Technical Analysis Techniques versus

Mathematical Models: Boundaries of Their

Validity Domains
Christophette Blanchet-Scalliet1 , Awa Diop2 , Rajna Gibson3 , Denis Talay2 ,
and Etienne Tanre2
1

Laboratoire Dieudonne, universite Nice Sophia-Antipolis Parc Valrose 06108

Nice Cedex 2, France [email protected]
INRIA, Projet OMEGA, 2004 route des Lucioles, BP93, 06902 Sophia-Antipolis,
France {Awa.Diop,Denis.Talay,Etienne.Tanre}@sophia.inria.fr
NCCR FINRISK, Swiss Banking Institute, University of Z
urich, Plattenstrasse
14, Z
urich 8032, Switzerland [email protected]

Abstract We aim to compare financial technical analysis techniques to strategies

which depend on a mathematical model. In this paper, we consider the moving
average indicator and an investor using a risky asset whose instantaneous rate of
return changes at an unknown random time. We construct mathematical strategies.
We compare their performances to technical analysis techniques when the model is
misspecified. The comparisons are based on Monte Carlo simulations.

1 Introduction
In the financial industry, there are three main approaches to investment: the
fundamental approach, where strategies are based on fundamental economic
principles, the technical analysis approach, where strategies are based on past
prices behavior, and the mathematical approach where strategies are based on
mathematical models and studies. The main advantage of technical analysis is
that it avoids model specification, and thus calibration problems, misspecification risks, etc. On the other hand, technical analysis techniques have limited
theoretical justifications, and therefore no one can assert that they are riskless, or even efficient (see [LMW00]).
Consider a nonstationary financial economy. It is impossible to specify and
calibrate models which can capture all the sources of instability during a long
time interval. Thus it might be useful to compare the performances obtained
by using erroneously calibrated mathematical models and the performances
obtained by technical analysis techniques.

C. Blanchet-Scalliet, A. Diop, R. Gibson, D. Talay, E. Tanre

To our knowledge, this question has not yet been investigated in the literature. The purpose of this paper is to present its mathematical complexity
and preliminary results.
Here we consider the case of an asset whose instantaneous expected rate
of return changes at an unknown random time. We compare the performances
of traders who respectively use:
a strategy which is optimal when the model is perfectly specified and
calibrated,
mathematical strategies for misspecified situations,
a technical analysis technique.
In all this paper, we limit ourselves to the case in which the traders utility function is logarithmic. Of course, it is a severe limitation from a financial point of view. This choice is also questionable from a numerical point
of view because logarithmic utilities tend to smoothen the effects of the different strategies. However, we will see that, even in this case and within a
simplified model, the analytical formulae are rather cumbersome and that our
analysis requires nonelementary mathematical and numerical tools. See also
the Remark 1 below.
Our study is divided into two parts: a mathematical part which, when possible, provides analytical formulae for portfolios managed by means of mathematical and technical analysis strategies; a numerical part which provides
quantitative comparisons between all these various strategies.

2 Description of the Setting

The financial market consists of two assets which are traded continuously.
The first one is an asset without systematic risk, typically a bond (or a bank
account), whose price at time t evolves according to
 0
dSt = St0 rdt,
(1)
S00 = 1.
The second asset is a stock subject to systematic risk. We model the evolution
of its price at time t by the linear stochastic differential equation



dSt = St 2 + (1 2 )1(t ) dt + St dBt ,
(2)
S0 = S 0 ,
where (Bt )0tT is a one-dimensional Brownian motion on a given probability
space (, F, P). At the random time , which is neither known, nor directly
observable, the instantaneous return rate changes from 1 to 2 . A simple
computation shows that


Z t
2
0
)t + (2 1 )
1( s) ds =: S 0 exp(Rt ),
St = S exp Bt + (1
2
0

Technical Analysis Techniques versus Mathematical Models

where the process (Rt )t0 is defined as



Z t
2
1( s) ds.
Rt = Bt + 1
t + (2 1 )
2
0

(3)

This model was considered by Shiryaev [Shi63] who studied the problem of
detecting the change time as early and reliably as possible when one only
observes the process (St )t0 .
Assumptions and Notation
The algebra generated by the observations at time t is denoted by
FtS := (Su , 0 u t) , t [0, T ].
Note that the Brownian motion (Bt )0tT is not adapted to the filtration
(FtS )t0 .
The Brownian motion (Bt )t0 and the random variable are independent.
The change time follows an exponential law 4 of parameter :
P ( > t) = et ,

t 0.

