Professional Documents
Culture Documents
J. T. Manhire
Texas A&M University, College Station, Texas
Abstract
This paper explores the possibility that asset prices, especially those traded in large volume on
public exchanges, might comply with specific physical laws of motion and probability. The paper
first examines the basic dynamics of price displacements in financial markets and finds one can
model this dynamic as a harmonic oscillator at local “slices” of elapsed time via a homogeneous
coordinate system. Price and time coordinates here are Wigner invariant meeting the threshold
of mathematical law discovery in markets. Based on this finding, the paper theorizes that price
displacements are constrained, meaning they have extreme values beyond which they cannot
go for defined periods. By showing that price displacements also comply with the principle of
stationary action, the paper explores a method for measuring specific probabilities of future price
displacements based on prior historical data. Testing this theory with limited markets suggests it
can make forecasts consistent with historical data as to constraints on extreme price movements
during market “crashes” and probabilities of specific price displacements at other times.
1
2 J. T. MANHIRE
An initial postulate follows from an obvious of the action principle to the asset’s movement
implication of the paper’s objective: through the price dimension. Section 5 derives
I. By analogy, financial markets comply with, probability measures from the dynamics dis-
even if they are not necessarily subject to, certain cussed. Section 6 attempts to test the theory
physical laws. by comparing results obtained from traditional
If one does not assume this, the entire enter- analyses with those obtained solely from the
prise would be without logical motivation. theoretical model. Initial comparisons with his-
Mantegna and Stanley [6] define a financial torical asset price data appear promising. Sec-
market as a system fueled by a large number tion 7 closes the paper with a brief summary
of interactions between buyers and sellers in and recommendations for future research.
perpetual search for an appropriate price for
an asset. Broken down further, this definition
explains that a financial market is a system, the
2. Local Systems
mechanics of which relate solely to changes
We begin with first principles by observing fi-
in price and the sole drivers of these mechan-
nancial market data fixed in a tangible medium
ics are the interactions of buyers and sellers.
and asking, “What can we deduce about the
The remaining two postulates follow from this
most fundamental aspects of a market’s change
definition:
in price over some elapsed period?” The an-
II. Observed market mechanics are the result of
swer seems clear. The price of an asset—the
local systems that can be defined by their respective
object of the system we call a financial market—
changes in price and time.
goes up, down, or stays the same when com-
III. The only phenomenon that directly affects an paring the asset’s final and initial prices over
asset’s price displacement is the net effect generated some period. Seen this way, it’s reasonable to
by the trading activities of buyers and sellers at each talk about the asset as moving through price
instance of elapsed time. over time.2
From these postulates, the paper attempts to Imagine a single dimension of an inertial
follow logical conclusions to a theoretical end. frame of reference for an asset. This is not a
As shared in the final section, the theoretical spatial dimension in the traditional sense, but
results from these assumptions are consistent rather a dimension of price. An asset moves
with historical data—at least for the few as- through this price dimension consistent with
sets against which the theory has been tested. our third postulate.
While every attempt is made throughout this
Let’s explore the basic dynamics of an as-
paper to clearly convey the process that leads
set’s movement through this price dimension,
from observation to theory, there is not attempt
that is, its price displacement. Displacement in-
at true mathematical rigor.
volves moving something from one position to
The paper is presented in seven sections, the
another over some elapsed period. Therefore,
first being this introduction. Section 2 explores
to arrive at the most basic dynamic of an as-
the basic dynamics of an asset’s price displace-
set’s price displacement we must first try to
ment at each instance in time. Section 3 in-
understand the most basic instance of elapsed
vestigates the forces that drive this price dis-
time.
placement. Section 4 explores the application
2 We can also talk about the price moving around the
with historical market data then perhaps there is more here asset if we wish. After all, does the ship move through the
for others to investigate and philosophical questions with water or does the water move around the ship? In the end,
which others can grapple. the chosen reference frame seems more intuitive.
THE ACTION PRINCIPLE IN MARKET MECHANICS 3
note these averages, such as h·i , however, such notation is ters, price displacement and all functions of it are vector
unnecessarily cumbersome given this clarification of what quantities even though we use scalar notation throughout
we must consider an average and why. with direction signified by positive or negative signs.
4 J. T. MANHIRE
translation, even though prices and time them- conventional units. Because of this, we will re-
selves are not. If the changes in price and time fer to our coordinate system of changes in time
are homogeneous then they are also isotropic, and price that sets t A = x A = 0 and t = 1 as
meaning they are invariant under rotation. a “homogeneous coordinate system,” although
For these reasons, we can always make both the reader should be aware that others might
t B and x B equivalent to t and x, respectively, define this term differently in the literature [9].
if we regard t A and x A as zero in relation
to t B and x B , respectively, for any time slice. 3. External Forces
Therefore, in this coordinate system, we can
assume four price-time principles: (1) transla- At this point we should be able to imagine the
tional invariance, (2) rotational invariance, (3) price displacement of an asset as the net effect
time translational invariance, and (4) Galilean of the interactions of buyers and sellers. In fact,
invariance (the rules apply equally to all assets based on our postulates, these interactions are
even if each is in its own coordinate system).5 the only source of a net “force” that moves the
These principles are important because they asset through the price dimension over a time
are the four invariances Wigner claims neces- slice τ.
