Slides ensae-2016-9
 
Slides ensae-2016-7
Slides ensae-2016-7Slides ensae-2016-7
Slides ensae-2016-7
 
Slides ensae-2016-5
Slides ensae-2016-5Slides ensae-2016-5
Slides ensae-2016-5
 
Slides ensae-2016-3
Slides ensae-2016-3Slides ensae-2016-3
Slides ensae-2016-3
 
Pricing Game, 100% Data Sciences
Pricing Game, 100% Data SciencesPricing Game, 100% Data Sciences
Pricing Game, 100% Data Sciences
 

Similar to Quantile Estimation and Optimal Portfolios Optimization

Options Portfolio Selection
Options Portfolio SelectionOptions Portfolio Selection
Options Portfolio Selectionguasoni
 
Solution to the Practice Test 3A, Normal Probability Distribution
Solution to the Practice Test 3A, Normal Probability DistributionSolution to the Practice Test 3A, Normal Probability Distribution
Solution to the Practice Test 3A, Normal Probability DistributionLong Beach City College
 
Solution to the Practice Test 3A, Chapter 6 Normal Probability Distribution
Solution to the Practice Test 3A, Chapter 6 Normal Probability DistributionSolution to the Practice Test 3A, Chapter 6 Normal Probability Distribution
Solution to the Practice Test 3A, Chapter 6 Normal Probability DistributionLong Beach City College
 
Error analysis statistics
Error analysis   statisticsError analysis   statistics
Error analysis statisticsTarun Gehlot
 
DDKT-Munich.pdf
DDKT-Munich.pdfDDKT-Munich.pdf
DDKT-Munich.pdfGRAPE
 
Litv_Denmark_Weak_Supervised_Learning.pdf
Litv_Denmark_Weak_Supervised_Learning.pdfLitv_Denmark_Weak_Supervised_Learning.pdf
Litv_Denmark_Weak_Supervised_Learning.pdfAlexander Litvinenko
 
Hands-On Algorithms for Predictive Modeling
Hands-On Algorithms for Predictive ModelingHands-On Algorithms for Predictive Modeling
Hands-On Algorithms for Predictive ModelingArthur Charpentier
 
asset-v1_MITx+18.6501x+2T2020+type@asset+block@lectureslides_Chap8-noPhantom.pdf
asset-v1_MITx+18.6501x+2T2020+type@asset+block@lectureslides_Chap8-noPhantom.pdfasset-v1_MITx+18.6501x+2T2020+type@asset+block@lectureslides_Chap8-noPhantom.pdf
asset-v1_MITx+18.6501x+2T2020+type@asset+block@lectureslides_Chap8-noPhantom.pdfPrasantaKumarMohapat2
 
ch7_lin_updatedApril13_2022.pdf
ch7_lin_updatedApril13_2022.pdfch7_lin_updatedApril13_2022.pdf
ch7_lin_updatedApril13_2022.pdfCharlieTsui5
 
DDKT-SAET.pdf
DDKT-SAET.pdfDDKT-SAET.pdf
DDKT-SAET.pdfGRAPE
 
Expanding further the universe of exotic options closed pricing formulas in t...
Expanding further the universe of exotic options closed pricing formulas in t...Expanding further the universe of exotic options closed pricing formulas in t...
Expanding further the universe of exotic options closed pricing formulas in t...caplogic-ltd
 
DDKT-Praga.pdf
DDKT-Praga.pdfDDKT-Praga.pdf
DDKT-Praga.pdfGRAPE
 

Similar to Quantile Estimation and Optimal Portfolios Optimization (20)

Options Portfolio Selection
Options Portfolio SelectionOptions Portfolio Selection
Options Portfolio Selection
 
Solution to the Practice Test 3A, Normal Probability Distribution
Solution to the Practice Test 3A, Normal Probability DistributionSolution to the Practice Test 3A, Normal Probability Distribution
Solution to the Practice Test 3A, Normal Probability Distribution
 
Slides ima
Slides imaSlides ima
Slides ima
 
Solution to the Practice Test 3A, Chapter 6 Normal Probability Distribution
Solution to the Practice Test 3A, Chapter 6 Normal Probability DistributionSolution to the Practice Test 3A, Chapter 6 Normal Probability Distribution
Solution to the Practice Test 3A, Chapter 6 Normal Probability Distribution
 