(4)

The value of the portfolio at time t is denoted by Wt .

We denote by Ft the conditional a posteriori probability (constructed by
means of the observation of the process S) that the change time has
occurred within the interval [0, t]:


Ft := P t/FtS .
(5)
We denote by (Lt )t0 the following exponential likelihood-ratio process :
(
1
Lt = exp
(2 1 )Rt
2
(6)
 )

1
2
2
2 (2 1 ) + 2(2 1 )(1
) t .
2
2
Finally, the parameters 1 , 2 , > 0 and r 0 are such that
1

2
2
< r < 2
.
2
2

Any other law is allowed to derive our main results.

C. Blanchet-Scalliet, A. Diop, R. Gibson, D. Talay, E. Tanre

3 A Technical Analysis Detection Strategy

3.1 Introduction
Technical analysis is an approach which is based on the prediction of the future evolution of a financial instrument price using only its price history. Thus,
technical analysts compute indicators which result from the past history of
transaction prices and volumes. These indicators are used as signals to anticipate future changes in prices (see, e.g., the book by Steve Achelis [Ach00]).
Here, we limit ourselves to the moving average indicator because it is
simple and often used to detect changes in return rates. To obtain its value,
one averages the closing prices of the stock during the most recent time
periods.
3.2 Moving Average Based on the Prices
Our trader takes decisions at discrete times. We thus consider a regular parT
:
tition of the interval [0, T ] with step t = N
0 = t0 < t1 < . . . < tN = T,

tn = nt.

We denote by t {0, 1} the proportion of the agents wealth invested in the

risky asset at time t, and by Mt the moving average of the prices. Therefore,
Mt =

Su du.

(7)

We suppose that, at time 0, the agent knows the history of the risky asset
prices before time 0 and has enough data to compute M0 .
At each tn , n [1 N ], the agent invests all his/her wealth into the risky
asset if the price Stn is larger than the moving average Mtn . Otherwise, he/she
invests all the wealth into the riskless asset. Consequently,
tn = 1(St M ) .
n
tn

(8)

Denote by x the initial wealth of the trader. The wealth at time tn+1 is
!
St0n+1
Stn+1
tn + 0 (1 tn ) ,
Wtn+1 = Wtn
Stn
Stn
and therefore, since St0n+1 /St0n = exp(rt),
WT = x

N
1
Y
n=0





tn exp(Rtn+1 Rtn ) exp(rt) + exp(rt) .

(9)

Technical Analysis Techniques versus Mathematical Models

3.3 The Particular Case of the Logarithmic Utility Function

One of our key results is
Proposition 1. The expectation of the logarithmic utility function of the
agents wealth is


2
(1)
r T p
E log(WT ) = log(x) + rT + 2
2



1 eT  (2)
2
(1)
(3)
(p p )e + p
r
+t 2
t
2
1e
1 eT (3)
p ,
t(2 1 )(et t)
1 et
where we have set


Z Z 2 3/2 (2 //2)2 (1+z2 )
z
z

2 2 y
(1)
2
e
i2 /2
dzdy,
(10)
p =
2y
2 y
0
y
Z Z
3/2
(2 //2)2 (v) (1+z22 )
z2 2
2 2 y

(2)
)
(
2
2
p =
1
e
z1
2y
4
2
+z2
y2
0
R
y1


 1 3/2 (1 //2)2 v (1+z2 )

z2
z1
z1
2 2 y1

2
1 i 2
i2 (v)/2
e
v/2
2 y2
2y1
2 y1
ev dy1 dz1 dy2 dz2 dv,


Z Z 1 3/2 (1 //2)2 (1+z2 )
z
z

2 2 y
(3)
2
e
i2 /2
dzdy,
p =
2y
2 y
0
y
Z
2
ze /4y z cosh(u)u2 /4y
iy (z) =
e
sinh(u) sin(u/2y)du.
y 0

(11)
(12)
(13)