sary for the discovery of mathematical rules of Are these interactions sometimes the result
a system [7]. Without these invariances, math- of traders reacting to external events such as
ematical rules might exist for a system, but we a significant terrorist attack or news of a ma-
would be hard pressed to root them out. The jor corporate scandal? Of course, but neither
perceived lack of these invariances in economic the attack nor the scandal directly affect the
systems is a common criticism for applying mechanics of this system we call a financial
physical laws, even if by analogy, to financial market. It might be the individual reactions
markets [8]. of traders to these events that in the aggregate
As a consequence, we can represent the end affect the direction of the price displacement
state of any elapsed period with the coordi- over some elapsed period, but it is important
nates (t, x ) and that should be sufficient to de- to remember that financial markets have only
scribe the remaining dynamics occurring over one “prime mover” when we examine the sys-
the time interval in terms of time and price. tem qua system. That mover is the net force
The only difference in the coordinates is that generated by the interactions of traders. There
t will always be positive since time only moves is no other source of direct force in this theory.6
in one direction for our purposes here, while x We use the term “force” here only by anal-
might be positive, negative, or zero since prices ogy; however, we will assume the net force
move up, down, or experience no change. to have the same properties and produce the
In this way, our coordinate system becomes same effects on an asset in the dimension of
one of net changes from state A to state B since price as a physical force has on an object in the
we can always set ǫ to A and ǫ + 1 to B for any dimension of space. Recall that we assume fi-
slice τ regardless of the actual elapsed time in nancial markets comply with certain principles
of physics, even if we agree that financial mar-
5 It
is well known that Newton’s Second Law is invariant
under a Galilean transformation so we need not prove this kets are not subject to all physical laws (e.g.,
again here. By the end of this paper it should be clear gravity).
that the price displacement of any asset is subject to the
same rules regardless of its individual coordinate system 6 This theory does not disregard the uniqueness of each
with the inertial coefficient being the main determinant individual agent’s behavior. The difference is in focus.
of price displacement magnitude given equal net trading Instead of focusing on the individual agent, this theory
interactions. focuses on the interaction of individual agents.
6 J. T. MANHIRE
We know from experience that buyers and activities would have caused separately.
sellers interact with an asset and each other We can also assume that something “carries”
for each time slice τ that an exchange is open these forces to produce the observed price dis-
for trading.7 Let’s call a buyer’s activity α placement of an asset. For now, we will refer
since the activity is physically manifested in to this net “carrying thing” as ψ, where ψ is
what’s called an “ask” and a seller’s activity the result of the superposition of the “carry-
β since the activity is physically manifested in ing thing” of the buyers’ force and that of the
what’s called a “bid.” Because we’ve decided to sellers’.
make the up direction positive and the down Let’s return to our proportional assumption
direction negative, we can notate a buyer’s that F ∝ x. Since both F and x are time func-
activity as +α and a seller’s activity as − β. tions, we know per Newton’s second law that
We assume that the price displacement for F = m ẍ, where m can be thought of as an iner-
some period τ is proportional to the sum of tial coefficient unique to each asset and ẍ is the
the buyers’ and sellers’ activities, or second time derivative of the price displace-
ment. This leaves us with m ẍ ∝ x.
x ∝ α + (− β) = α − β,
If m ẍ displaces the asset in the price dimen-
sion over, let’s say, the period τ1 it means the
since these are the only activities that affect the
equivalent of t A is t1 and the asset moves from
asset’s movement through the price dimension.
price x1 to price x2 at t2 . For the sake of argu-
Consequently, if α = β then the actions cancel,
ment let’s assume the displacement over this
producing a price displacement x = 0 for that
slice is in the positive direction. Yet, at pre-
period. However, if α > β then x > 0, and if
cisely t2 the price x2 that was the close price
α < β then x < 0.
for the previous slice τ1 becomes the opening
More specifically, we can think of the inter-
price for the next slice τ2 . Our homogeneous
active dynamic α − β as generating some net
coordinate system is such that t A and x A can
force that is solely responsible for the price
always be considered zero since we care only
displacement we witness for the period. In
about the net displacement over the period.
other words, the net force is proportional to
For this reason, when x2 goes from being the
the change in price, or F ∝ x.
closing price for τ1 to being the opening price
Here, F is the net force resulting from the su-
for τ2 the asset must reclaim the value zero
perposition of Fα and Fβ where Fα is positively
since we regard the starting price for any slice
directed and Fβ is negatively directed, or
as zero.
F = Fα + (− Fβ ) = Fα − Fβ . By doing this, one of two things happen
(both of which are equivalent): either the as-
The system as we have defined it appears in set moves from the price x2 = x2 at the end
all respects to be linear. Therefore, it is appro- of τ1 to the price x2 = 0 at the start of τ2 , or
priate to assume that the net response of the the coordinate system moves to set x2 = 0 at
system to the interactions of traders is the sum the start of τ2 . Either way, the price displace-
of the responses that the buyers’ and sellers’ ment for the first slice, x (τ1 ), must be traversed
7 Some “penny” stocks might go for long periods with-
again in full, but in the opposite direction, so
out buyer or seller activity. The markets we examine here that x2 = 0 for the start of τ2 . This means if
are those with high trade volumes for all relevant times. m ẍ is the net force that displaces the asset in
The only reason for there to not be a trade in these markets
would be that the exchange is closed so the interactions
the positive direction over τ1 , there must be a
cannot occur. negatively-directed force of equal magnitude
THE ACTION PRINCIPLE IN MARKET MECHANICS 7
returning price x2 to zero for the start of τ2 . so that the positive time-axis is heading into the
We’ll express this negatively-directed force as page, the positive ℜ-axis is pointing to the top
kx. of the page, and the positive ℑ-axis is pointing
This turns our proportional statement m ẍ ∝ to the right of the page. We’re now observ-
x into ing only the two dimensions of the complex
m ẍ = −kx. (1) plane. This configuration is slightly unconven-
tional, but it maintains the visual of positive
Since m ẍ is a positively-directed force and kx is
price changes going “up” and negative changes
a negatively-directed force of equal magnitude,
going “down.”
we must add a negative sign to kx so that the
We find a complex solution for Eq. (1) in
two sides of the equation balance.