Error analysis statistics
Error analysis   statisticsError analysis   statistics
Error analysis statistics
 
DDKT-Munich.pdf
DDKT-Munich.pdfDDKT-Munich.pdf
DDKT-Munich.pdf
 
Classification
ClassificationClassification
Classification
 
Litv_Denmark_Weak_Supervised_Learning.pdf
Litv_Denmark_Weak_Supervised_Learning.pdfLitv_Denmark_Weak_Supervised_Learning.pdf
Litv_Denmark_Weak_Supervised_Learning.pdf
 
Side 2019, part 2
Side 2019, part 2Side 2019, part 2
Side 2019, part 2
 
report
reportreport
report
 
Hands-On Algorithms for Predictive Modeling
Hands-On Algorithms for Predictive ModelingHands-On Algorithms for Predictive Modeling
Hands-On Algorithms for Predictive Modeling
 
asset-v1_MITx+18.6501x+2T2020+type@asset+block@lectureslides_Chap8-noPhantom.pdf
asset-v1_MITx+18.6501x+2T2020+type@asset+block@lectureslides_Chap8-noPhantom.pdfasset-v1_MITx+18.6501x+2T2020+type@asset+block@lectureslides_Chap8-noPhantom.pdf
asset-v1_MITx+18.6501x+2T2020+type@asset+block@lectureslides_Chap8-noPhantom.pdf
 
Slides erm-cea-ia
Slides erm-cea-iaSlides erm-cea-ia
Slides erm-cea-ia
 
pattern recognition
pattern recognition pattern recognition
pattern recognition
 
ch7_lin_updatedApril13_2022.pdf
ch7_lin_updatedApril13_2022.pdfch7_lin_updatedApril13_2022.pdf
ch7_lin_updatedApril13_2022.pdf
 
Side 2019, part 1
Side 2019, part 1Side 2019, part 1
Side 2019, part 1
 
DDKT-SAET.pdf
DDKT-SAET.pdfDDKT-SAET.pdf
DDKT-SAET.pdf
 
Inequality #4
Inequality #4Inequality #4
Inequality #4
 
Expanding further the universe of exotic options closed pricing formulas in t...
Expanding further the universe of exotic options closed pricing formulas in t...Expanding further the universe of exotic options closed pricing formulas in t...
Expanding further the universe of exotic options closed pricing formulas in t...
 
DDKT-Praga.pdf
DDKT-Praga.pdfDDKT-Praga.pdf
DDKT-Praga.pdf
 

More from Arthur Charpentier (20)

Family History and Life Insurance
Family History and Life InsuranceFamily History and Life Insurance
Family History and Life Insurance
 
ACT6100 introduction
ACT6100 introductionACT6100 introduction
ACT6100 introduction
 
Family History and Life Insurance (UConn actuarial seminar)
Family History and Life Insurance (UConn actuarial seminar)Family History and Life Insurance (UConn actuarial seminar)
Family History and Life Insurance (UConn actuarial seminar)
 
Control epidemics
Control epidemics Control epidemics
Control epidemics
 
STT5100 Automne 2020, introduction
STT5100 Automne 2020, introductionSTT5100 Automne 2020, introduction
STT5100 Automne 2020, introduction
 
Family History and Life Insurance
Family History and Life InsuranceFamily History and Life Insurance
Family History and Life Insurance
 
Machine Learning in Actuarial Science & Insurance
Machine Learning in Actuarial Science & InsuranceMachine Learning in Actuarial Science & Insurance
Machine Learning in Actuarial Science & Insurance
 
Reinforcement Learning in Economics and Finance
Reinforcement Learning in Economics and FinanceReinforcement Learning in Economics and Finance
Reinforcement Learning in Economics and Finance
 
Optimal Control and COVID-19
Optimal Control and COVID-19Optimal Control and COVID-19
Optimal Control and COVID-19
 