Proof. The tedious calculation involves an explicit formula, due to Yor [Yor01],
Rt
for the density of ( 0 exp(2Bs )ds, Bt ). See [BSDG+ 05] for details.
3.4 Empirical Determination of a Good Windowing
One can optimize the choice of by using Proposition 1 and deterministic
numerical optimization procedures, or by means of Monte Carlo simulations.
In this subsection we present results obtained from Monte Carlo simulations,
which show that bad choices of may weaken the performance of the technical
analyst strategy. For each value of we have simulated 500,000 trajectories
of the asset price and computed the expectation E log(WT ) by a Monte Carlo
method. The parameters used to obtain Figure 1(a) and Figure 1(b) are all
equal but the volatility. It is clear from the figures that the optimal choice of
varies. When the volatility is 5 percent, the optimal choice of is around 0.3

C. Blanchet-Scalliet, A. Diop, R. Gibson, D. Talay, E. Tanre

whereas, when the volatility is 15 percent, the optimal choice of is around

0.8.
The parameters used to obtain Figure 1(b) and Figure 1(c) are all identical
but the maturity. The optimal choice of is around 0.3 when the maturity is
2 years, and is around 0.4 when the maturity is 3 years.
The empirical variance of log(WT ) is around 0.04. Thus, the Monte Carlo
error on E log(WT ) is of order 5.104 with probability 0.99. The number of
trajectories used for these simulations seems to be too large; however, considered as a function of , the quantity E log(WT ) varies very slowly, so that
we really need a large number of simulations to obtain the smooth curves
(Figure 1).

4.8

4.88
4.87

4.79

E(log(W_T))

4.77
4.76
4.75
4.74

4.85
4.84
4.83
4.82
4.81
4.8
4.79

4.73
4.72

4.78
0

0.2

0.4

0.6

0.8

1.2

1.4

4.77

1.6

order of moving average = delta

0.1

0.2

0.3

(b)
5.07

5.06

5.05

5.04

5.03

5.02

5.01

0.1

0.2

0.3

0.4

0.5

0.6

order of moving average = delta

(a)

E(log(W_T))

E(log(W_T))

4.86
4.78

0.4

0.5

0.6

0.7

0.8

0.9

order of moving average = delta

(c)

Fig. 1. E(log(WT )) as a function of .

Parameter 1 2
Figure 1(a) 0.2 0.2 2
Figure 1(b) 0.2 0.2 2
Figure 1(c) 0.2 0.2 2

0.15
0.05
0.05

r
0.0
0.0
0.0

T
2.0
2.0
3.0

0.7

Technical Analysis Techniques versus Mathematical Models

Figure 2 below illustrates the impact of the parameters 1 , 2 and on

the optimal choice of .
1=0.22=0.2

1=0.12=0.2
4.86

4.82

4.84

E[log(WT)]

4.84

4.8

4.82

4.78

4.8

4.76

4.78

4.74

0.5

1.5

4.76

0.5

1.5

1.5

1=0.12=0.1

1=0.22=0.1
4.7

4.7
4.69
E[log(WT)]

4.69

4.68
4.68

4.67
4.67

4.66
4.65

0.5

1
delta (years)

1.5

4.66

0.5

1
delta (years)

Fig. 2. Volatility and Optimal Moving Average Window Size: Plot of Expected
Value of the Log of Terminal Wealth vs. Window size with T = 2, = 2, - = 0.1,
- - = 0.15, and -. = 0.2

Remark 1. One can observe that the empirical optimal choices of are close
to the classical values used by the technical analysts, that is, around 200 days
or 50 days. One can also observe from Monte Carlo simulations that these
optimal values also hold when the traders utility function belongs to the
HARA family: see [BSDG+ 05].

