MAR KET ANALYSIS · MAR KET ANALYSIS Charles B. Epstein, Editor Applying Chaos Theory to Investment...
Transcript of MAR KET ANALYSIS · MAR KET ANALYSIS Charles B. Epstein, Editor Applying Chaos Theory to Investment...
-
WIL
EY
FIN
AN
CE
ED
ITIO
NS
POR
TFO
LIO
MA
NA
GE
ME
NT
FO
RM
UL
AS
Ral
ph V
ince
TR
AD
ING
AN
D I
N V
EST
ING
IN
BO
ND
OPT
ION
SM
. Ant
hony
Won
g
FRA
CT
AL
MA
R K
ET
AN
ALY
SIS
Cha
rles
B. E
pste
in, E
dito
rA
pply
ing
Cha
os T
heor
y to
Inve
stm
ent
AN
ALY
ZIN
GA
ND
FO
RE
CA
STIN
G F
UT
UR
ES
PRIC
ES
Ant
hony
F. H
erbs
tan
dE
cono
mic
sC
HA
OS
AN
D O
RD
ER
IN
TH
E C
API
TA
L M
AR
KE
TS
Edg
arE
.Pe
ters
____
____
____
____
____
____
____
____
____
____
____
____
____
____
____
____
___
INSI
DE
TH
E F
INA
NC
IAL
FU
TU
RE
S M
AR
KE
TS,
3R
D E
DIT
ION
Mar
k J.
Pow
ers
and
Mar
k G
. Cas
telin
oR
EL
AT
IVE
DIV
IDE
ND
YIE
LD
Edg
arE
. Pet
ers
Ant
hony
E. S
pare
SEL
LIN
G S
HO
RT
Jose
ph A
. Wal
ker
TR
EA
SUR
Y O
PER
AT
ION
S A
ND
TH
E F
OR
EIG
N E
XC
HA
NG
E C
HA
LL
EN
GE
Dim
itris
N. C
hora
fas
TH
E F
OR
EIG
N E
XC
HA
NG
E A
ND
MO
NE
Y M
AR
KE
TS
GU
IDE
Julia
n W
alm
sley
CO
RPO
RA
TE
FIN
AN
CIA
L R
ISK
MA
NA
GE
ME
NT
Dia
ne B
. Wun
nick
e, D
avid
R. W
ilson
, Bro
oke
Wun
nick
eM
ON
EY
MA
NA
GE
ME
NT
ST
RA
TE
GIE
S FO
R F
UT
UR
ES
TR
AD
ER
SN
auze
r J.
Bal
sara
TH
E M
AT
HE
MA
TIC
S O
F M
ON
EY
MA
NA
GE
ME
NT
Ral
ph V
ince
TH
E N
EW
TE
CH
NO
LO
GY
OF
FIN
AN
CIA
L M
AN
AG
EM
EN
TD
imitr
is N
. Cho
rafa
sT
HE
DA
Y T
RA
DE
R'S
MA
NU
AL
Will
iam
F. E
ngO
PTIO
N M
AR
KE
T M
AK
ING
Alle
n J.
Bai
rdT
RA
DIN
G F
OR
A L
IVIN
GD
r. A
lexa
nder
Eld
erC
OR
POR
AT
E F
INA
NC
IAL
DIS
TR
ESS
AN
D B
AN
KR
UPT
CY
, SE
CO
ND
ED
ITIO
NE
dwar
d I.
Altm
anFI
XE
D.I
NC
OM
E A
RB
ITR
AG
EM
. Ant
hony
Won
gT
RA
DIN
G A
PPL
ICA
TIO
NS
OF
JAPA
NE
SE C
AN
DL
EST
ICK
CH
AR
TIN
GG
ary
S. W
agne
r an
d B
rad
L. M
athe
nyFR
AC
TA
L M
AR
KE
T A
NA
LY
SIS:
APP
LY
ING
CH
AO
S T
HE
OR
Y T
O I
NV
EST
ME
NT
AN
D E
CO
NO
MIC
SE
dgar
E. P
eter
sU
ND
ER
STA
ND
ING
SW
APS
John
F. M
arsh
all a
nd K
enne
th R
. Kap
ner
JOH
NW
ILE
Y &
SON
S,IN
C.
GE
NE
NT
ICA
LG
OR
ITH
MS
AN
D I
NV
EST
ME
NT
ST
RA
TE
GIE
SR
icha
rd J
Bau
er, J
rN
ew Y
ork
•C
hich
este
r•
Bris
bane
•T
oron
to•
Sin
gapo
re
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
Thi
s te
xt is
prin
ted
on a
cid-
free
pap
er.
Cop
yrig
ht ©
199
4 by
Joh
n W
iley
& S
ons,
Inc.
All
right
s re
serv
ed. P
ublis
hed
sim
ulta
neou
sly
in C
anad
a.
Rep
rodu
ctio
n or
tran
slat
ion
of a
ny p
art
of th
is w
ork
beyo
nd
that
per
mitt
ed b
y S
ectio
n 10
701
108
of th
e 19
76 U
nite
d
Sta
tes
Cop
yrig
ht A
ct w
ithou
t the
perm
issi
on o
f the
cop
yrig
bt
owne
r is
unl
awfu
l. R
eque
sts
for
perm
issi
on o
r fu
rthe
r
info
rmat
ion
shou
ld b
e ad
dres
sed
to th
eP
erm
issi
ons
Dep
artm
ent,
John
Wile
y &
Son
s, In
c., 6
05 T
hird
Ave
nue,
New
Yor
k, N
Y
1015
8-00
12.
Thi
s pu
blic
atio
n is
des
igne
d to
pro
vide
acc
urat
ean
d
auth
orita
tive
info
rmat
ion
in r
egar
d to
the
subj
ect
mat
ter
cove
red.
It is
sol
d w
ithth
e un
ders
tand
ing
that
the
publ
ishe
r is
not
eng
aged
inre
nder
ing
lega
l, ac
coun
ting,
or o
ther
pro
fess
iona
l ser
vice
s.If
lega
l adv
ice
or o
ther
expe
rt a
ssis
tanc
e is
req
uire
d, th
ese
rvic
es o
f a c
ompe
tent
prof
essi
onal
per
son
shou
ld b
e so
ught
. Fro
m a
Dec
lara
tion
ofPr
inci
ples
join
tly a
dopt
ed b
y a
Com
mitt
ee o
fthe
Am
eric
an B
ar A
ssoc
iatio
n an
d a
Com
mitt
eeof
Pub
lishe
rs.
Lib
rary
of
Con
gres
s C
atal
ogin
g-in
-Pub
licat
ion
Dat
a:
Pete
rs,E
dgar
E.,
1952
—F
ract
al m
arke
t ana
lysi
sap
plyi
ng c
haos
theo
ry to
inve
stm
ent a
nd
econ
omic
s / E
dgar
E. P
eter
s.
p.cm
.
Incl
udes
inde
x.IS
8N 0
-471
-585
24-6
I. In
vest
men
ts—
Mat
hem
atic
s.2.
Fra
ctal
s.3.
Cha
otic
beh
avio
r in
syst
ems.
I. T
itle.
II. T
itle:
Cha
os th
eory
.
HG
45I5
.3.P
4719
94
332.
6015
l474
—dc
2O93
-285
98
Prin
ted
in th
e U
nite
d S
tate
s of
Am
eric
a
10 9
8 7
6 5
4 3
2
To
Sher
yl
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
Pre
face
In19
91, I
fin
ishe
d w
ritin
g a
book
ent
itled
, Cha
osan
d O
rder
in th
e C
apita
l
Mar
kets
. It
was
pub
lishe
d in
the
Fall
of th
at y
ear
(Pet
ers,
199
Ia).
My
goal
was
to w
rite
a c
once
ptua
l int
rodu
ctio
n, f
or th
ein
vest
men
t com
mun
ity, t
o ch
aos
the-
ory
and
frac
tal s
tatis
tics.
I a
lso
wan
ted
to p
rese
nt s
ome
prel
imin
ary
evid
ence
that
, con
trar
y to
acc
epte
d th
eory
, mar
kets
are
not
wel
l-de
scri
bed
by th
e ra
n-
dom
wal
k m
odel
, and
the
wid
ely
taug
ht E
ffic
ient
Mar
ketH
ypO
thes
is (
EM
H)
is
not w
ell-
supp
orte
d by
em
piri
cal e
vide
nce.