Quadrant I of the complex plane (upper right)
This is a very familiar equation. It describes
in the following form:
the motion of a simple harmonic oscillator.8
This is consistent with observations of a finan- r r !
k k
cial market from the perspective of the homo- ψB = R cos t + i sin t , (2)
m m
geneous coordinate system we have created.
At each time slice τ the basic dynamic is that where R is the modulus of ψB .
the asset moves through the price dimension in The principal value of the argument of the
one direction and then must return (or the co- complex number ψ at t B , or arg(ψB ), is φB .
ordinate system reset) to the equilibrium point Let’s call φǫ the phase of system, where the
for the next slice thereby covering the same phase is defined geometrically in the complex
distance in price. plane as the angle from the positive ℜ-axis to a
So far our coordinate system has the dimen- vector rotated toward the ℑ-axis. In our coordi-
sions of time and price. Yet, there are well- nate system, this means the phase is positively
known solutions to Eq. (1) in the complex directed if it moves clockwise from the +ℜ-axis
plane. If a solution to an equation of motion de- towards the +ℑ axis.9
rived from our observations of how a financial Reviewing the complex solution for Eq. (1)
market moves in price resides in the complex we hold that the phase φǫ is equivalent to
plane, we must assume that the complex plane √
t k/m. Let’s denote the change in the phase
is a necessary component of the lens through from t A to t B as
which we observe this real phenomenon even if
it ends up being an intermediate mathematical φ := ∆φǫ = φB − φ A .
contrivance.
To observe this solution we must add one Defined this way, at t A = 0 where x A = 0,
more dimension—an imaginary dimension—to φ A must always equal π2 . This means
our coordinate system. This third dimension is
π
represented by the imaginary-axis ℑ, thereby φ = φB − .
2
making our price dimension equivalent to the
real-axis ℜ. Let’s rotate the coordinate system Thus, we see that the principle value of the
phase displacement φ is always the comple-
8 More generally, the system can be expressed as a Van
either the total or very near total conservation of energy positively-directed phase moving counterclockwise. This is
with any net energy introduced into the system wholly simply a result of how one chooses to set up the coordinate
converted to motion in the price dimension. system, so the variation is irrelevant for our examination.
8 J. T. MANHIRE
x = R sin φ. (4) 1
x= ( ψ + ψ ∗ ). (5)
2
The real solution for the price displacement
We can say the same for x in the negative di-
in Eq. (4) is typically a result of our solving
rection using −ψ and −ψ∗ .
Eq. (1) and ignoring the imaginary parts of the
Given our definitions of ψ and ψ∗ we see
complex solution ψ in Eq. (3). Taken as a phys-
that
ical description, this approach implies that we
ψ + ψ∗ = 2R sin φ.
only observe one part of a much richer physical
reality, the imaginary part of which somehow Note that the imaginary terms i cos φ in ψ and
still lingers in existence but beyond our abil- −i cos φ in ψ∗ cancel when added. Substituting
ity to observe it. Such an approach certainly this sum into Eq. (5) we recover Eq. (4).
stirs the imagination to conceive of a “hidden By thinking of the price displacement
dimension” that exists beyond our senses yet as the one-dimensional median of the two-
is still responsible for the phenomena we expe- dimensional region between the wave func-
rience in real-world financial markets. tion and its complex conjugate, we arrive at
THE ACTION PRINCIPLE IN MARKET MECHANICS 9
the same solution using a mathematical con- superposition of things that carry the forces of
trivance that cancels the imaginary terms in- buyers and sellers with a resulting amplitude
stead of admitting a lingering dimension that of ±R. This is consistent with our assumption
physically exists but which we simply ignore. that some “carrying thing” is responsible for
It is not merely “more complete” to regard the transporting the net force generated by the
price displacement this way, it’s necessary. Sim- trading interactions of buyers and sellers that
ply ignoring the imaginary parts deprives us ultimately effects a price change for an asset.
of a fuller understanding of how the price dis- As with many things in the physical world, the
placement, the probabilities of these displace- thing that carries the force in our theory is akin
ments, and the action relate to one another. to a wave, or at least its analog.10
Because we are dealing with a sinusoid we Throughout this paper we are ultimately in-
can define the average frequency ν in cycles terested in the price displacement x, which we
per unit of elapsed time and the average phase know is equivalent to x B given that x A can al-
displacement per unit of elapsed time from ways be considered zero. Yet, ψ A cannot be
states A to B as ω, where considered zero since ψ A = i R, so ψ 6= ψB .
Still, we will use ψ to denote ψB for the remain-
φ
ω= = 2πν. (6) der of this paper with the clear understanding
t
that when we use ψ hereafter we are really
We will return to these terms later in our dis- talking about the complex number ψ at state
cussion of probabilities and our approximation B and not the change in the complex numbers
for the extreme price displacements ±R. from states A to B.
We now have the elements of our three- Looking at our solution for the price dis-
dimensional homogeneous coordinate system: placement in Eq. (4) we can make two related
elapsed time, displaced price, and displaced observations. The first is that there exists a
phase, or (t, x, φ). From these three coordinates unique phase displacement measure φ for each
we can determine all relevant dynamics of the possible price displacement over period t. The
financial market that occur over the period t. second is that the net movement in price over
Note that the price displacement is a func- period t is some maximum absolute measure
tion of elapsed time and the phase displace- multiplied by a function that oscillates between
ment is a functional of the price displacement. −1 and +1. In other words, xmin = −R and
This gives us the homogeneous coordinates xmax = +R.