4 The Optimal Portfolio Allocation Strategy

4.1 A General Formula
In this section our aim is to make explicit the optimal wealth and strategy of
a trader who perfectly knows all the parameters 1 , 2 , and . Of course,

C. Blanchet-Scalliet, A. Diop, R. Gibson, D. Talay, E. Tanre

this situation is unrealistic. However it is worth computing the best financial

performances that one can expect within our model. To be able to compare
this optimal strategy to a technical analyst strategy, we impose constraints on
the portfolio. Indeed, a technical analyst is only allowed to invest all his/her
wealth in the stock or the bond. Therefore the proportions of the traders
wealth invested in the stock are constrained to lie within the interval [0, 1].
To compute the constrained optimal wealth we use the martingale approach to stochastic control problems as developed by Karatzas, Shreve, Cvitanic, etc. More precisely, we follow and carefully adapt the martingale approach to the celebrated Merton problem [Mer71]. We emphasize that our
situation differs from the Merton problem by two aspects:
The drift coefficient of the dynamics of the risky asset is not constant over
time (since it changes at the random time ).
Here we must face some subtle measurability issues since the traders strategy needs to be adapted with respect to the filtration generated by (St ): as
already noticed, the drift change at the random time makes this filtration
different from the filtration generated by the Brownian motion (Bt ).
Let t be the proportion of the traders wealth invested in the stock at time
t; the remaining proportion 1 t is invested in the bond. For a given nonrandom initial capital x > 0, let Wx, denote the wealth process corresponding
to the portfolio ( ). Let A(x) denote the set of admissible portfolios, that is,
A(x) := { FtS progressively measurable process s.t.
W0x, = x, Wtx, > 0 for all t > 0, [0, 1]}.
The investors objective is to maximize his/her expected utility U of wealth
at the terminal time T . The value function thus is
V (x) :=

sup E U (WT ).

A(x)

As in Karatzas-Shreve [KS98], we introduce an auxiliary unconstrained

market defined as follows. We first decompose the process R in its own filtration as


2
) + (2 1 )Ft dt + dB t ,
dRt = (1
2
where B is the innovation process, i.e., the FtS - Brownian motion defined as


Z t
1
2
Bt =
Rt (1
)t (2 1 )
Fs ds , t 0,

2
0
where F is the conditional a posteriori probability (5).
Let D the subset of the {FtS } progressively measurable processes :
[0, T ] R such that

Technical Analysis Techniques versus Mathematical Models

Z
E

(t)dt < , where (t) := inf(0, (t)).

The bond price process S 0 () and the stock price S() satisfy
Rt
St0 () = 1 + 0 Su0 ()(r + (u))du,


Rt
St () = S0 + 0 Su () (1 + (2 1 )Fu + (u) + (u))du + dB u .
For each auxiliary unconstrained market driven by a process , the value
function is
V (, x) := sup Ex U (WT ()),
A(,x)

where
dWt () = Wt () (r + (t))dt + t (t)dt + (2 1 )Ft dt + (1 r)dt + dB t
Proposition 2. If there exists e such that
V (e
, x) = inf V (, x)

(14)

then there exists an optimal portfolio for which the optimal wealth is

Wt = Wt (e
).

(15)

An optimal portfolio allocation strategy is

t

:=

t
1 r + (2 1 )Ft + e(t)
Rt
+
rt

e (s)ds

0
Ht W t e

!
, (16)

where Ft defined in (5) satisfies

Ft =

Rt

es L1
s ds
,
Rt
t
s
1 + e Lt 0 e
L1
s ds
et Lt

and Hte is the exponential process defined by



R t 1 r + e(s) (2 1 )Fs

e
+
Ht = exp 0
dB s

!
2

1 R t 1 r + e(s) (2 1 )Fs
0
+
ds ,
2

and is a FtS adapted process which satisfies

Z t


RT
RT
E HTe erT 0 e (t)dt (U 0 )1 (HTe erT 0 e (t)dt ) / FtS = x +
s dB s .
0

Here, v is the Lagrange multiplier which makes the expectation of the left hand
side equal to x for all x.
Proof. See Karatzas-Shreve [KS98, p. 275] to prove (15). We obtain (16) by
solving the classical unconstrained problem for e.