I ha
ve r
ecei
ved,
in g
ener
al, a
ver
y po
sitiv
e re
spon
se to
that
boo
k. M
any
read
ers
have
com
mun
icat
ed th
eir
appr
oval
—an
d so
me,
thei
rdis
appr
oval
—an
d
have
ask
ed d
etai
led
ques
tions
. The
que
stio
ns f
ell i
nto
two
cate
gori
es: (
I) te
ch-
nica
l, an
d (2
) co
ncep
tual
. In
the
tech
nica
l cat
egor
y w
ere
the
requ
ests
for
mor
e
deta
il ab
out t
he a
naly
sis.
My
book
had
not
bee
n in
tend
ed to
be
ate
xtbo
ok, a
nd
I ha
d gl
osse
d ov
er m
any
tech
nica
l det
ails
invo
lved
in th
ean
alys
is. T
his
ap-
proa
ch im
prov
ed th
e re
adab
ility
of
the
book
, but
it le
ft m
any
read
ers
won
der-
ing
how
to p
roce
ed.
In th
e se
cond
cat
egor
y w
ere
ques
tions
con
cern
ed w
ith c
once
ptua
lis
sues
. If
the
EM
H is
fla
wed
, how
can
we
fix
it? O
r be
tter
still
, wha
t is
avi
able
rep
lace
-
men
t? H
ow d
o ch
aos
theo
ry a
nd f
ract
als
fit i
n w
ith tr
adin
g st
rate
gies
and
with
the
dich
otom
y be
twee
n te
chni
cal a
nd f
unda
men
tal a
naly
sis?
Can
thes
e se
em-
ingl
y di
spar
ate
theo
ries
be
unite
d? C
an tr
aditi
onal
theo
ry b
ecom
eno
nlin
ear?
In th
is b
ook,
Lam
add
ress
ing
both
cat
egor
ies
of q
uest
ions
. Thi
s bo
okis
dif
fer-
ent f
rom
the
prev
ious
one
, but
it r
efle
cts
man
ysi
mila
r fe
atur
es. F
ract
al M
arke
t
Ana
lysi
s is
an
atte
mpt
to g
ener
aliz
e C
apita
l Mar
ket T
heor
y (C
MT
)an
d to
ac-
coun
t for
the
dive
rsity
of
the
inve
stm
ent c
omm
unity
.O
ne o
f th
e fa
iling
s of
trad
i-
tiona
l the
ory
is it
s at
tem
pt to
sim
plif
y "t
he m
arke
t" in
to a
n av
erag
epr
otot
ypic
al VI'
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
viii
Pre
face
Pre
face
—ix
ratio
nal i
nves
tor.
The
rea
sons
for
setti
ng o
ut o
n th
is r
oute
wer
e no
ble.
In th
e tr
adi-
tion
of W
este
rn s
cien
ce, t
he fo
undi
ng fa
ther
s of
CM
T a
ttem
pted
to le
arn
som
e-th
ing
abou
t the
who
le b
y br
eaki
ng d
own
the
prob
lem
into
its
basi
c co
mpo
nent
s.T
hat a
ttem
pt w
as s
ucce
ssfu
l. B
ecau
se o
f the
fars
ight
ed w
ork
of M
arko
witz
,S
harp
e, F
ama,
and
oth
ers,
we
have
mad
e en
orm
ous
prog
ress
ove
r th
e pa
st 4
0 ye
ars.
How
ever
, the
red
uctio
nist
app
roac
h ha
s its
lim
its, a
nd w
e ha
ve r
each
ed th
em.
It is
tim
e to
take
a m
ore
holis
tic v
iew
of h
ow m
arke
ts o
pera
te. I
n pa
rtic
ular
, it i
stim
e to
rec
ogni
ze th
e gr
eat d
iver
sity
that
und
erlie
s m
arke
ts. A
ll in
vest
ors
do n
otpa
rtic
ipat
e fo
r th
e sa
me
reas
on, n
or d
o th
ey w
ork
thei
r st
rate
gies
ove
r th
e sa
me
inve
stm
ent h
oriz
ons.
The
sta
bilit
y of
mar
kets
is in
evita
bly
tied
to th
e di
vers
ityof
the
inve
stor
s. A
mat
ure"
mar
ket i
s di
vers
e as
wel
l as
old.
If a
ll th
e pa
rtic
i-pa
nts
had
the
sam
e in
vest
men
t hor
izon
, rea
cted
equ
ally
to th
e sa
me
info
rmat
ion,
and
inve
sted
for
the
sam
e pu
rpos
e, in
stab
ility
wou
ld r
eign
. Ins
tead
, ove
r th
e lo
ngte
rm, m
atur
e m
arke
is h
ave
rem
arka
ble
stab
ility
. A d
ay tr
ader
can
trad
e an
ony-
mou
sly
with
a p
ensi
on fu
nd: t
he fo
rmer
trad
es fr
eque
ntly
for
shor
t-te
rm g
ains
;th
e la
tter
trad
es in
freq
uent
ly fo
r lo
ng-t
erm
fina
ncia
l sec
urity
. The
day
trad
er r
e-ac
ts to
tech
nica
l tre
nds;
the
pens
ion
fund
inve
sts
base
d on
long
-ter
m e
cono
mic
grow
th p
oten
tial.
Yet
, eac
h pa
rtic
ipat
es s
imul
tane
ousl
y an
d ea
ch d
iver
sifie
s th
eot
her.
The
red
uctio
nist
app
roac
h, w
ith it
s ra
tiona
l inv
esto
r, c
anno
t han
dle
this
dive
rsity
with
out c
ompl
icat
ed m
ultip
art m
odel
s th
at r
esem
ble
a R
ube
Gol
dber
gco
ntra
ptio
n. T
hese
mod
els,
with
thei
r m
ultip
le li
miti
ng a
ssum
ptio
ns a
nd r
estr
ic-
tive
requ
irem
ents
, ine
vita
bly
fail.
The
y ar
e so
com
plex
that
they
lack
flex
ibili
ty,
and
flexi
bilit
y is
cru
cial
to a
ny d
ynam
ic s
yste
m.
The
firs
t pur
pose
of t
his
book
is to
intr
oduc
e th
e F
ract
al M
arke
t Hyp
othe
sis—
a ba
sic
refo
rmul
atio
n of
how
, and
why
, mar
kets
func
tion.
The
sec
ond
purp
ose
ofth
e bo
ok is
to p
rese
nt to
ols
for
anal
yzin
g m
arke
ts w
ithin
the
frac
tal f
ram
ewor
k.M
any
exis
ting
tool
s ca
n be
use
d fo
r th
is p
urpo
se. I
will
pre
sent
new
tool
s to
add
to
the
anal
yst's
tool
box,
and
will
rev
iew
exi
stin
g on
es.
Thi
s bo
ok is
not
a n
arra
tive,
alth
ough
its
prim
ary
emph
asis
is s
till c
once
p-tu
al. W
ithin
the
conc
eptu
al fr
amew
ork,
ther
e is
a r
igor
ous
cove
rage
of a
naly
ti-ca
l tec
hniq
ues.
As
in m
y pr
evio
us b
ook,
I be
lieve
that
any
one
with
a fi
rmgr
ound
ing
in b
usin
ess
stat
istic
s w
ill fi
nd m
uch
that
is u
sefu
l her
e. T
he p
rimar
yem
phas
is is
not
on
dyna
mic
s, b
ut o
n em
piric
al s
tatis
tics,
that
is, o
n an
alyz
ing
time
serie
s to
iden
tify
wha
t we
are
deal
ing
with
.
TH
E S
TR
UC
TU
RE
OF
TH
E B
OO
K
The
boo
k is
div
ided
into
five
par
ts, p
lus
appe
ndic
es. T
he fi
nal a
ppen
dix
con-
tain
s fr
acta
l dis
trib
utio
n ta
bles
. Oth
er r
elev
ant t
able
s, a
nd fi
gure
s co
ordi
nate
d
to th
e di
scus
sion
, are
inte
rspe
rsed
inth
e te
xt. E
ach
part
bui
lds
on th
e pr
evio
us
part
s, b
ut th
e bo
ok c
an b
e re
adno
nseq
uent
ially
by
thos
e fa
mili
ar w
ith th
e co
n-
cept
s of
the
first
boo
k.