Slides OICA 2020
Slides OICA 2020Slides OICA 2020
Slides OICA 2020
 
Lausanne 2019 #3
Lausanne 2019 #3Lausanne 2019 #3
Lausanne 2019 #3
 
Lausanne 2019 #4
Lausanne 2019 #4Lausanne 2019 #4
Lausanne 2019 #4
 
Lausanne 2019 #2
Lausanne 2019 #2Lausanne 2019 #2
Lausanne 2019 #2
 
Lausanne 2019 #1
Lausanne 2019 #1Lausanne 2019 #1
Lausanne 2019 #1
 
Side 2019 #10
Side 2019 #10Side 2019 #10
Side 2019 #10
 
Side 2019 #11
Side 2019 #11Side 2019 #11
Side 2019 #11
 
Side 2019 #12
Side 2019 #12Side 2019 #12
Side 2019 #12
 
Side 2019 #9
Side 2019 #9Side 2019 #9
Side 2019 #9
 
Side 2019 #8
Side 2019 #8Side 2019 #8
Side 2019 #8
 
Side 2019 #7
Side 2019 #7Side 2019 #7
Side 2019 #7
 

Recently uploaded

Sales & Marketing Alignment: How to Synergize for Success
Sales & Marketing Alignment: How to Synergize for SuccessSales & Marketing Alignment: How to Synergize for Success
Sales & Marketing Alignment: How to Synergize for SuccessAggregage
 
Yaroslav Rozhankivskyy: Три складові і три передумови максимальної продуктивн...
Yaroslav Rozhankivskyy: Три складові і три передумови максимальної продуктивн...Yaroslav Rozhankivskyy: Три складові і три передумови максимальної продуктивн...
Yaroslav Rozhankivskyy: Три складові і три передумови максимальної продуктивн...Lviv Startup Club
 
Regression analysis: Simple Linear Regression Multiple Linear Regression
Regression analysis:  Simple Linear Regression Multiple Linear RegressionRegression analysis:  Simple Linear Regression Multiple Linear Regression
Regression analysis: Simple Linear Regression Multiple Linear RegressionRavindra Nath Shukla
 
VIP Kolkata Call Girl Howrah 👉 8250192130 Available With Room
VIP Kolkata Call Girl Howrah 👉 8250192130  Available With RoomVIP Kolkata Call Girl Howrah 👉 8250192130  Available With Room
VIP Kolkata Call Girl Howrah 👉 8250192130 Available With Roomdivyansh0kumar0
 
Socio-economic-Impact-of-business-consumers-suppliers-and.pptx
Socio-economic-Impact-of-business-consumers-suppliers-and.pptxSocio-economic-Impact-of-business-consumers-suppliers-and.pptx
Socio-economic-Impact-of-business-consumers-suppliers-and.pptxtrishalcan8
 
Call Girls Navi Mumbai Just Call 9907093804 Top Class Call Girl Service Avail...
Call Girls Navi Mumbai Just Call 9907093804 Top Class Call Girl Service Avail...Call Girls Navi Mumbai Just Call 9907093804 Top Class Call Girl Service Avail...
Call Girls Navi Mumbai Just Call 9907093804 Top Class Call Girl Service Avail...Dipal Arora
 
Monte Carlo simulation : Simulation using MCSM
Monte Carlo simulation : Simulation using MCSMMonte Carlo simulation : Simulation using MCSM
Monte Carlo simulation : Simulation using MCSMRavindra Nath Shukla
 
Monthly Social Media Update April 2024 pptx.pptx
Monthly Social Media Update April 2024 pptx.pptxMonthly Social Media Update April 2024 pptx.pptx
Monthly Social Media Update April 2024 pptx.pptxAndy Lambert
 
Mysore Call Girls 8617370543 WhatsApp Number 24x7 Best Services
Mysore Call Girls 8617370543 WhatsApp Number 24x7 Best ServicesMysore Call Girls 8617370543 WhatsApp Number 24x7 Best Services
Mysore Call Girls 8617370543 WhatsApp Number 24x7 Best ServicesDipal Arora
 
Catalogue ONG NUOC PPR DE NHAT .pdf
Catalogue ONG NUOC PPR DE NHAT      .pdfCatalogue ONG NUOC PPR DE NHAT      .pdf
Catalogue ONG NUOC PPR DE NHAT .pdfOrient Homes
 
Call Girls In Panjim North Goa 9971646499 Genuine Service
Call Girls In Panjim North Goa 9971646499 Genuine ServiceCall Girls In Panjim North Goa 9971646499 Genuine Service
Call Girls In Panjim North Goa 9971646499 Genuine Serviceritikaroy0888
 