10

C. Blanchet-Scalliet, A. Diop, R. Gibson, D. Talay, E. Tanre

4.2 The Particular Case of the Logarithmic Utility Function

Proposition 3. If U () = log() and the initial endowment is x, then the
optimal wealth process and strategy are
RT

xer(T t)+ t e (t)dt

=
,
Hte


1 r + (2 1 )Ft + e(t)
,
t =
2
Wt,x

(17)

where

(1 r + (2 1 )Ft )

1 r + (2 1 )Ft
< 0,
if
2
1 r + (2 1 )Ft
e(t) =
(18)
[0, 1],
0
if

2
(1 r + (2 1 )Ft )
otherwise,
and, as above,
e (t) = inf (0, e(t)) .
Remark 2. The optimal strategies for the constrained problem are the projections on [0, 1] of the optimal strategies for the unconstrained problem.
Remark 3. In the case of the logarithmic utility function, when t is small and
thus before the change time with high probability, one has Ft close to 0; as,
by hypothesis, one also has 1r
0, the optimal strategy is close to 0 ; after
2
the change time , one has Ft close to 1, and the optimal strategy is close to
min(1, 2r
2 ). In both cases, we approximately recover the optimal strategies
of the constrained Merton problem with drift parameters equal to 1 or 2
respectively.
Using (18) one can obtain an explicit formula for the value function corresponding to the optimal strategy:
E log(WT ) = log(x) + rT

Z T Z "
2
a

1
+
1 r + (2 1 )

2 1 + r
1+a
2
0
0
a>

2 2 + r
#

2
1
a

1 r + (2 1 )
1
1 r
2 1 + r
2
1+a

<a<
r
2 2 + r
2
et (1 + a)g(a, t)dadt.
(19)
Here, g(a, t)da is the density function of

Technical Analysis Techniques versus Mathematical Models


exp

11


2 1
(2 1 )2
s
+
s
ds,
Bs

2 2

is a Brownian Motion. This density admits a rather complex explicit

where B
expression which involves the function iy (z) defined as in (13): see Yor [Yor01]
and Borodin and Salminen [BS02].

5 A Model and Detect Strategy

The optimal strategy of the preceding section assumes that the trader has
chosen a mathematical model and controls the investment policy to optimize
the expected utility function of his/her wealth, and that the investment policy
is time continuous, whereas the technical analyst does not control the policy
and invests at discrete times according to a rupture detection rule. We now
consider the case of a trader who chooses a mathematical model and wants
to reinvest the portfolio only once, namely at the time where the change time
is optimally detected owing to the price history. In this section we describe
the wealth of such a trader, supposing that the reinvestment rule is the same
as the technical analysts one: at the detected change time from 1 to 2 , all
the portfolio is reinvested in the risky asset. We also continue to suppose for
a while that the trader perfectly knows all the parameters of the model.
In the full report [BSDG+ 05], we consider two detection methods proposed
by Karatzas [Kar03] and by Shiryaev [Shi02]. Here, we limit ourselves to
consider Karatzas optimal stopping rule K which minimizes the expected
miss
R() := E| |
(20)
over all stopping rules , where is a positive random variable. In view of
the results in [Kar03] one has here
Proposition 4. The stopping rule K that minimizes the expected miss E|
| over all the stopping rules with E() < is


Z t

p
 t
K
s 1
= inf t 0 e Lt
e
Ls ds
,
1 p
0
where Lt is defined as in (6), and p is the unique solution in ( 12 , 1) of the
equation
Z
0

1/2

(1 2s)e/s 2
s
ds =
(1 s)2+

with = 2 2 /(2 1 )2 .

1/2

(2s 1)e/s 2
s
ds
(1 s)2+

12

C. Blanchet-Scalliet, A. Diop, R. Gibson, D. Talay, E. Tanre

We thus are in a position to compute the wealth of the trader who uses
the strategy consisting in investing all of his/her money in the bond until K
and in the stock after K . The value of the portfolio at maturity T is
WT =

xS0K
ST 1(K T ) + xST0 1(K >T ) .
S K

In the particular case of the logarithmic utility function, one can exhibit an
exact formula for E(log(WT )). Unfortunately this new formula has a complexity similar to (19): see [BSDG+ 05]. However we can numerically compare
the performances of the two change time detection strategies (one based on
technical analysis, the other one based on a mathematical model), and the
optimal portfolio allocation strategy. Figure 3 illustrates, based on the typical
results that we have obtained so far, that the methods using mathematical
models have better performances than the technical analyst method.
4.85

Optimal allocation
Model and detect
Technical Analyst

E[log(Wt )]