Par
t One
: Fra
ctal
Tim
e S
erie
s
Cha
pter
1 in
trod
uces
frac
tal t
ime
serie
s an
d de
fines
both
spa
tial a
nd te
mpo
ral
frac
tals
. The
re is
a p
artic
ular
em
phas
is o
n w
hat
frac
tals
are
, con
cept
ually
and
phys
ical
ly. W
hy d
o th
ey s
eem
cou
nter
intu
itive
, eve
n th
ough
frac
tal g
eom
etry
is
muc
h cl
oser
to th
e re
al w
orld
than
the
Euc
lidea
n ge
omet
ry w
eal
l lea
rned
in
high
sch
ool?
Cha
pter
2 is
a b
rief r
evie
w o
f Cap
ital
Mar
ket T
heor
y (C
MT
) an
d
of th
e ev
iden
ce o
f pro
blem
s w
ith th
e th
eory
. Cha
pter
3 is
, in
man
y w
ays,
the
hear
t of t
he b
ook:
I de
tail
the
Fra
ctal
Mar
ket
Hyp
othe
sis
as a
n al
tern
ativ
e to
the
trad
ition
al th
eory
dis
cuss
ed in
Cha
pter
2.
As
a F
ract
alM
arke
tH
ypot
hesi
s,
it co
mbi
nes
elem
ents
of f
ract
als
from
Cha
pter
1w
ith p
arts
of t
radi
tiona
l CM
T
in C
hapt
er 2
. The
Fra
ctal
Mar
ket H
ypot
hesi
s se
tsth
e co
ncep
tual
fram
ewor
k
for
frac
tal m
arke
t ana
lysi
s.
Par
t Tw
o: F
ract
al (
R/S
) A
naly
sis
Hav
ing
defin
ed th
e pr
oble
m in
Par
t One
, I o
ffer
tool
s fo
r an
alys
is in
Par
t
Tw
o—in
par
ticul
ar, r
esca
led
rang
e (R
IS)
anal
ysis
.M
any
of th
e te
chni
cal
ques
tions
I re
ceiv
ed a
bout
the
first
boo
k de
alt w
ithR
/S a
naly
sis
and
re-
ques
ted
deta
ils a
bout
cal
cula
tions
and
sig
nific
ance
test
s.P
arts
Tw
o an
d
Thr
ee a
ddre
ss th
ose
issu
es. R
/S a
naly
sis
is a
rob
ust
anal
ysis
tech
niqu
e fo
r un
-
cove
ring
long
mem
ory
effe
cts,
frac
tal s
tatis
tical
str
uctu
re,
and
the
pres
ence
of c
ycle
s. C
hapt
er 4
sur
veys
the
conc
eptu
al b
ackg
roun
dof
R/S
ana
lysi
s an
d
deta
ils h
ow to
app
ly it
. Cha
pter
5 g
ives
bot
h st
atis
tical
test
sfo
r ju
dgin
g th
e
sign
ifica
nce
of th
e re
sults
and
exa
mpl
es o
f how
R/S
ana
lysi
s re
acts
tokn
own
stoc
hast
ic m
odel
s. C
hapt
er 6
sho
ws
how
R/S
ana
lysi
s ca
n be
used
to u
ncov
er
both
per
iodi
c an
d no
nper
iodi
c cy
cles
.
Par
t Thr
ee: A
pply
ing
Fra
ctal
Ana
lysi
s
Thr
ough
a n
umbe
r of
cas
e st
udie
s, P
art T
hree
det
ails
how
R/S
anal
ysis
tech
-
niqu
es c
an b
e us
ed. T
he s
tudi
es, i
nter
estin
g in
thei
r ow
nrig
ht, h
ave
been
se-
lect
ed to
illu
stra
te th
e ad
vant
ages
and
dis
adva
ntag
es o
fusi
ng R
IS a
naly
sis
on
diffe
rent
type
s of
tim
e se
ries
and
diffe
rent
mar
kets
. Alo
ng th
e w
ay,
inte
rest
ing
thin
gs w
ill b
e re
veal
ed a
bout
tick
dat
a, m
arke
t vol
atili
ty, a
ndho
w c
urre
ncie
s ar
e
diffe
rent
from
oth
er m
arke
ts.
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
Pre
face
Pre
face
Par
t Fou
r: F
ract
al N
oise
Hav
ing
used
R/S
ana
lysi
s to
fin
d ev
iden
ce to
sup
port
the
Frac
tal M
arke
t Hy-
poth
esis
, 1 s
uppl
y m
odel
s to
exp
lain
thos
e fi
ndin
gs. P
art F
our
appr
oach
es m
arke
tac
tivity
fro
m th
e vi
ewpo
int o
f st
ocha
stic
pro
cess
es; a
s su
ch, i
t con
cent
rate
s on
frac
tal n
oise
. In
Cha
pter
13,
usi
ng R
/S a
naly
sis,
dif
fere
nt "
colo
red"
noi
ses
are
anal
yzed
and
com
pare
d to
the
mar
ket a
naly
sis.
The
fin
ding
s ar
e re
mar
kabl
ysi
mila
r. I
n ad
ditio
n, th
e be
havi
or o
f vo
latil
ity is
giv
en a
sig
nifi
cant
exp
lana
tion.
Cha
pter
14
disc
usse
s th
e st
atis
tics
of f
ract
al n
oise
pro
cess
es, a
nd o
ffer
s th
em a
san
alte
rnat
ive
to th
e tr
aditi
onal
Gau
ssia
n no
rmal
dist
ribu
tion.
The
impa
ct o
ffr
acta
l dis
trib
utio
ns o
n m
arke
t mod
els
is d
iscu
ssed
. Cha
pter
15
show
s th
e im
-pa
ct o
f fr
acta
l sta
tistic
s on
the
port
folio
sel
ectio
n pr
oble
m a
ndop
tion
pric
ing.
Met
hods
for
ada
ptin
g th
ose
mod
els
for
frac
tal d
istr
ibut
ions
are
rev
iew
ed.
Part
Fou
r is
a v
ery
deta
iled
sect
ion
and
will
not
be
appr
opri
ate
for
all r
eade
rs.
How
ever
, bec
ause
the
appl
icat
ion
of tr
aditi
onal
CM
T h
as b
ecom
e in
grai
ned
into
mos
t of
the
inve
stm
ent c
omm
unity
, I b
elie
ve th
at m
ost r
eade
rssh
ould
rea
d th
esu
mm
ary
sect
ions
of
each
cha
pter
, if
noth
ing
else
, in
Part
Fou
r.C
hapt
er 1
3, w
ith
its s
tudy
of
the
natu
re o
f vo
latil
ity, s
houl
d be
of
part
icul
ar in
tere
st.
Whi
le r
eadi
ng th
e bo
ok, m
any
of y
ou w
ill w
onde
r,w
here
is th
is le
adin
g?
Will
this
hel
p m
e m
ake
mon
ey?
Thi
s bo
ok d
oes
not
offe
r ne
w tr
adin
g te
ch-
niqu
es o
r fi
nd p
ocke
ts o
f in
effi
cien
cy th
at th
e sa
vvy
inve
stor
can
pro
fit f
rom
.
It is
not
a b
ook
of s
trat
egy
for
mak
ing
bette
rpr
edic
tions
. Ins
tead
, it o
ffer
s a
new
vie
w o
f ho
w m
arke
ts w
ork
and
how
to te
st ti
me
seri
es f
or p
redi
ctab
ility
.
Mor
e im
port
antly
, it g
ives
add
ition
al in
form
atio
nab
out t
he r
isks
inve
stor
s
take
, and
how
thos
e ri
sks
chan
ge o
ver
time.
If
know
ledg
e is
pow
er, a
s th
e ol
d
clic
hé g
oes,
then
the
info
rmat
ion
here
sho
uld
beco
nduc
ive,
if n
ot to
pow
er, a
t
leas
t to
bette
r pr
ofits
.