Ensure the security of your HCL environment by applying the Zero Trust princi...
Ensure the security of your HCL environment by applying the Zero Trust princi...Ensure the security of your HCL environment by applying the Zero Trust princi...
Ensure the security of your HCL environment by applying the Zero Trust princi...Roland Driesen
 
MONA 98765-12871 CALL GIRLS IN LUDHIANA LUDHIANA CALL GIRL
MONA 98765-12871 CALL GIRLS IN LUDHIANA LUDHIANA CALL GIRLMONA 98765-12871 CALL GIRLS IN LUDHIANA LUDHIANA CALL GIRL
MONA 98765-12871 CALL GIRLS IN LUDHIANA LUDHIANA CALL GIRLSeo
 
Tech Startup Growth Hacking 101 - Basics on Growth Marketing
Tech Startup Growth Hacking 101  - Basics on Growth MarketingTech Startup Growth Hacking 101  - Basics on Growth Marketing
Tech Startup Growth Hacking 101 - Basics on Growth MarketingShawn Pang
 
Insurers' journeys to build a mastery in the IoT usage
Insurers' journeys to build a mastery in the IoT usageInsurers' journeys to build a mastery in the IoT usage
Insurers' journeys to build a mastery in the IoT usageMatteo Carbone
 
Creating Low-Code Loan Applications using the Trisotech Mortgage Feature Set
Creating Low-Code Loan Applications using the Trisotech Mortgage Feature SetCreating Low-Code Loan Applications using the Trisotech Mortgage Feature Set
Creating Low-Code Loan Applications using the Trisotech Mortgage Feature SetDenis Gagné
 
Eni 2024 1Q Results - 24.04.24 business.
Eni 2024 1Q Results - 24.04.24 business.Eni 2024 1Q Results - 24.04.24 business.
Eni 2024 1Q Results - 24.04.24 business.Eni
 
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999Tina Ji
 
A DAY IN THE LIFE OF A SALESMAN / WOMAN
A DAY IN THE LIFE OF A  SALESMAN / WOMANA DAY IN THE LIFE OF A  SALESMAN / WOMAN
A DAY IN THE LIFE OF A SALESMAN / WOMANIlamathiKannappan
 

Recently uploaded (20)

Sales & Marketing Alignment: How to Synergize for Success
Sales & Marketing Alignment: How to Synergize for SuccessSales & Marketing Alignment: How to Synergize for Success
Sales & Marketing Alignment: How to Synergize for Success
 
Yaroslav Rozhankivskyy: Три складові і три передумови максимальної продуктивн...
Yaroslav Rozhankivskyy: Три складові і три передумови максимальної продуктивн...Yaroslav Rozhankivskyy: Три складові і три передумови максимальної продуктивн...
Yaroslav Rozhankivskyy: Три складові і три передумови максимальної продуктивн...
 
Regression analysis: Simple Linear Regression Multiple Linear Regression
Regression analysis:  Simple Linear Regression Multiple Linear RegressionRegression analysis:  Simple Linear Regression Multiple Linear Regression
Regression analysis: Simple Linear Regression Multiple Linear Regression
 
VIP Kolkata Call Girl Howrah 👉 8250192130 Available With Room
VIP Kolkata Call Girl Howrah 👉 8250192130  Available With RoomVIP Kolkata Call Girl Howrah 👉 8250192130  Available With Room
VIP Kolkata Call Girl Howrah 👉 8250192130 Available With Room
 
Socio-economic-Impact-of-business-consumers-suppliers-and.pptx
Socio-economic-Impact-of-business-consumers-suppliers-and.pptxSocio-economic-Impact-of-business-consumers-suppliers-and.pptx
Socio-economic-Impact-of-business-consumers-suppliers-and.pptx
 
Call Girls Navi Mumbai Just Call 9907093804 Top Class Call Girl Service Avail...
Call Girls Navi Mumbai Just Call 9907093804 Top Class Call Girl Service Avail...Call Girls Navi Mumbai Just Call 9907093804 Top Class Call Girl Service Avail...
Call Girls Navi Mumbai Just Call 9907093804 Top Class Call Girl Service Avail...
 