4.8
4.75
4.7

4.65
4.6

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

Time
Fig. 3. Comparison

6 The Performances of the Strategies Based on

Misspecified Models
6.1 Introduction
In practice, it is extremely difficult to know parameters exactly. If one may
hope to calibrate 1 and relatively well owing to historical data, the value
of 2 cannot be determined a priori (i.e. before the occurrence of the drift
change), and the law of cannot be calibrated accurately because of the lack
of data concerning .
Consider a trader who believes that the stock price is


dSt = St 2 + (1 2 )1(t ) dt + St dBt ,
(21)

Technical Analysis Techniques versus Mathematical Models

13

where the law of is exponential with parameter . We suppose that the

true stock price is given by (2). Our aim is to study the misspecified optimal
allocation strategy and the misspecified model and detect strategy.
Notation. As above we set Rt = log(St ), where St is the actual price. We
define Lt and F t as follows:


 
1
1
2
2
Lt = exp
(

)R

)
+
2(

)(

)
t ,
t
2
1
2
1
1
1
2
2 2
2 2
R
1
t
et Lt 0 es Ls ds
Ft =
.
Rt
1
1 + et Lt 0 es Ls ds
6.2 On the Misspecified Optimal Allocation Strategy
Observing the stock price St , the trader computes a pseudo optimal allocation
by using the erroneous parameters 1 , 2 , and . Thus the value of his/her
misspecified optimal allocation strategy is
t = proj[0,1]

(1 r + (2 1 )F t )
,
2

and the corresponding wealth is

Wt

rt

Z

= e exp

u d(eru Su )

Numerical Example
In this section, we compare numerically the performance of two traders who
respectively use a misspecified model and the true model. We fix the value
of 1 = 0.2, = 0.15, r = 0.0 and = 2.0, and we assume that they
are perfectly known by the trader. A contrario 2 is misspecified. Its true
value is 2 = 0.2. Figure 4 shows the functions t E(log(Wt )) for three
values of 2 . It suggests that it is better to overestimate 2 (2 > 2 ) than
to underestimate it (2 < 2 ).
6.3 On Misspecified Model and Detect Strategies
The erroneous stopping rule is

Z t
K
1
= inf t 0, et Lt
es Ls ds
0

p
1 p

1
where p is the unique solution in ( , 1) of the equation
2
Z 1/2
Z p
(1 2s)e/s 2
(2s 1)e/s 2
s
ds =
s
ds
(1 s)2+
(1 s)2+
0
1/2

14

C. Blanchet-Scalliet, A. Diop, R. Gibson, D. Talay, E. Tanre

2 = 0.2
2 = 0.3
2 = 0.1

4.85

E[log(Wt )]

4.8
4.75
4.7
4.65
4.6

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

Time
Fig. 4. Error on 2 for the optimal trader

with = 2 2 /(2 1 )2 .
The value of the corresponding portfolio is
0
W T = xS
K

ST
+ xST0 1(K >T ) .
1 K
SK ( T )

6.4 A Comparison Between Misspecified Strategies and the

Technical Analysis Technique
Our main question is: Is it better to invest according to a mathematical strategy based on a misspecified model, or according to a strategy which does not
depend on any mathematical model? Because of the analytical complexity of
all the explicit formulae that we have obtained for the various expected utilities of wealth at maturity, we have not yet succeeded to find a mathematical
1
answer to this question, even in asymptotic cases (when 2
is large, e.g.).
2
As this part of our work is still in progress, we present here a few numerical
results obtained from Monte Carlo simulations. Consider the following study
case.
Parameters of the model
1 2 r
True values
0.2 0.2 2 0.15 0.0
Parameters used by the trader 1 2 r
Misspecified values (case I) 0.3 0.1 1.0 0.25 0.0
Misspecified values (case II) 0.3 0.1 3.0 0.25 0.0
Figure 5 shows that the technical analyst overperforms misspecified optimal allocation strategies when the parameter is underestimated.
We have looked for other cases where the technical analyst is able to
overperform the misspecified optimal allocation strategies. Consider the case
where the true values of the parameters are in Table 1. Table 2 summarizes
our results. It must be read as follows. For the misspecified values 2 = 0.1,

Technical Analysis Techniques versus Mathematical Models

15

4.85

MSP Case II
Technical
Analyst

E[log(Wt )]