Con
cord
, Mas
sach
uset
ts
ED
GA
R E
. PE
TE
RS
Par
t Fiv
e: N
oisy
Cha
os
Par
tFi
ve o
ffer
s a
dyna
mic
al s
yste
ms
alte
rnat
ive
to th
e st
ocha
stic
pro
cess
es o
fPa
rt F
our.
In
part
icul
ar, i
t off
ers
nois
y ch
aos
as a
pos
sibl
e ex
plan
atio
n of
the
frac
-ta
l str
uctu
re o
f m
arke
ts. C
hapt
er 1
6, w
hich
giv
es R
/S a
naly
sis
of c
haot
ic s
ys-
tem
s, r
evea
ls r
emar
kabl
e si
mila
ritie
s w
ith m
arke
t and
oth
er ti
me
seri
es. A
part
icul
ar e
mph
asis
is p
lace
d on
dis
tingu
ishi
ng b
etw
een
frac
tal n
oise
and
noi
sych
aos.
A r
evie
w is
giv
en o
f th
e B
DS
(Bro
ck—
Dec
hert
—Sc
hein
kman
) te
st, w
hich
,w
hen
used
in c
onju
nctio
n w
ith R
/S a
naly
sis,
can
giv
e co
nclu
sive
evi
denc
ew
ay o
r th
e ot
her.
Cha
pter
17
appl
ies
frac
tal s
tatis
tics
to n
oisy
chao
s, r
econ
cilin
gth
e tw
o ap
proa
ches
. An
expl
anat
ion
is o
ffer
ed f
or w
hy e
vide
nce
of b
oth
frac
tal
nois
e an
d no
isy
chao
s ca
n ap
pear
sim
ulta
neou
sly.
The
res
ult i
s cl
osel
y tie
d to
the
Frac
tal M
arke
t Hyp
othe
sis
and
the
theo
ry o
f m
ultip
le in
vest
men
t hor
izon
s.C
hapt
er 1
8 is
a r
evie
w o
f th
e fi
ndin
gs o
n a
conc
eptu
al le
vel.
Thi
s fi
nal
chap
ter
unite
s th
e Fr
acta
l Mar
ket H
ypot
hesi
s w
ith th
e em
piri
cal w
ork
and
theo
retic
al m
odel
s pr
esen
ted
thro
ugho
ut th
e bo
ok. F
or r
eade
rs w
ho u
nder
-st
and
a pr
oble
m b
ette
r w
hen
they
kno
w th
e so
lutio
n, it
may
be
appr
opri
ate
tore
ad C
hapt
er 1
8 fi
rst.
The
app
endi
ces
offe
r so
ftw
are
that
can
be
used
for
ana
lysi
s an
d re
prod
uce
tabl
es o
f th
e fr
acta
l dis
trib
utio
ns.
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
L
Ack
now
ledg
men
ts
Iw
ould
like
to th
ank
the
follo
win
g pe
ople
for
thei
rinv
alua
ble
advi
ce a
nd a
ssis
-
tanc
e: F
rom
Pan
Ago
ra A
sset
Man
agem
ent:
Ric
hard
Cro
wel
l, Pe
ter
Rat
hjen
s,
John
Lew
is, B
ruce
Cla
rke,
Ter
ry N
orm
an, A
lan
Bro
wn,
and
Cliv
e L
ang.
Als
o
War
ren
Spro
ul, E
arl K
eefe
r, R
ober
t Mel
len,
Gui
doD
eBoe
ck, a
nd R
on B
rand
es
for
help
, ide
as, a
nd r
efer
ence
s. T
hank
s al
so to
Kri
stin
e ("
with
a K
") L
ino,
Dav
id A
rrig
hini
, Jim
Rul
lo, a
nd C
huck
LeV
ine
ofth
e Pa
nAgo
ra A
sset
Allo
ca-
tion
team
, for
thei
r in
dulg
ence
and
hel
p.I
wou
ld a
lso
like
to th
ank,
bel
ated
ly, m
y or
igin
alW
iley
edito
r, W
endy
Gra
u,
who
per
suad
ed m
e to
wri
te b
oth
book
s an
d sa
w m
eth
roug
h th
e fi
rst p
ublic
a-
tion.
I w
ould
als
o lik
e to
than
k M
yles
Tho
mps
on, m
y cu
rren
tedi
tor,
for
see
ing
me
thro
ugh
this
one
.Fi
nally
, I w
ould
onc
e ag
ain
like
to th
ank
my
wif
e,Sh
eryl
, and
my
child
ren,
Ian
and
Luc
ia, f
or th
eir
cont
inue
d su
ppor
t dur
ing
the
man
y ho
urs
I ne
eded
to
mys
elf
to c
ompl
ete
this
pro
ject
.
E.E
.P.
XII
I
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
L
Con
tent
s
PA
RT
ON
E F
RA
CT
AL
TIM
E S
ER
IES
Intr
oduc
tion
to F
ract
al T
ime
Ser
ies
3
Fra
ctal
Spac
e, 4
Frac
tal T
ime,
5Fr
acta
l Mat
hem
atic
s, 9
The
Cha
os G
ame,
10
Wha
t Is
a Fr
acta
l?, 1
2T
he F
ract
al D
imen
sion
, 15
Frac
tal M
arke
t Ana
lysi
s, 1
7
2Fa
ilure
of th
e G
auss
ian
Hyp
othe
sis
18
Cap
ital
Mar
ket T
heor
y, 1
9St
atis
tical
Cha
ract
eris
tics
of M
arke
ts, 2
1T
he T
erm
Str
uctu
re o
f V
olat
ility
, 27
The
Bou
nded
Set
, 37
Sum
mar
y, 3
8
3A
Fra
ctal
Mar
ket H
ypot
hesi
s39
Effi
cien
tM
arke
ts R
evis
ited,
39
Stab
le M
arke
ts v
ersu
s E
ffic
ient
Mar
kets
, 41
The
Sou
rce
of L
iqui
dity
, 42
Info
rmat
ion
Sets
and
Inv
estm
ent H
oriz
ons,
43
Stat
istic
al C
hara
cter
istic
s of
Mar
kets
, Rev
isite
d, 4
3T
he F
ract
al M
arke
t Hyp
othe
sis,
44
Sum
mar
y, 4
9
xv
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
xvi
Con
tent
sC
onte
nts
Xvi
i
PA
RT
TW
O F
RA
CT
AL
(R/S
) A
NA
LYS
IS10
Vol
atili
ty: A
Stu
dy in
Ant
iper
sist
ence
143
Rea
lized
Vol
atili
ty, 1
46
4M
easu
ring
Mem
ory—
The
Hur
st P
roce
ss a
nd R
/S A
naly
sis
53Im
plie
dV
olat
ility
, 148
Bac
kgro
und:
Dev
elop
men
t of
R/S
Ana
lysi
s, 5
4Su
mm
ary,
150
The
Jok
er E
ffec
t, 60
11P
robl
ems
with
Und
ersa
mpl
ing:
Gol
d an
d U
.K. I
nfla
tion
151
Ran
dom
ness
and
Pers
iste
nce:
Int
erpr
etin
g th
e H
urst
Exp
onen
t, 61
R/S
Ana
lysi
s: A
Ste
p-by
-Ste
p G
uide
, 61
Typ
e I
Und
ersa
mpl
ing:
Too
Litt
le T
ime.