Monte Carlo simulation : Simulation using MCSM
Monte Carlo simulation : Simulation using MCSMMonte Carlo simulation : Simulation using MCSM
Monte Carlo simulation : Simulation using MCSM
 
Monthly Social Media Update April 2024 pptx.pptx
Monthly Social Media Update April 2024 pptx.pptxMonthly Social Media Update April 2024 pptx.pptx
Monthly Social Media Update April 2024 pptx.pptx
 
Mysore Call Girls 8617370543 WhatsApp Number 24x7 Best Services
Mysore Call Girls 8617370543 WhatsApp Number 24x7 Best ServicesMysore Call Girls 8617370543 WhatsApp Number 24x7 Best Services
Mysore Call Girls 8617370543 WhatsApp Number 24x7 Best Services
 
Catalogue ONG NUOC PPR DE NHAT .pdf
Catalogue ONG NUOC PPR DE NHAT      .pdfCatalogue ONG NUOC PPR DE NHAT      .pdf
Catalogue ONG NUOC PPR DE NHAT .pdf
 
Call Girls In Panjim North Goa 9971646499 Genuine Service
Call Girls In Panjim North Goa 9971646499 Genuine ServiceCall Girls In Panjim North Goa 9971646499 Genuine Service
Call Girls In Panjim North Goa 9971646499 Genuine Service
 
Ensure the security of your HCL environment by applying the Zero Trust princi...
Ensure the security of your HCL environment by applying the Zero Trust princi...Ensure the security of your HCL environment by applying the Zero Trust princi...
Ensure the security of your HCL environment by applying the Zero Trust princi...
 
MONA 98765-12871 CALL GIRLS IN LUDHIANA LUDHIANA CALL GIRL
MONA 98765-12871 CALL GIRLS IN LUDHIANA LUDHIANA CALL GIRLMONA 98765-12871 CALL GIRLS IN LUDHIANA LUDHIANA CALL GIRL
MONA 98765-12871 CALL GIRLS IN LUDHIANA LUDHIANA CALL GIRL
 
KestrelPro Flyer Japan IT Week 2024 (English)
KestrelPro Flyer Japan IT Week 2024 (English)KestrelPro Flyer Japan IT Week 2024 (English)
KestrelPro Flyer Japan IT Week 2024 (English)
 
Tech Startup Growth Hacking 101 - Basics on Growth Marketing
Tech Startup Growth Hacking 101  - Basics on Growth MarketingTech Startup Growth Hacking 101  - Basics on Growth Marketing
Tech Startup Growth Hacking 101 - Basics on Growth Marketing
 
Insurers' journeys to build a mastery in the IoT usage
Insurers' journeys to build a mastery in the IoT usageInsurers' journeys to build a mastery in the IoT usage
Insurers' journeys to build a mastery in the IoT usage
 
Creating Low-Code Loan Applications using the Trisotech Mortgage Feature Set
Creating Low-Code Loan Applications using the Trisotech Mortgage Feature SetCreating Low-Code Loan Applications using the Trisotech Mortgage Feature Set
Creating Low-Code Loan Applications using the Trisotech Mortgage Feature Set
 
Eni 2024 1Q Results - 24.04.24 business.
Eni 2024 1Q Results - 24.04.24 business.Eni 2024 1Q Results - 24.04.24 business.
Eni 2024 1Q Results - 24.04.24 business.
 
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
Russian Faridabad Call Girls(Badarpur) : ☎ 8168257667, @4999
 
A DAY IN THE LIFE OF A SALESMAN / WOMAN
A DAY IN THE LIFE OF A  SALESMAN / WOMANA DAY IN THE LIFE OF A  SALESMAN / WOMAN
A DAY IN THE LIFE OF A SALESMAN / WOMAN
 