(t : x (t) : φ[ x (t)]), with each coordinate having This has significant implications for our the-
some proportionality to a unit of elapsed time, ory of market mechanics as it suggests there is
or some net price displacement measure beyond
t x φ which an asset cannot go for a defined period.
: : .
t t t The absolute price displacement can be greater
Since we can regard t = 1 with our unit than R for a period less than t (e.g., tn− j − t A
assumption, we can reduce this to a two- where j is a positive, non-zero integer); how-
dimensional coordinate system with no loss in ever, it cannot be greater than R for t B − t A .
generality. This gives us the simplified homo- This means an asset’s net price movement is
geneous coordinates ( x : φ) that become equiv- constrained for a specified period, not by some
alent to the “speed” of the market’s mechanics
10 Although it is more accurate to refer to these as wave-
over each unit of elapsed time, or ( ẋ : ω ).
like functions, for the sake of brevity we will call them wave
Set up this way, ψ = ψ A + ψB is the complex functions throughout.
10 J. T. MANHIRE
external regulation, but by the properties of Alternatively, the Lagrangian is often de-
the asset itself. We will cover this more fully in fined as the difference between the kinetic term
a later section. K and the potential term V of a system, each
measured at tǫ , since K expresses a property
in terms of ẋǫ and V in terms of xǫ . There-
4. The Action Principle
fore, L = Kǫ − Vǫ .12 It is well known that
Our next endeavor is to examine whether the in any physical system the observable path
motion of an asset through price is consistent of an object (the trajectory the object actually
with the principle of stationary action, mean- takes through configuration space over time)
ing we will investigate whether the action of minimizes this differential L over time.13 This
the market system is stationary under small is where the term “principle of least action”
perturbations along its path from x A at t A to comes from.
x B at t B , at least to a first order approximation Still speaking analogously, recall from the
[10][11]. third postulate that this theory posits that an
We would be remiss if we did not acknowl- asset contains none of its own “force” and,
edge that the discussion in this section of a therefore, none of its own “energy.” Any dis-
financial market as a system with “energies” placement of the asset in price is a result of
(even a discussion by analogy) understandably work done on the asset, which is equal to the
appears as a Samuelsonian nightmare for some net external force over the amount of any dis-
readers [12]. Please keep in mind that, per our placement. The theory further holds that there
first postulate, we use these concepts analo- is only one aggregate source that generates this
gously but hold that the analogous concepts net external force and introduces it to the asset:
might help us better understand the dynamics the net effect of buyers’ and sellers’ trading
of the market mechanics we observe. interactions.
We do not make this examination for its own From the previous sections we see that at
sake. We will use the calculated value of the each time slice τ the asset moves linearly in
market’s action to derive a method for approx- price. As applied here, the principle of station-
imating the constraints of the system’s price ary action holds that for each elapsed time t
displacement for any time slice that are a con- the “path” taken by an asset (i.e., the curve
sequence of its oscillating mechanics. traced out in the relative configuration space of
the price dimension) between times t A and t B
A. The Action and the Lagrangian Generally is the one for which the action does not change
(i.e., is stationary) under small changes in the
The Lagrangian is a way of describing a finan- relative configuration of the asset related to the
cial market as a function of the conditions at relative price dimension.
tǫ (or any initial time for a slice τ) relating to Since the price dimension is our vertical co-
the price at the beginning of the time slice (xǫ ) ordinate axis, we can go further and state that
and the first derivative of that price (ẋǫ ). It such a path is a straight line in either the up
contains the complete information of both the or down direction, although under rotation we
system and the effects of forces acting upon can generalize it as the positive or negative di-
the system.11
12 Note that the kinetic term is simply the integral of
11 Dirac would most likely consider the Lagrangian not the left-hand side of Eq. (1) and the potential term is the
as a function of the asset’s initial coordinates and its first negative integral of the right-hand side, both with respect
derivative, but instead as a function of the asset’s price at to price displacement.
time tǫ and its price at time tǫ+1 , i.e., as endpoints [13]. 13 This is shown by the Euler-Lagrange equation.
THE ACTION PRINCIPLE IN MARKET MECHANICS 11
rection. This is consistent with the oscillating Dirac’s approach [13] and, later, Feynman’s def-
dynamic expressed earlier. inition [14]. Observe, however, that the price
It is important to note this is only consistent displacement is not only a result of the phase
with observed market mechanics when t = τ; displacement of ψ. It is also a result of the
that is, the asset moves strictly up or down in phase displacement belonging to the complex
price only at local time slices. What is “local?” conjugate of ψ∗ , which we’ll denote here as φ∗
Well, that depends on what measure we wish to to avoid confusion.15 In fact, both φ and φ∗
give n and our definition of a temporal interval contribute in equal measure to x since, as we
unit. showed earlier, a complex number and its com-
Thought of another way, the action—again, plex conjugate share the same ℜ-axis value x B ,
the time integral of the Lagrangian—over any which in our homogeneous coordinate system
region of our price-time coordinate system is equivalent to the price displacement x. So
must be stationary for any small changes in if the Lagrangian for each time slice τ makes
the coordinates in that region. If we keep di- some contribution to the total price displace-
viding the regions until we get to a collection ment x measured between t A and t B so does
of time slices, we observe this stationary ac- twice the phase displacement for that time slice,
tion as the binary “up-down” oscillations of which we’ll denote as a functional of the price
typical market mechanics at each time slice τ, displacement for that interval τ, or 2φ[ x (τ )].