152
An
Exa
mpl
e: T
he Y
en/D
olla
r E
xcha
nge
Rat
e, 6
3T
ype
II U
nder
sam
plin
g: T
oo L
ow a
Fre
quen
cy, 1
54T
wo
Inco
nclu
sive
Stu
dies
, 154
5T
estin
gR
/S A
naly
sis
65S
umm
ary,
158
The
Ran
dom
Nul
l Hyp
othe
sis,
66
12C
urre
ncie
s:A
Tru
e H
urst
Pro
cess
159
Sto
chas
ticM
odel
s, 7
5Su
mm
ary,
85
The
Dat
a, 1
60Y
en/D
olla
r, 1
61M
ark/
Dol
lar,
163
6F
indi
ng C
ycle
s: P
erio
dic
and
Non
perio
dic
86P
ound
/Dol
lar,
163
Peri
odic
Cyc
les,
88
Yen
/Pou
nd, 1
65
Non
peri
odic
Cyc
les,
93
Sum
mar
y. 1
66
Sum
mar
y, 1
02PA
RT
FO
UR
FR
AC
TA
L N
OIS
E
PA
RT
TH
RE
E A
PP
LYIN
G F
RA
CT
AL
AN
ALY
SIS
13F
ract
iona
l Noi
se a
nd R
/S A
naly
sis
169
7C
ase
Stu
dy M
etho
dolo
gy10
7T
he C
olor
of N
oise
, 170
Met
hodo
logy
, 108
Pink
Noi
se: 0
<H
<0.
50, 1
72
Dat
a, 1
09B
lack
Noi
se: 0
.50<
H <
1.0,
183
Stab
ility
Ana
lysi
s, 1
10T
he M
irro
r E
ffec
t, 18
7Fr
actio
nal D
iffe
renc
ing:
AR
FIM
A M
odel
s, 1
88
8D
owJo
nes
Indu
stria
ls, 1
888—
1990
: An
Idea
l Dat
a S
et11
2S
umm
ary,
196
Num
ber
of O
bser
vatio
ns v
ersu
s L
engt
h of
Tim
e, 1
1214
Frac
tal
Sta
tistic
s19
7
Tw
enty
-Day
Ret
urns
, 113
Frac
tal (
Stab
le)
Dis
trib
utio
ns, 1
99Fi
ve-D
ay R
etur
ns, 1
17St
abili
ty u
nder
Add
ition
, 206
Dai
ly R
etur
ns, 1
19C
hara
cter
istic
s of
Fra
ctal
Dis
trib
utio
ns, 2
08St
abili
ty A
naly
sis,
125
Mea
suri
ng a
, 209
Raw
Dat
a an
d Se
rial
Cor
rela
tion,
127
Mea
suri
ng P
roba
bilit
ies,
214
Sum
mar
y, 1
30In
fini
te D
ivis
ibili
ty a
nd I
GA
RC
H, 2
14Su
mm
ary,
216
9S
&P
500
Tic
k D
ata,
198
9—19
92: P
robl
ems
with
Ove
rsam
plin
g13
215
App
lyin
g F
ract
al S
tatis
tics
217
The
Una
djus
ted
Dat
a, 1
33Po
rtfo
lio S
elec
tion,
218
The
AR
(l)
Res
idua
ls, 1
37O
ptio
n V
alua
tion,
224
Impl
icat
ions
. 141
Sum
mar
y, 2
32
L
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
xviii
Con
tent
s
PA
RT
FIV
EN
oisy
Cha
os
16N
oisy
Cha
os a
nd R
/S A
naly
sis
235
Info
rmat
ion
and
Inve
stor
s, 2
37C
haos
, 239
App
lyin
g R
/S A
naly
sis,
241
Dis
tingu
ishi
ng N
oisy
Cha
os f
rom
Fra
ctio
nal N
oise
, 246
FRA
CT
AL
TIM
E S
ER
IES
17Fr
acta
lS
tatis
tics,
Noi
sy C
haos
, and
the
FM
H25
2
Fre
quen
cyD
istr
ibut
ions
, 253
Vol
atili
ty T
erm
Str
uctu
re, 2
57Se
quen
tial S
tand
ard
Dev
iatio
n an
d M
ean,
259
Mea
suri
ng a
, 263
The
Lik
eiih
ood
of N
oisy
Cha
os, 2
64O
rbita
l Cyc
les,
265
Self
-Sim
ilari
ty, 2
68A
Pro
posa
l: U
nitin
g G
AR
CH
, FB
M, a
nd C
haos
, 270
18U
nder
stan
ding
Mar
kets
271
Info
rmat
ion
and
Inve
stm
ent H
oriz
ons,
272
Stab
ility
, 272
Ris
k, 2
73L
ong
Mem
ory,
274
Cyc
les,
274
Vol
atili
ty, 2
75T
owar
d a
Mor
e C
ompl
ete
Mar
ket T
heor
y, 2
75
App
endi
x1:
The
Cha
os G
ame
277
App
endi
x2:
GA
US
S P
rogr
ams
279
App
endi
x 3:
Fra
ctal
Dis
trib
utio
n T
able
s28
7
Bib
liogr
aphy
296
Glo
ssar
y30
6
Inde
x31
3
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
L
1 Intr
oduc
tion
to F
ract
alT
ime
Ser
ies
Wes
tern
cultu
re h
as lo
ng b
een
obse
ssed
by
the
smoo
th a
ndsy
mm
etri
c. N
ot a
ll
cultu
res
are
sim
ilarl
y ob
sess
ed, b
ut th
e W
est (
mea
ning
Eur
opea
n de
rive
d) h
as
long
reg
arde
d pe
rfec
t for
ms
as s
ymm
etri
c, s
moo
th,
and
who
le. W
e lo
ok f
or
patte
rns
and
sym
met
ry e
very
whe
re. O
ften
, we
impo
se p
atte
rns
whe
re n
one
ex-
ists
, and
we
deny
pat
tern
s th
at d
o no
t con
form
to o
urov
eral
l con
cept
ual f
ram
e-
wor
k. T
hat i
s, w
hen
patte
rns
are
not s
ymm
etri
cal
and
smoo
th, w
e cl
assi
fy
them
as
illus
ions
.T
his
conf
lict c
an b
e tr
aced
bac
k to
the
anci
ent G
reek
s.T
o de
scri
be o
ur
phys
ical
wor
ld, t
hey
crea
ted
a ge
omet
ry b
ased
on
pure
,sym
met
ric,
and
sm
ooth
form
s. P
lato
sai
d th
at th
e "r
eal"
wor
ld c
onsi
sted
of
thes
esh
apes
. The
se f
orm
s
wer
e cr
eate
d by
a f
orce
, or
entit
y,ca
lled
the
"Goo
d."
The
wor
ld o
f th
e G
ood
coul
d be
glim
psed
onl
y oc
casi
onal
ly, t
hrou
gh th
em
ind.
The
wor
ld w
e in
habi
t
is a
n im
perf
ect c
opy
of th
e re
al w
orld
, and
was
cre
ated
by a
dif
fere
nt e
ntity
,
calle
d th
e "D
emiu
rge.
" T
he D
emiu
rge,
a le
sser
bein
g th
an th
e G
ood,
was
doom
ed to
cre
ate
infe
rior
cop
ies
of th
e re
al w
orld
.T
hese
cop
ies
wer
e ro
ugh,
asym
met
ric,
and
sub
ject
to d
ecay
. In
this
way
, Pla
tore
conc
iled
the
inab
ility
of
the
Gre
ek g
eom
etry
, lat
er f
orm
aliz
ed b
y E
uclid
, to
desc
ribe
our
wor
ld. T
he
prob
lem
was
not
with
the
geom
etry
, but
with
our
wor
ldits
elf.
3
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
4In
trod
uctio
n to
Fra
ctal
Tim
e S
erie
sF
ract
al T
ime
5
FR
AC
TA
L S
PA
CE
Fra
ctal
geom
etry
is th
e ge
omet
ry o
f th
eD
emiu
rge.
Unl
ike
Euc
lidea
n ge
ome-
try,
it th
rive
s on
rou
ghne
ssan
d as
ymm
etry
. Obj
ects
are
not
vari
atio
ns o
n a
few
perf
ect a
nd s
ymm
etri
cal f
orm
s,bu
t are
infi
nite
ly c
ompl
ex. T
he m
ore
clos
ely
they
are
exa
min
ed, t
he m
ore
deta
il is
rev
eale
d. F
or e
xam
ple,
a tr
ee is
afr
acta
l
form
. Im
agin
e a
pine
tree
like
the
Dou
glas
fir
, com
mon
ly u
sed
for
Chr
istm
as
tree
s. C
hild
ren
ofte
n dr
awD
ougl
as f
irs
as tr
iang
les
(the
bra
nche
s)w
ith
rect
angu
lar
base
s (t
he tr
ee tr
unks
),gi
ving
the
tree
s as
muc
h sy
mm
etry
as
pos-
sibl
e. L
ogos
of
Chr
istm
as tr
ees
have
the
sam
e ap
pear
ance
or
may
subs
titut
e
cone
s fo
r th
e tr
iang
les.