Quantile Estimation and Optimal Portfolios Optimization

  • 1. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Quantile estimation and optimal portfolios Arthur Charpentier & Abder Oulidi ENSAI-ENSAE-CREST & IMA Angers Journée SFdS, Mai 2007 1
  • 2. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Portfolio management and optimal allocations Idea: allocating capital among a set of assets to maximize return and minimize risk. If diversification effects were intuited early, and Markowitz (1952) proposed a mathematical model. • return is measured by the expected value of the portfolio return, • risk is quantified by the variance of this return. Agenda 1. statistical issue in the mean-variance framework 2. portfolio optimization with general risk measures 2
  • 3. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Portfolio optimization (parametric framework) Consider a risk measure R (variance or Value-at-Risk). Solve ω∗ =    argmin{R(ωt X)}, u.c. ωt 1 = 1 and E(ωt X) ≥ η, where X ∼ L(θ), θ unknown. θ is unknown but can be estimated using a sample {X1, . . . , Xn}. “The parameters governing the central tendency and dispersion of returns are usually not known, however; and are often estimated or guessed at using observed returns and other available data. In empirical applications, the estimated parameters are used as if they were the true value” (Coles & Loewenstein (1988)). If ω∗ = ψ(θ) (e.g. mean-variance) ω∗ = ψ(θ). 3
  • 4. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Optimization of standard deviation (or variance) Allocation in the first asset Allocation in the second asset Standard deviation of the portfolio −200 −100 0 100 200 −200−1000100200 Allocation in the first asset Allocationinthesecondasset Figure 1: Portfolio variance optimization problem. 4
  • 5. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Portfolio optimization (parametric framework) In the case of no explicit expression of the optimum, solve (numerically) ω∗ =    argminR(ωt X), u.c.ω ∈ {(ωk)k∈{1,...,m}} where X ∼ L(θ). The idea is to generate samples Xi’s, • either from a parametric distribution L(θ), • or from a nonparametric distribution (bootstrap approach). 5
  • 6. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Optimization of Value-at-Risk VaR of the portfolio −4 −2 0 2 4 6 −2−1012345 −4 −2 0 2 4 6 VaR of the portfolio −4 −2 0 2 4 6 012345 −4 −2 0 2 4 6 Figure 2: Optimization in the mean-VaR framework. 6
  • 7. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Classical mean-variance allocation problem Consider d risky assets, with weekly returns X = (X1, . . . , Xd). Denote µ = E(X) and Σ = var(X). Let ω = (ω1, . . . , ωd) ∈ Rd denote the weights in all risky assets. • the expected return of the portfolio is E(ωt X) = ωt µ, • the variance of the portfolio is var(ωt X) = ωt Σω.    ω∗ ∈ argmin{ωt Σω} u.c. ωt µ ≥ η and ωt 1 = 1 convex ⇐⇒    ω∗ ∈ argmax{ωt µ} u.c. ωt Σω ≤ η and ωt 1 = 1 7
  • 8. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Classical mean-variance allocation problem The solution can be given explicitly (see Markowitz (1952)) as ω∗ = ψ(µ, Σ) = p + ηq where µ = E(X), Σ = var(X), p = 1 d bΣ−1 1 − aΣ−1 µ and q = 1 d cΣ−1 µ − aΣ−1 1 , and a = 1t Σ−1 µ, b = µt Σ−1 µ, c = 1t Σ−1 1, d = bc − a2 . Note that p is an allocation, and q indicates how the original portfolio should be modified. 8
  • 9. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Efficient frontier first asset −0.2 0.0 0.2 0.4 −0.4 0.0 0.4 −0.20.00.2 −0.20.00.20.4 second asset third asset −0.20.20.6 −0.2 0.0 0.2 −0.40.00.4 −0.2 0.2 0.6 fourth asset Portfolio with 4assets 0.010 0.015 0.020 0.025 0.030 0.0000.0010.0020.0030.0040.0050.006 Efficient Frontier Standard deviation Expectedvalue Figure 3: Solving a variance optimization problem. 9
  • 10. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Inference issues In practice, µ = [µi] and Σ = [Σi,j] are unknown, and should be estimated. A natural idea is to define µi = 1 n n t=1 Xi,t et Σi,j = 1 n − 1 n t=1 (Xi,t − µi)(Xj,t − µj). Given n observed observed returns, µ|Σ ∼ N µ, Σ n and nΣ|Σ ∼ W (n − 1, Σ) . where the two random variables µ and Σ are independent, given Σ. 10
  • 11. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios 0.05 0.10 0.15 0.20 0.25 0.000.050.100.150.200.25 Efficient Frontier, with 250 past observations Standard deviation Expectedvalue 0.05 0.10 0.15 0.20 0.25 0.000.050.100.150.200.25 Efficient Frontier, with 1000 past observations Standard deviation Expectedvalue Figure 4: Efficient frontiers and estimation. 11