which again is the elapsed time from tǫ to tǫ+1 . Since the local elements of Eq. (7) sum to the
Therefore, for each time slice in our price-time total time integral of the Lagrangian between
coordinate system there exists a Lagrangian t A and t B , and therefore the total action, we
Rt
that is a function of its coordinates and their can state that each element tǫǫ+1 L dt ≡ dS (if
first derivatives with respect to time. n is sufficiently large) and that each element
Does this mean we can’t determine the ac- is minimized under the principle of stationary
tion between t A and t B when the elapsed time action.
is not local? No, in fact, quite the opposite. It
simply means we must first examine the spe-
B. Measuring the Action
cific dynamics at each slice τ between t A and
t B and then sum over all periods τ to find mea- We begin measuring the action by examining
surable results that match observed data. In the potential and kinetic terms of the system.
other words, the action S from t A to t B is equal Any positively-directed force can be expressed
to the following series of time slices: as the negative spatial derivative of the sys-
Z tB Z t Z t2 Z tn tem’s potential. Thus, by analogy, we can ex-
1
Ldt = Ldt + Ldt + · · · + Ldt, press the negatively-directed force kx from Eq.
tA t0 t1 t n −1
(1) as the positive price derivative of the po-
(7)
tential (or its analog) of the system we call
where τǫ = tǫ+1 − tǫ just as on a larger scale
a financial market. This potential V then be-
we find t = t B − t A .14
comes
The Lagrangian over each time slice τǫ in 1
V = kx2 . (8)
our price-time coordinate system makes some 2
contribution to the total price displacement x
15 Technically, the complex wave function ψ has the con-
measured between t A and t B . This is similar to
jugate ψ∗ and both have identical phase displacements φ,
although they are oppositely directed; however, it’s easier
14 Of course, we can always set t to τ , t to t
ǫ B ǫ+1 , and t A to talk about the phase displacements attributed to each
to tǫ if we choose. by denoting them, for the moment, as φ and φ∗ .
12 J. T. MANHIRE
Note that the potential term of the asset is the movement of the asset through the price di-
simply the price integral of the right-hand side mension. In other words, W = Vǫ = Kǫ+1 . For
of Eq. (1). The kinetic term should then be this to be true, Vǫ+1 and Kǫ must equal zero,
the price integral of the left-hand side, which with V and K “trading off” measures over each
becomes elapsed period but the sum of V and K always
1
K = m ẋ2 . (9) equaling W for that time slice.
2
So the answer to the question, “What is the
At this point we should already begin to sus- potential of the system?” depends entirely on
pect an equivalence between the potential and when we ask the question over the elapsed time
kinetic terms in financial markets since both are slice τ. For this reason, it becomes impossible
results of price integrals of the balance of the to express the potential and kinetic terms in
net interaction of buyers and sellers expressed any way except as averages over τ. The only
in Eq. (1). Still, we will dive more deeply into exceptions are expressions of the potential and
this over the coming pages in order to be sure. kinetic terms at the endpoints tǫ and tǫ+1 of the
The next question becomes, “What is the generic slice τǫ . Since Kǫ = 0, the Lagrangian
measure of this action, and by extension, the becomes
measure of the phase displacement?” We at-
tempt to answer this by again examining the L = Kǫ − Vǫ = 0 − Vǫ = −Vǫ .
dynamics at each time slice.
At each slice τ the net external force causes a We’ve made a bit of progress. We now know
tiny amount of work to be done with respect to that the Lagrangian is the negative initial po-
a tiny displacement in price, especially when tential generated by the interactions of buyers
n is very large, or F (τǫ ) = dxdW . To figure and sellers for any time slice τǫ . But what is
(τǫ )
out the total amount of work done on the asset the measure of this potential? We know it must
over the period we take the price integral of be the same as the average measure of work,
the net external force, or W = 21 Fx. which is equal to 12 Fx. We also know from
Newton’s second law that F = m ẍ = mx t2
since
The work W is equivalent to the total poten- 2
ẍ = x/t when taken as an average. Therefore,
tial necessary to be introduced into the asset
from the net interactions of buyers and sell- 1 mx mx2
ers in order to make the asset move in price. Fx = 2 x = .
2 2t 2t2
Since the asset neither has nor stores any of its
own “energy,” any potential used to move the But this is the same as the measure of the
asset through the price dimension must come average kinetic term of a system since the av-
from the net potential introduced into the asset erage ẋ is always x/t. It seems, therefore,
by these external interactions. In other words, that Kmax = Vmax . Since Kmax implies that
the kinetic term K resulting from any price Vmin = 0, Vmax implies that Kmin = 0, and
displacement must come from an external po- Vǫ ≡ Vmax , we can conclude that
tential V. This means the final kinetic term
kx2 mx2
must equal the initial potential as measured Vǫ = = 2
2 2t
over any elapsed period.
Consequently, we can hold that the total as an average measure for t = τǫ .16 Conse-
work performed on the asset equals the to- quently, we can express the Lagrangian of an
tal potential initially introduced into the asset. 16 There are some who might argue that if we are ex-
This is also equal to the final kinetic term from pressing the average kinetic term where τ is small we must
THE ACTION PRINCIPLE IN MARKET MECHANICS 13
Since ẍ = − kx
m , we get S(ξ ) = S( x + η ) = S( x ) + S(η ),
Z tB Z tB
t k which is valid for an oscillating system. There-
ẋ2 dt = | x ẋ |tBA + xx dt.
tA m tA fore, S > 0 when η (t A ) = η (t B ) = 0.
The action then becomes The action of the arbitrary system then be-
comes
k tB k tB
m
Z Z
S= | x ẋ |ttBA + xx dt − xx dt , m
Z t
B k tB
Z
2 m tA m tA S( x ) + S(η ) + ẋ η̇ dt − xη dt .