Yet
, Dou
glas
fir
s ar
e no
t tri
angl
es o
r co
nes.
The
y ar
e a
netw
ork
of b
ranc
hes
qual
itativ
ely
sim
ilar
to th
e sh
ape
of th
e ov
eral
l tre
e,bu
t
each
indi
vidu
al b
ranc
h is
dif
fere
nt.
The
bra
nche
s on
bra
nche
s (s
ucce
ssiv
e ge
n-
erat
ions
of
bran
ches
) be
com
epr
ogre
ssiv
ely
smal
ler.
Yet
, with
in e
ach
gene
ra-
tion
ther
e is
act
ually
a r
ange
of
size
s. A
nd, e
ach
tree
is d
iffe
rent
.
Euc
lidea
n ge
omet
ry c
anno
t rep
licat
e a
tree
.U
sing
Euc
lidea
n ge
omet
ry, w
e
can
crea
te a
nap
prox
imat
ion
of a
tree
, but
it a
lway
slo
oks
artif
icia
l, lik
e a
child
's d
raw
ing
or a
logo
. Euc
lidea
n ge
omet
ryre
crea
tes
the
perc
eive
d sy
mm
e-
try
of th
e tr
ee, b
ut n
ot th
eva
riet
y th
at a
ctua
lly b
uild
s its
str
uctu
re.
Und
erly
ing
this
per
ceiv
ed s
ymm
etry
is a
cont
rolle
d ra
ndom
ness
, and
incr
easi
ngco
mpl
ex-
ity a
t fin
er le
vels
of
reso
lutio
n. T
his
"sel
f-si
mila
r" q
ualit
y is
the
defi
ning
cha
r-
acte
rist
ic o
f fr
acta
ls. M
ost n
atur
al s
truc
ture
s,pa
rtic
ular
ly li
ving
thin
gs, h
ave
this
cha
ract
eris
tic.
A s
econ
d pr
oble
m, w
hen
we
appl
yE
uclid
ean
geom
etry
to o
ur w
orld
, is
one
of d
imen
sion
ality
. We
live
in a
thre
e-di
men
sion
al s
pace
, but
onl
y so
lidfo
rms
are
trul
y th
ree-
dim
ensi
onal
,ac
cord
ing
to th
e de
fini
tions
that
are
the
basi
s of
Euc
lidea
n ge
omet
ry. I
n m
athe
mat
ical
term
s, a
nob
ject
mus
t be
diff
eren
tiabl
e
acro
ss it
s en
tire
surf
ace.
Aw
iffl
e ba
ll, f
or in
stan
ce, i
s no
t ath
ree-
dim
ensi
tmal
obje
ct, a
lthou
gh it
res
ides
in a
thre
e-di
men
sion
al s
pace
.
In a
dditi
on, o
ur p
erce
ptio
n of
dim
ensi
on c
an c
hang
e, d
epen
ding
on
our
dis-
tanc
e fr
om a
n ob
ject
. Fro
m a
dist
ance
, a D
ougl
as f
ir lo
oks
like
atw
o-di
men
sion
al
tria
ngle
. As
we
com
e cl
oser
, it a
ppea
rs a
s a
thre
e-di
men
sion
al c
one.
Clo
ser
still
,
we
can
see
its b
ranc
hes,
and
it lo
oks
like
a ne
twor
k of
one-
dim
ensi
onal
line
s.
Clo
ser
exam
inat
ion
reve
als
the
bran
ches
as
thre
e-di
men
sion
al tu
bes.
Euc
lidea
n
geom
etry
als
o ha
s di
ffic
ulty
with
the
dim
ensi
onal
ity o
f cr
eatio
nsof
the
Dem
iurg
e an
d w
ith in
crea
sing
com
plex
ity.
By
cont
rast
, Euc
lidea
n st
ruct
ures
be-
com
e si
mpl
er a
t sm
alle
r an
dsm
alle
r sc
ales
. The
thre
e-di
men
sion
also
lid r
e-
duce
s to
a tw
o-di
men
sion
al p
lane
. The
two-
dim
ensi
onal
pla
ne is
mad
e up
of
one-
dim
ensi
onal
line
s an
d, f
inal
ly, n
ondi
men
sion
alpo
ints
. Our
per
cept
ion
of th
e
tree
, on
the
othe
r ha
nd, w
ent
from
two-
dim
ensi
onal
toth
ree-
dim
ensi
onal
to
one-
dim
ensi
onal
, and
bac
k to
thre
e-di
men
sion
al.
Thi
s is
dif
fere
nt f
rom
the
Euc
lidea
n pe
rcep
tion.
Euc
lidea
n ge
omet
ry is
onl
y us
eful
as
a gr
oss
sim
plif
icat
ion
of th
e w
orld
of
the
Dem
iurg
e. F
ract
al g
eom
etry
, by
cont
rast
,is
char
acte
rize
d by
sel
f-si
mila
rity
and
incr
ease
d co
mpl
exity
und
erm
agni
fica
tion.
Its
maj
or a
pplic
atio
n as
a g
eom
e-
try
of s
pace
has
bee
n in
gen
erat
ing
real
istic
look
ing
land
scap
es v
ia c
ompu
ters
.
The
Der
niur
ge c
reat
ed n
ot o
nly
frac
tal s
pace
but f
ract
al ti
me
as w
ell.
Al-
thou
gh o
ur p
rim
ary
focu
s w
ill b
e on
frac
tal t
ime
seri
es, f
ract
al s
pace
will
help
us u
nder
stan
d fr
acta
l tim
e.W
e w
ill s
ee th
e di
ffer
ence
bet
wee
nth
e sm
ooth
ness
of th
e E
uclid
ean
wor
ld a
nd th
e ro
ughn
ess
of o
ur w
orld
, whi
ch li
mits
the
usef
ul-
ness
of
Euc
lid's
geo
met
ry a
s a
met
hod
of d
escr
iptio
n.
FRA
CT
AL
TIM
E
Thi
sco
nflic
t bet
wee
n th
e sy
mm
etry
of
Euc
lidea
nge
omet
ry a
nd th
e as
ymm
e-
try
of th
e re
al w
orld
can
be
furt
her
exte
nded
to o
ur c
once
pt o
f tim
e.T
radi
tion-
ally
, eve
nts
are
view
ed a
s ei
ther
ran
dom
or
dete
rmin
istic
. In
the
dete
rmin
istic
view
, all
even
ts th
roug
h tim
e ha
ve b
een
fixe
d fr
om th
e m
omen
t of
crea
tion.
Thi
s vi
ew h
as b
een
give
n a
theo
logi
cal
basi
s by
den
omin
atio
ns s
uch
as th
e
Cal
vini
sts,
and
sci
entif
ic e
ndor
sem
ent b
yce
rtai
n "b
ig b
ang"
theo
rist
s. I
n co
n-
tras
t, ni
hilis
t gro
ups
cons
ider
all
even
tsto
be
rand
om, d
eriv
ing
from
no
stru
c-
ture
or
orde
r th
roug
h tim
e.In
fra
ctal
tim
e, r
ando
mne
ss a
ndde
term
inis
m, c
haos
and
ord
er c
oexi
st. I
n
frac
tal s
hape
s, w
e se
e a
phys
ical
repr
esen
tatio
n of
how
thes
e op
posi
tes
wor
k
toge
ther
. The
pin
e tr
ee h
as g
loba
l str
uctu
rean
d lo
cal r
ando
mne
ss. I
n ge
nera
l,
we
know
wha
t a p
ine
tree
look
s lik
e, a
nd w
e ca
n pr
edic
t the
gen
eral
org
loba
l
shap
e of
any
pin
e tr
ee w
ith a
hig
h de
gree
of a
ccur
acy.
How
ever
, at t
he in
divi
d-
ual b
ranc
h le
vel,
each
bra
nch
isdi
ffer
ent.