4.8
4.75

MSP Case I
4.7

4.65
4.6

0.2

0.4

0.6

0.8

Time

1.2

1.4

1.6

1.8

Fig. 5. A technical analyst may overperform misspecified optimal allocation strategies

= 0.25, = 1, if the trader chooses 1 in the interval (0.5, 0.05) then

the misspecified optimal strategy is worse than the technical analysts one.
In fact, other numerical studies show that a single misspecified parameter is
not sufficient to allow the technical analyst to overperform the Model and
Detect traders. Astonishingly, other simulations show that the technical anTable 1. True values of the parameters
Parameter True Value
1
-0.2
2
0.2

0.15

Table 2. Misspecified values and range of the parameters

1 (-0.5,-0.05)
2
0.1

0.25

1 -0.3
2 (0,0.13)
0.25

1 -0.3
2 0.1
(0.2,)

1 -0.3
2 0.1
0.25
(0,1.5)

alyst may overperform the misspecified optimal allocation strategy but not
the misspecified model and detect strategy. One can also observe that, when
2 /1 decreases, the performances of well specified and misspecified model
and detect strategies decrease. See [BSDG+ 05].

16

C. Blanchet-Scalliet, A. Diop, R. Gibson, D. Talay, E. Tanre

7 Conclusions and Remarks

We have compared strategies designed from possibly misspecified mathematical models and strategies designed from technical analysis techniques. We
have made explicit the traders expected logarithmic utility of wealth in all
the cases under study. Unfortunately, the explicit formulae are not propitious
to mathematical comparisons. Therefore we have used Monte Carlo numerical experiments, and observed from these experiments that technical analysis
techniques may overperform mathematical techniques in the case of severe
misspecifications. Our study also brings some information on the range of
misspecifications for which this observation holds true.
Jointly with M. Martinez (INRIA) and S. Rubenthaler (University of Nice
Sophia Antipolis) we are now considering the infinite time case where the
instantaneous expected rate of return of the stock changes at the jump times
of a Poisson process and the values after each change time are unknown. We
also plan to consider technical analysis techniques different from the moving
average considered here.

Acknowledgment
This research is part of the Swiss national science foundation research program
NCCR Finrisk which has funded A. Diop during her postdoc studies at INRIA
and University of Z
urich.

References
[Ach00]
[BS02]

S. Achelis. Technical Analysis from A to Z. McGraw Hill, 2000.

A. N. Borodin and P. Salminen. Handbook of Brownian MotionFacts
and Formulae. Probability and its Applications. Birkh
auser Verlag,
Basel, second edition, 2002.
[BSDG+ 05] C.
Blanchet-Scalliet,
A.
Diop,
R.
Gibson,
D.
Talay,
E. Tanre, and K. Kaminski.
Technical Analysis Compared
to Mathematical Models Based Methods Under Misspecification.
NCCR-FINRISK Working Paper Series 253, 2005.
http://www.nccr-finrisk.unizh.ch/wp/index.php
[Kar03]
I. Karatzas. A note on Bayesian detection of change-points with an
expected miss criterion. Statist. Decisions, 21(1):313, 2003.
[KS98]
I. Karatzas and S. E. Shreve. Methods of Mathematical Finance, volume 39 of Applications of Mathematics. Springer-Verlag, New York,
1998.
[LMW00] A. W. Lo, H. Mamaysky, and J. Wang. Foundations of technical analysis: Computational algorithms, statistical inference, and empirical implementation. Journal of Finance, LV(4):17051770, 2000.
[Mer71]
R. C. Merton.
Optimum consumption and portfolio rules in a
continuous-time model. J. Econom. Theory, 3(4):373413, 1971.

Technical Analysis Techniques versus Mathematical Models

[Shi63]
[Shi02]

[Yor01]

17

A. N. Shiryaev. On optimum methods in quickest detection problems.

Theory Probab. Applications, 8:2246, 1963.
A. N. Shiryaev. Quickest detection problems in the technical analysis of
the financial data. In Mathematical FinanceBachelier Congress, 2000
(Paris), Springer Finance, pages 487521. Springer, Berlin, 2002.
M. Yor. Exponential Functionals of Brownian Motion and Related Processes. Springer Finance. Springer, Berlin, 2001.