2 tA m tA
which simplifies to
If we rewrite the term ẋ η̇ in terms of η, we find
m m that the integrated terms disappear since η = 0.
S = | x ẋ |ttBA = ( x B ẋ B − x A ẋ A ).
2 2 Therefore, S(η ) = 0 meets the condition that
Because we set x A = 0 in our homogeneous co- S > 0 when η (t A ) = η (t B ) = 0.
ordinate system, the x A terms disappear. Fur- We can also define the action as stationary
ther, because we can only talk about the kinetic if there is no change in S in the first approx-
imation. In general, if we have a minimum
quantity of a function then any perturbation
do so as away from that minimum in the first order will
x ǫ +1 − x ǫ x ǫ − x ǫ −1
m
·
2 t ǫ +1 − t ǫ t ǫ − t ǫ −1 only have a deviation in the second order. At
2
not as m2 · xt if the mean value of the square of the price any other place besides the minimum a small
displacement is dt instead of dt2 . While this exception is perturbation will show up in the first order, but
noted, the notation used in this paper seems adequate for
purposes of examining the price displacement of financial
at the minimum any small perturbation will
markets. make no difference in the first approximation.
14 J. T. MANHIRE
We can examine this by looking at the ac- absolute value of the phase displacement is in
2
tion as a series. The action S = mx
2t can be the range [0, π2 ]. It is the same for any negative
expressed as the infinite series price displacement. Therefore, if solutions to
|φ| include these two extremes when the par-
∞
1 n
h
n 2
i
tial derivative of the action with respect to the
∑2 ( t − 1 ) (− 1 ) mx (11)
n =0 phase displacement is equal to zero, we can
reasonably assume the action is stationary for
for |1 − t| = 0, which is what we’re looking for
all values of the price displacement | x |.
since we want the action to be stationary for
Recall from Eq. (4) that x = R sin φ. Thus,
any unit of elapsed time. Because we only need
we can set the partial derivative of the action
to calculate the value to the first approximation
with respect to the phase displacement equal
to judge if the action is stationary we can re-
to zero with the expression
strict this series from n ∈ {0, ∞} to n ∈ {0, 1}.
This gives us ∂
m(R sin φ)2
= 0. (13)
∂φ 2t
1 h i 1 h i
(t − 1)0 (−1)0 mx2 + (t − 1)1 (−1)1 mx2 ,
2 2 This becomes
which becomes mR2 sin φ cos φ
= 0, (14)
1 h i 1 h i t
(1) (1)mx2 + (t − 1) (−1)mx2 .
2 2
which generates two possible solutions: gπ
This reduces to and gπ + π2 , where g is an integer scalar. We’re
interested in solutions at g = 0 since all other
mx2
(2 − t ). (12) solutions are simply positive or negative inte-
2
ger multiples. This leaves us with the solutions
Note that at t = 1 this equation is equivalent
π
to Eq. (10). Thus, we see that δS = 0; the |φ| = 0 and |φ| = , (15)
2
action remains stationary to the first approxi-
mation for any temporal unit. Therefore, we which are the extremes of the absolute value of
can conclude the differential represented by each phase displacement we just mentioned.
the Lagrangian over t is minimized. Accordingly, we can conclude that all pos-
sible paths for the price displacement x will
From Eq. (10) we see that the action is
result in a stationary action to a first order
parabolic with regards to the price displace-
approximation. Therefore, the mechanics of
ment. Therefore, we can look for solutions for
financial markets viewed through the lens of
the price displacement that maintain the action
our homogeneous coordinate system conform
as stationary. One way to do this is to take
to the principle of least action for all possible
the partial derivative of the action with respect
price paths.
to the price displacement and find solutions
when the partial derivative is zero. Our hope is
that the action remains stationary for all possi- 5. Deriving Probabilities
ble values of the price displacement, meaning
We now understand that there exists a unique
from zero to |R|.
complex number ψ and its complex conjugate
For any positive price displacement on the ψ∗ for every discrete price displacement x in
complex plane ranging from zero to |R| the the inclusive range x ∈ [−R, R]. We can ex-
THE ACTION PRINCIPLE IN MARKET MECHANICS 15
press ψ as an exponential function of φ in the method for doing this by defining multiple re-
form gions and then splitting them up, but we can
ψ = i exp[−iφ]. (16) also view this as a basic probability problem.