  • 12. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Parametric bootstrap Assume that X ∼ L(θ). estimate θ by θn. The procedure is the following 1. generate n returns X1, . . . , Xn from L(θn); 2. estimate µ and Σ, i.e. µn and Σn, 3. solve the minimization problem, i.e. ω∗ = 1 d bΣ −1 1 − aΣ −1 µ + η 1 d cΣ −1 µ − aΣ −1 1 , Using several simulations, the distribution of the ω∗ k and vark(ω∗ k t X) can be obtained. 12
  • 13. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Nonparametric bootstrap A nonparametric procedure can also be considered. Consider a n sample {X1, . . . , Xn} 1. generate a bootstrap sample from {X1, . . . , Xn} 2. estimate µ and Σ, i.e. µn and Σn, 3. solve the minimization problem, i.e. ω∗ = 1 d bΣ −1 1 − aΣ −1 µ + η 1 d cΣ −1 µ − aΣ −1 1 , Using several simulations, the distribution of the ω∗ k and vark(ω∗ k t X) can be obtained. 13
  • 14. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios 0.4 0.6 0.8 1.0 012345 Allocation in the first asset Allocation weight Density 0.0 0.2 0.4 012345 Allocation in the second asset Allocation weight Density −0.3 −0.1 0.0 0.1 02468 Allocation in the third asset Allocation weight Density 0.05 0.15 0.25 0246810 Allocation in the fourth asset Allocation weightDensity Figure 5: Distributions of optimal allocations ω∗ k’s. 14
  • 15. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinsecondasset Joint distribution of optimal allocations (1−2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinthirdasset Joint distribution of optimal allocations (1−3) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinfourthasset Joint distribution of optimal allocations (1−4) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the second asset Allocationinfourthasset Joint distribution of optimal allocations (2−4) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the second asset Allocationinthirdasset Joint distribution of optimal allocations (2−3) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the third asset Allocationinfourthasset Joint distribution of optimal allocations (3−4) Figure 6: Joint distributions of optimal allocations ω∗ k’s. 15
  • 16. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios 0.05 0.06 0.07 0.08 0.09 0.10 0.11 020406080100 Density of estimated optimal standard deviation Optimal standard deviation Density Figure 7: Distribution of vark(ω∗ k t X) 16
  • 17. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Value-at-Risk minimization With V aR(X, p) = F−1 (p) = sup{x, F(x) < p}, the program is    ω∗ ∈ argmin{VaR(ωt X, α)} u.c. E(ωt X) ≥ η,ωt 1 = 1 nonconvex    ω∗ ∈ argmax{E(ωt X)} u.c. {VaR(ωt X, α)} ≤ η ,ωt 1 = 1 In the previous framework (mean-variance), it could be done easily since • there are only a few estimates of the variance • there exists an analytical expression of the optimal allocation, In the case of Value-at- Risk minimization, • there are several estimators of quantiles (see Charpentier & Oulidi (2007)), • numerical optimization should be considered. 17
  • 18. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Quantile estimation • raw estimator of the quantile, X[pn]:n = Xi:n = F−1 n (i/n) such that i ≤ pn < i + 1. • weighted average of F−1 n (p), e.g. αXi:n + (1 − α)Xi+1:n, • weighted average of F−1 n (p), e.g. n i=1 αiXi:n = 1 0 αuF−1 n (u)du, • smoothed version of the cdf, F−1 K (p) where FK(x) = 1 nh n i=1 K x − Xi h • semiparametric approach, based on Hill’s estimator, Xn−k:n n k (1 − p) −ξk , where ξk = 1 k k i=1 log Xn+1−i:n − log Xn−k:n (if ξ > 0), • fully parametric approach, Xn + u1−pvar(X) (if X ∼ N(µ, σ2 )) 18
  • 19. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios 0.0 0.5 1.0 1.5 2.0 0.00.20.40.60.81.0 Empirical quantile estimation Value Probability 0.0 0.5 1.0 1.5 2.0 0.00.20.40.60.81.0 Empirical quantile estimation Value Probability Figure 8: Classical estimation of the quantile, based on F−1 (·). 19
  • 20. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios 0.0 0.5 1.0 1.5 2.0 0.00.51.01.5 Smoothed empirical quantile estimation Value Smootheddensity 0.0 0.5 1.0 1.5 2.0 0.00.20.40.60.81.0 Smoothed empirical quantile estimation ValueProbability Figure 9: Smoothed estimation of the quantile, based on F−1 K (·). 20
  • 21. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios A short extention to general risk measures In a much more general setting, spectral risk measures can be considered, i.e. R(X) = 1 0 φ(p)F−1 X (p)dp, for some distortion function φ : [0, 1] → [0, 1]. 21