We
do n
ot k
now
how
long
it is
, or
its
diam
eter
. Eac
h tr
ee is
dif
fere
nt, b
ut s
hare
sce
rtai
n gl
obal
pro
pert
ies.
Eac
h ha
s
loca
l ran
dom
ness
and
glo
bal d
eter
min
ism
.In
this
sec
tion,
we
will
exa
min
e
how
the
conc
ept o
f fr
acta
l tim
e ev
olve
d,an
d w
hat i
t mea
ns.
Mos
t cul
ture
s fa
vor
the
dete
rmin
istic
view
of
time.
We
like
to th
ink
that
we
have
a p
lace
in th
eun
iver
se, t
hat w
e ha
ve a
des
tiny.
Yet
, we
see
rand
om,
cata
stro
phic
eve
nts
that
can
thw
art u
s fr
omfu
lfill
ing
our
purp
ose.
Nat
ural
dis
-
aste
rs c
an d
estr
oy o
ur e
nvir
onm
ent.
Eco
nom
ic d
isas
ters
can
take
wha
t we
own.
Indi
vidu
als
we
do n
ot k
now
can
rob
us
of o
urliv
es, b
y ac
cide
nt o
r w
ith m
alic
e.
Con
vers
ely,
goo
d fo
rtun
e ca
n ar
rive
by
bein
g at
the
righ
t pla
ce a
t the
rig
ht
time.
A c
hanc
e m
eetin
g ca
n op
en n
ew d
oors
.Pi
ckin
g th
e ri
ght l
otte
ry n
umbe
rs
can
brin
g us
a f
ortu
ne.
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
6
Gre
atev
ents
als
o se
em to
res
t on
chan
ce. N
ewto
n sa
w a
n ob
ject
fal
ling
(leg
-en
d sa
ys it
was
an
appl
e) a
nd f
orm
ulat
ed th
e ca
lcul
us a
nd th
e la
ws
of g
ravi
ty.
Flem
ing
left
a p
etri
dis
h ex
pose
d an
d di
scov
ered
pen
icill
in. D
arw
in d
ecid
ed to
go o
n a
voya
ge a
nd f
orm
ulat
ed a
theo
ry o
f ev
olut
ion
beca
use
of h
is e
xper
ienc
es
on th
e jo
urne
y.T
hese
and
sim
ilar
even
ts s
eem
ed to
hap
pen
by c
hanc
e, a
nd th
ey c
hang
ed h
is-
tory
. Yet
,th
e ca
lcul
us in
depe
nden
tly o
f N
ewto
n, a
t alm
ost t
he
sam
e tim
e—in
fac
t, w
e us
e L
iebn
ez's
not
atio
n. W
alla
cede
velo
ped
the
theo
ry o
fna
tura
l sel
ectio
n in
depe
nden
tly o
f D
arw
in, t
houg
h la
ter.
Bec
ause
of
a pa
per
byW
alla
ce, D
arw
in f
ound
the
ener
gy to
wri
te O
rigi
n of
Spe
cies
so
he w
ould
rec
eive
cred
it as
the
theo
ry's
ori
gina
l dev
elop
er. I
n ou
r ow
n fi
eld
of f
inan
cial
eco
nom
ics,
wha
t is
know
n as
the
Cap
ital A
sset
Pri
cing
Mod
el (
CA
PM)
was
dev
elop
ed in
de-
pend
ently
by
no f
ewer
than
thre
e pe
ople
—Sh
arpe
(19
64),
Lin
tner
(19
65),
and
Mos
sin
(196
6)—
at a
lmos
t the
sam
e tim
e. T
his
wou
ld im
ply
that
thes
e di
scov
erie
sw
ere
mea
nt to
hap
pen.
His
tory
dem
ande
d it.
It w
as th
eir
dest
iny.
It h
as b
een
diff
icul
t to
reco
ncile
ran
dom
ness
and
ord
er, c
hanc
e an
d ne
ces-
sity
, or
free
will
and
det
erm
inis
m. I
s th
is d
icho
tom
y on
ce a
gain
the
Dem
iurg
eim
perf
ectly
cop
ying
the
Goo
d?E
vent
s ar
e pe
rcei
ved
as e
ither
ran
dom
, and
ther
efor
e un
pred
icta
ble,
or
de-
term
inis
tic, a
nd p
erfe
ctly
pre
dict
able
. Unt
il th
e be
ginn
ing
of th
is c
entu
ry, i
tw
as g
ener
ally
acc
epte
d th
at th
e un
iver
se r
an li
ke a
clo
ck. E
vent
ually
,sc
ient
ists
wer
e to
dis
cove
r th
e eq
uatio
ns u
nder
lyin
g th
e un
iver
se, a
nd b
ecom
eab
le to
pre
-
dict
its
even
tual
cou
rse.
Tim
e w
as o
f no
con
sequ
ence
in N
ewto
nian
mec
hani
cs;
theo
retic
ally
, tim
e co
uld
be r
ever
sed,
bec
ause
New
ton'
s eq
uatio
ns w
orke
d fi
new
heth
er ti
me
ran
forw
ard
or b
ackw
ard.
Irr
ever
sibl
e tim
e, th
e fi
rst b
low
to th
isde
term
inis
tic v
iew
, cam
e in
the
mid
-I9t
h ce
ntur
y fr
om th
e em
ergi
ng f
ield
of
ther
mod
ynam
ics.
The
rmod
ynam
ics
bega
n as
the
stud
y of
hea
t was
te p
rodu
ced
by m
achi
nes.
It
was
som
e tim
e be
fore
ther
mod
ynam
ics,
an
appl
ied
scie
nce,
was
take
n se
riou
sly
by s
cien
tists
. The
initi
al s
tudy
foc
used
on
how
ene
rgy
is c
onve
rted
into
use
ful
wor
k. I
n a
syst
em s
uch
as th
e st
eam
eng
ine,
ste
am tu
rns
whe
els
and
perf
orm
sa
func
tion
such
as
pow
erin
g a
boat
's p
addl
e w
heel
. Not
all
the
ener
gypr
oduc
edis
con
vert
ed in
to w
ork.
Som
e is
lost
or
is d
issi
pate
d as
fri
ctio
n. T
he s
tudy
of
thes
e "d
issi
pativ
e sy
stem
s" e
vent
ually
gre
w to
incl
ude
flui
d dy
nam
ics.
Flu
iddy
nam
ics,
whi
ch in
vest
igat
ed th
e he
atin
g an
d m
ixin
g of
flu
ids,
gav
e us
tim
e-
irre
vers
ible
sys
tem
s.Su
ppos
e tw
o lit
ers
of f
luid
are
sep
arat
ed b
y a
wat
erpr
oof,
rem
ovab
le p
arti-
tion.
On
one
side
is a
lite
r of
red
flu
id; o
n th
e ot
her,
a li
ter
of b
lue.
We
deci
deto
use
the
term
ent
ropy
as
the
mea
sure
of
the
degr
ee o
f m
ixin
g of
the
red
and
blue
flu
ids.
As
long
as
the
part
ition
is in
pla
ce, w
eha
ve lo
w e
ntro
py. I
f w
elif
t
the
part
ition
, the
red
and
blue
flu
ids
will
flo
w in
to o
nean
othe
r, a
nd th
e le
vel o
f
entr
opy
will
ris
e as
the
flui
ds b
ecom
e m
ore
mix
ed. E
vent
ually
, whe
nth
e re
d
and
blue
bec
ome
thor
ough
lym
ixed
, all
of th
e fl
uid
will
beco
me
purp
le.
Whe
n fu
lly m
ixed
, the
flu
idha
s re
ache
d a
stat
e of
equi
libri
um. I
t can
not
beco
me
"mor
e m
ixed
." I
tha
s re
ache
d a
leve
l of
max
imum
ent
ropy
. How
ever
,
we
cann
ot "
unm
ix"
the
flui
d. D
espi
te th
e fa
ctth
at th
e m
ixin
g of
the
flui
ds is
unde
rsla
ndab
le in
dyn
amic
al te
rms,
it is
tim
e-de
pend
ent a
ndir
reve
rsib
le. T
he
flui
d's
stat
e of
hig
h en
trop
y, o
run
cert
aint
y, w
hich
com
esfr
om th
e m
axim
um
mix
ing
of tw
o st
ates
(in
this
cas
e, th
e st
ates
are
labe
led
"red
" an
d "b
lue"
),
cann
ot b
e de
scri
bed
bytim
e-re
vers
ible
, New
toni
aneq
uatio
ns. T
he f
luid
will
neve
r be
com
eun
mix
ed; i
ts e
ntro
py w
ill n
ever
decl
ine,
eve
n if
we
wai
t for
ete
r-
nity
. In
ther
mod
ynam
ics,
time
has
an a
rrow
that
poin
ts o
nly
tow
ard
the
futu
re.