By asking, “What is the probability of x having
This becomes a probability (Gaussian) function
the specific value X?,” we’re essentially ask-
by simply squaring the exponent
ing, “What is the probability of finding both ψ
h i h i and ψ∗ each with specific phase displacement
i exp (−iφ)2 = i exp −φ2 (17)
values φ = Φ where Φ is the unique phase
and it becomes normalized by dividing the displacement for X?” Formally, this becomes
Gaussian by the sum of all possible values of Pr(ψ ∩ ψ∗ ). We know how to solve this since it
the Gaussian is well known that
action for each elapsed period then becomes 2πν(Φ), where ν(Φ) is the average frequency
!2 at the specific phase displacement value Φ. We
i exp[−S] 2 can express this as
R∞ = Q x e−S (21)
i −∞ exp[−S] dx
Pr(|φ| ≥ Φ) ≡ Pr(|φ| ≥ 2πν(Φ)). (23)
where Q x is the corresponding normalization
constant for the price displacement in terms of We can also express this equation using the
the action with the value Gauss complementary error function as
r
m erfc(Φ) ≡ erfc(2πν(Φ)), (24)
Qx = . (22)
2πt
which implies
This approach implies that the total proba-
bility measure of the absolute price displace- Φ = 2πν(Φ) (25)
ment | x | for any region of relative configura-
tion space with boundary coordinates (| x |, |φ|) or
Φ
is the product of two identically split regions of ν(Φ) = . (26)
2π
three-dimensional relative configuration space
with elapsed time t = 1. This is because the Let’s next try to express the probability of
probabilities as we’ve constructed them are the absolute price displacement | x | being at
exponential functions relating to the absolute least the value | X | in terms of the action S . We
phase displacement |φ|. The entire probability know the value of the action from Eq. (10). To
measure is the combination of two evenly-split find the normalized probability of | x | ≥ | X |
regions with respect to |ψ| and |ψ∗ | along the in terms of the action we integrate the action
ℜ-axis of our homogeneous coordinate system as an exponent from | X | to infinity. We then
for each time slice τ. normalize the result by dividing by the integral
In short, we must square the traditional prob- of the same from zero to infinity. This gives us
ability function for |ψ| effectively doubling the R∞
absolute phase displacement since there are |X| e−S dx √
R∞ = erfc S . (27)
two possible ways in the complex plane we can 0 e−S dx
get the exact same price displacements for x,
namely, ψ and ψ∗ . The same is true for − x But recall that this is only one-half of the me-
with −ψ and −ψ∗ . This approach does not vio- chanics since it represents the probability in
late the principle of unitarity since probability terms of |ψ| only. To complete the picture, we
measures remain between zero and unity.17 must add the |ψ∗ | half, thereby squaring the
Recall our expressions of the frequencies in entire quotient. This results in
Eq. (6) and our use of an elapsed time unit. The √ 2
probability of the phase displacement being at Pr(| x | ≥ | X |) = erfc S . (28)
least some value Φ is equivalent to the proba-
This implies
bility of the phase displacement being at least
17 One might argue that this approach assumes that price r 2
m
displacements in the positive and negative directions are Pr(| x | ≥ | X |) = erfc | X | (29)
symmetric, which in most cases they are not. However, 2t
if we only measure the probabilities of the absolute price
displacements and the corresponding absolute phase dis- from (10).
placements, this asymmetry should not matter for our
purposes. Since there exist 4|φ| for any | x |, we can state
THE ACTION PRINCIPLE IN MARKET MECHANICS 17
If | X | = |R| then |Φ| = π2 . From Eq. (26) we We will next apply these theoretical expres-
see that sions to historical data from various financial
π 1 markets and see to what extent the historical
ν = . (33)
2 4 and theoretical results are consistent.
Substituting this into Eq. (32) we get
lected well in advance of, sometimes decades The third step is to code the data with either
before, an historical “crash” and see if the pre- a “0” or “1.” A code of 0 means the value of
dicted extreme is violated by the crash many | x | for a particular week w is less than some
years later. If it is not violated, the results arbitrary value | X |. A code of 1 means the
suggest two things: that (1) again, the theory value of | x | is greater than or equal to | X |. We
might not be incorrect; and (2) each financial can then take the average of all the 0’s and 1’s
market has an inertial coefficient (or its analog) for each | X | and call it the relative frequency
that is an internal property of that asset. of the market meeting the condition | x | ≥ | X |.
Limited by time and access to free market We denote this relative frequency function ρ.
data extending back multiple decades, we test The fourth step is to approximate the aver-
here only a few financial markets for the unit age value of the inertial coefficient m for each
period t = 1 trading week. We use represen- asset, which we denote as m̂. We do this by em-
tative stock indices from American, European, ploying Eq. (29), performing some algebraic
and Asian exchanges, a commodity futures manipulation, and concluding that the value
contract, a foreign currency exchange pair, and of m for a specific week w is
a single publicly-traded company. Each of !2
erfc−1 (
p
these has sufficient trading volumes and freely ρ(| x | ≥ | X |)
mw = 2t (40)
accessible historical data reaching back into at | x |w
least the 1990s and some into the early 1970s.
recalling, of course, that t = 1 trading week.
The approximation of the average value of the
A. Probabilities and Inertial Coefficients
inertial coefficient for a market then becomes
The first step in this test is to find weekly his- the value of m̂ that yields the highest coeffi-
torical price data for the financial markets we cient of determination value r2 for the depen-
wish to examine. The following are the weekly dent variable Pr(| x | ≥ | X |) and the indepen-
data that are freely available for each market dent variable ρ(| x | ≥ | X |) where | X | = | x |w .
from the date listed to the present:18 The following are the m̂ and r2 values for
each asset using this method sampled from the
• S&P 500 (U.S.): 15 June 1980 first 100 weeks of the data set with the dates
• Dow Jones 30 (U.S.): 10 February 1985 listed previously (e.g., the S&P 500 sample is
• NASDAQ 100 (U.S.): 15 June 1980 from the beginning of the week starting 15 June
• DAX (Germany): 03 January 1988 1980 to the end of the week starting 16 May
• Nikkei 225 (Japan): 15 January 1984 1982):
• WIG20 (Poland): 24 April 1994
• Gold Futures: 27 January 1980 • S&P 500: m̂ = 977.73; r2 = 0.9992
• USD/JPY: 10 January 1971 • Dow Jones 30: m̂ = 982.21; r2 = 0.9997
• Johnson & Johnson (JNJ): 23 March 1980 • NASDAQ 100: m̂ = 683.00; r2 = 0.9988
• DAX: m̂ = 504.66; r2 = 0.9998
The second step is to calculate the change • Nikkei 225: m̂ = 474.14; r2 = 0.9990
in price from the previous week’s close to the • WIG20: m̂ = 355.92; r2 = 0.9988
current week’s close (x B − x A ). We can then • Gold Futures: m̂ = 820.35; r2 = 0.9994
calculate the absolute price displacement ratio • USD/JPY: m̂ = 2513.76; r2 = 0.9977
using | x | = | x B /x A − 1| for each week. • JNJ: m̂ = 511.80; r2 = 0.9995
each market, the fact that the coefficient of • Actual | x | for Crash Week: 1252.72
determination is so high for most samples sug- • Interval from Pred. to Crash: ≈ 18 years
gests that our method of calculation is very
close to accurate. Nikkei 225 (N225)
• Data from: 1988-1990 Note that in all cases the data used to ap-
• Crash Week: 05 October 2008 proximate the extreme value of a market crash
• Predicted |R| for Crash Week: 2292.98 was from years and sometimes decades prior
20 J. T. MANHIRE
to the crash itself. This implies that each mar- data for those periods.