  • 22. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Parametric bootstrap Assume that X ∼ L(θ). The procedure is the following 1. generate n returns X1, . . . , Xn from L(θ); 2. estimate for all ω on a finite grid, estimate VaR(ωt X), 3. solve the minimization problem on the grid to get numerically ω∗ n. Using several simulations, the distribution of ω∗ n and var(ω∗ nX) can be obtained. 22
  • 23. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinsecondasset Joint distribution of optimal allocations (1−2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinthirdasset Joint distribution of optimal allocations (1−3) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinfourthasset Joint distribution of optimal allocations (1−4) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the second asset Allocationinfourthasset Joint distribution of optimal allocations (2−4) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the third asset Allocationinfourthasset Joint distribution of optimal allocations (3−4) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the second asset Allocationinthirdasset Joint distribution of optimal allocations (2−3) Figure 10: Joint distributions of optimal allocations ω∗ k’s, smoothed quantile estimator. 23
  • 24. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios 0.16 0.18 0.20 0.22 0.24 05101520 Density of estimated optimal 99% quantile Optimal Value−at−Risk Density Figure 11: Distribution of VaRk(ω∗ k t X, 95%). 24
  • 25. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinsecondasset Joint distribution of optimal allocations (1−2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinthirdasset Joint distribution of optimal allocations (1−3) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the first asset Allocationinfourthasset Joint distribution of optimal allocations (1−4) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the second asset Allocationinfourthasset Joint distribution of optimal allocations (2−4) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the third asset Allocationinfourthasset Joint distribution of optimal allocations (3−4) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 −0.20.00.20.40.60.81.0 Allocation in the second asset Allocationinthirdasset Joint distribution of optimal allocations (2−3) Figure 12: Joint distributions of optimal allocations ω∗ k’s, raw quantile estima- tor. 25
  • 26. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios 0.16 0.18 0.20 0.22 0.24 010203040 Density of estimated optimal 95% quantile Optimal Value−at−Risk Density Figure 13: Distribution of VaRk(ω∗ k t X, 95%). 26
  • 27. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Conclusion Dealing with only 4 assets, it is difficult to get robust optimal allocation, only because of statistical uncertainty of classical estimators. Remark: this was mentioned in Liu (2003) on high frequency data (every 5 minutes, i.e. n = 10, 000) with 100 assets. 27
  • 28. Arthur CHARPENTIER & Abder OULIDI - Quantile estimation and optimal portfolios Some references Coles, J.L. & Loewenstein, U. (1988). Equilibrium pricing and portfolio composition in the presence of uncertain parameters. Journal of Financial Economics, 22, 279-303. Dowd, K. & Blake, D.. (2006). After VaR: the theory, estimation, and insurance applications of quantile-based risk measures. Journal of Risk & Insurance, 73, 193-229. Duarte, A. (1999). Fast computation of efficient portfolios. Journal of Risk, 1, 71-94. Duffie, D. & Pan, J. (1997). An overview of Value at Risk. Journal of Derivatives, 4, 7-49. Gaivoronski, A.A. & Pflug, G. (2000). Value-at-Risk in portfolio optimization: properties and computational approach. Working Paper 00-2, Norwegian University of Sciences & Technology. Jorion, P. (1997). Value at Risk: the new benchmark for controlling market risk. McGraw-Hill. Kast, R., Luciano, E. & Peccati, L. (1998). VaR and optimization: 2nd international workshop on preferences and decisions. Trento, July 1998. Klein, R.W. & Bawa, V.S. (1976). The effect of estimation risk on optimal portfolio choice. Journal of Financial Economics, 3, 215-231. Larsen, N., Mausser, H. & Uryasev, S. (2002). Algorithms for optimization of Value at Risk. in Financial engineering, e-commerce and supply-chain, Pardalos and Tsitsiringos eds., Kluwer Academic Publichers, 129-157. Litterman, R. (1997). Hot spots and edges II. Risk, 10, 38-42. Rockafellar, R.T. & Uryasev, S. (2000). Optimization of Conditional Value-at-Risk. Journal of Risk, 2, 21-41. 28