The
fir
st b
low
had
bee
nst
ruck
aga
inst
the
cloc
kwor
kvi
ew o
f th
e un
iver
se.
The
sec
ond
blow
cam
e w
ithth
e em
erge
nce
of q
uant
umm
echa
nics
. The
re-
aliz
atio
n th
at th
e m
olec
ular
str
uctu
reof
the
univ
erse
can
be
desc
ribe
d on
ly b
y
stat
es o
f pr
obab
ility
furt
her
unde
rmin
ed th
ede
term
inis
tic v
iew
. But
con
fusi
on
rem
aine
d. W
as th
e un
iver
sede
term
inis
tic o
r ra
ndom
?
Slow
ly, i
t has
bec
ome
appa
rent
that
mos
t nat
ural
sys
tem
s ar
ech
arac
teri
zed
by
loca
l ran
dom
ness
and
glo
bal
dete
rmin
ism
. The
se c
ontr
ary
stat
esm
ust c
oexi
st.
Det
erm
inis
m g
ives
us
natu
ral l
aw.R
ando
mne
ss in
duce
s in
nova
tion
and
vari
ety.
A h
ealth
y, e
volv
ing
syst
emis
one
that
not
onl
y ca
nsu
rviv
e ra
ndom
sho
cks,
but
also
can
abs
orb
thos
e sh
ocks
toim
prov
e th
e ov
eral
l sys
tem
,w
hen
appr
opri
ate.
.
For
inst
ance
, it h
as b
een
post
ulat
ed b
y W
est a
nd G
oldb
erge
r(1
987)
that
phy
s-
ical
fra
ctal
str
uctu
res
are
gene
rate
d by
nat
ure
beca
use
they
are
mor
e er
ror-
tole
rant
than
sym
met
rica
l str
uctu
res
in th
eir
crea
tion.
Tak
eth
e m
amm
alia
n
lung
. Its
mai
n br
anch
, the
trac
hea,
div
ides
into
two
subb
ranc
hes.
The
se tw
o
halv
es c
ontin
ue b
ranc
hing
. At
each
bra
nchi
ng g
ener
atio
n,th
e av
erag
e di
amet
er
decr
ease
s ac
cord
ing
to a
pow
erla
w. T
hus,
the
diam
eter
of
each
gen
erat
ion
is d
e-
pend
ent o
n th
e di
amet
ers
of th
e pr
evio
us g
ener
atio
n.In
add
ition
, eac
h br
anch
gene
ratio
n ac
tual
ly h
as a
ran
geof
dia
met
ers
with
in it
. The
ave
rage
diam
eter
of
each
gen
erat
ion
scal
es d
own
acco
rdin
g to
a p
ower
law
, but
any
indi
vidu
al b
ranc
h
can
be d
escr
ibed
only
pro
babi
listic
ally
. We
have
glo
bal d
eter
min
ism
(the
ave
r-
age
bran
ch s
ize)
and
loca
l ran
dom
ness
(th
edi
amet
er o
f in
divi
dual
bra
nche
s).
Why
doe
s na
ture
fav
or th
isst
ruct
ure,
whi
ch a
ppea
rsin
all
mam
mal
ian
lung
s?
Wes
t and
Gol
dber
ger
have
show
n th
at th
is f
ract
al s
truc
ture
is m
ore
stab
le a
nd
erro
r-to
lera
nt th
an o
ther
str
uctu
res.
Rem
embe
r th
at e
ach
bran
chge
nera
tion
is
depe
nden
t on
the
gene
ratio
nsbe
fore
it. I
f di
amet
ers
scal
edex
pone
ntia
lly, n
ot
only
wou
ld a
n er
ror
in th
efo
rmat
ion
of o
ne g
ener
atio
n af
fect
the
next
bra
nchi
ng
Intr
oduc
tion
to F
ract
al T
ime
Ser
ies
Fra
ctal
Tim
e7
PDF compression, OCR, web-optimization with CVISION's PdfCompressor
http://www.cvisiontech.com/pdf_compressor_31.html
-
8In
trod
uctio
n to
Fra
ctal
Tim
e S
erie
sfr
acta
l Mat
hem
atic
s-
gene
ratio
n, b
ut th
e er
ror
wou
ld g
row
with
eac
h su
cces
sive
gen
erat
ion.
A s
mal
ler
ror
mig
ht c
ause
the
lung
to b
ecom
e m
alfo
rmed
and
non
func
tiona
l. H
owev
er,
by f
ract
al s
calin
g, th
e er
ror
has
less
impa
ct b
ecau
se o
f th
e po
wer
law
, as
wel
l as
the
loca
l pro
babi
listic
str
uctu
re. B
ecau
se e
ach
gene
ratio
n ha
s a
rang
e of
dia
me-
ters
, one
mal
form
ed b
ranc
h ha
s le
ss im
pact
on
the
form
atio
n of
the
othe
rs.
Thu
s, th
e fr
acta
l str
uctu
re (
glob
al d
eter
min
ism
, loc
al r
ando
mne
ss)
is m
ore
erro
r-to
lera
nt d
urin
g fo
rmat
ion
than
oth
er s
truc
ture
s.T
o lo
ok a
head
, if
we
chan
ge th
is c
once
pt f
rom
a s
tatic
str
uctu
re (
the
lung
) to
a dy
nam
ic s
truc
ture
like
the
stoc
k m
arke
t, w
e ca
n m
ake
som
e in
tere
stin
g co
n-je
ctur
es. C
hang
e br
anch
gen
erat
ion
to "
inve
stm
ent h
oriz
on."
The
sto
ck m
arke
tis
mad
e up
of
inve
stor
s, f
rom
tick
trad
ers
to lo
ng-t
erm
inve
stor
s. E
ach
has
adi
ffer
ent i
nves
tmen
t hor
izon
that
can
be
orde
red
in ti
me.
A s
tabl
e m
arke
t is
one
in w
hich
all
the
part
icip
ants
can
trad
e w
ith o
ne a
noth
er, e
ach
faci
ng a
ris
kle
vel l
ike
the
othe
rs',
adju
sted
for
thei
r tim
e sc
ale
or in
vest
men
t hor
izon
. We
will
see
in C
hapt
er 2
that
the
freq
uenc
y di
stri
butio
n of
ret
urns
is th
e sa
me
for
day
trad
ers
as it
is f
or o
ne-w
eek
or e
ven
90-d
ay r
etur
ns, o
nce
an a
djus
tmen
t is
mad
e fo
r sc
ale.
Tha
t is,
fiv
e-m
inut
e tr
ader
s fa
ce th
e sa
me
risk
of
a la
rge
even
tas
doe
s a
wee
kly
trad
er. I
f da
y tr
ader
s ha
ve a
cra
sh a
t the
ir ti
me
scal
e, li
ke a
four
-sig
ma
even
t, th
e m
arke
t rem
ains
sta
ble
if th
e ot
her
trad
ers,
who
hav
e di
f-fe
rent
trad
ing
hori
zons
, see
that
cra
sh a
s a
buyi
ng o
ppor
tuni
ty, a
nd s
tep
in a
ndbu
y. T
hus,
the
mar
ket r
emai
ns s
tabl
e be
caus
e it
has
no c
hara
cter
istic
tim
esc
ale,
just
as
the
lung
had
no
char
acte
rist
ic d
iam
eter
sca
le. W
hen
the
mar
ket's
entir
e in
vest
men
t hor
izon
sho
rten
s, a
nd e
very
one
beco
mes
a o
ne-m
inut
e tr
ader
(inv
esto
rs h
ave
lost
thei
r fa
ith in
long
-ter
m in
form