ket might have its own unique properties that What are some of the implications of this
affect the probability of its price displacement theory and the preliminary results we’ve seen?
and its price displacement extremes. The first is that asset price displacements might
We call the property m̂ an “inertial coeffi- comply with certain physical laws. We showed
cient” by analogy only, but it appears to act here theoretically, and the historical data do not
much the same way an object’s mass would in contradict the conclusion, that asset price dis-
the physical world. An asset with a relatively placements are perhaps constrained by extreme
high inertial coefficient is less likely to move positive and negative values beyond which
as far in price as an asset with a relatively low whey cannot go for specified intervals of time.
inertial coefficient. This market property is, at Might this theoretical discovery act as a sort
a minimum, consistent with certain behaviors of “Black Swan” predictor [15], at least in mag-
of objects we experience in the physical world. nitude? While there is nothing in this theory
that would tell us when a low-probability event
would occur such as the market crash during
7. Conclusions
the week of 5 October 2008, the theory did ac-
Asset prices move either up or down during curately predict the magnitude of the constraint
some period of elapsed time. This up-and- of the price displacement for any trading week,
down motion, by definition, is linear for the including the first week of the October 2008
specified period. Linear motion in one dimen- crash.
sion, in this case the dimension of price, is Another way for researchers to test this the-
the result we would expect if the asset’s me- ory is to look at the correlation between price
chanics comply with the principle of stationary displacements and the net trading volume for
action. Yet, linear motion in one dimension appropriate samples of periods t. If one ex-
is also achievable through circular motion in amines the correlation between the total trad-
two dimensions. If the single dimension we ing volume and the price displacement for the
observe is “real” and the unobserved second sample one should find little to no correlation.
dimension “imaginary,” then circular motion There should be a sense of randomness. Yet,
in the complex plane can explain the observ- if one examines the net trading volume (i.e.,
able linear motion of assets through the price the volume attributed to α less the volume at-
dimension. tributed to β) one should find a fairly strong
Given this construct, the phase displace- correlation. Your author’s guess is that the
ment of the complex wave function and the strongest correlation will be nonlinear.
phase displacement of its complex conjugate Hopefully this theory also inspires a bet-
are equally likely to produce an observable ter understanding of market mechanic deriva-
price displacement. The square of the wave’s tives, especially asset and options pricing the-
phase displacement is then responsible for any ory. While this paper does not explore these
observable linear motion, and therefore, the areas it seems only natural that other work
stationary action. As a result, we hold that the specifically examining asset and options pric-
square of a function of the phase displacement ing might benefit from this different point of
is equal to the action of the asset, or S = u2 . view.
Applying the action principle to defined pe- Lastly, even though these results seem
riods produces price displacement results for promising, we should remember that the menu
assets that are consistent with historical price is not the meal and the map is not the terrain.
THE ACTION PRINCIPLE IN MARKET MECHANICS 21
Simply because the results of employing such [6] R. N. Mantegna, H. E. Stanley, An Introduction
a model suggest that financial markets might to Econophysics: Correlations and Complexity
comply with certain physical laws assumed in Finance, Cambridge Univ. Press, 2000.
in the model does not mean that this is an [7] E. P. Wigner, Symmetries and Reflections, Univ.
exact explanation of phenomena we witness of Indiana Press, Bloomington, 1967.
in everyday market mechanics; that is, these [8] J. L. McCauley, Response to Worrying Trends
very well might not be the “actual descriptions in Econophysics, Physica A 371 (2006) 601-609.
of the forces and interactions at hand [16].” [9] R. Casse, Projective Geometry: An Introduc-
Furthermore, this theory might only explain tion, New York, Oxford University Press, 2006.
macro-market movements and might not be
[10] W. R. Hamilton, First Essay On a General
as applicable to all individual assets. It’s pos- Method in Dynamics, Phil. Trans. Royal Soc’y
sible that not all individual financial markets (1835) 95.
have the same mechanical patterns. Again, fur-
[11] W. R. Hamilton, Second Essay On a General
ther research is need to find the limits of this
Method in Dynamics, Phil. Trans. Royal Soc’y
theory’s application. (1835) 247.
I sincerely appreciate the help, review, and
[12] P. A. Samuelson, Maximum Principles in Ana-
advice of my dear friend Joseph R. Hanley,
lytical Economics, in: A. Lindbeck (Ed.), Nobel
my father Dr. John T. Manhire, and Professors
Lectures, Economics 1969-1980, World Scien-
James McGrath, Lisa Rich, Saurabh Vishnubkat,
tific Publishing Co., Singapore, 1992, p. 68.
and Nuno Garoupa at Texas A&M. I also wish
to thank Andrew P. Morriss for our catalytic [13] P. A. M. Dirac, The Lagrangian in Quantum
discussions during the early stages of this work Mechanics, Physik. Zeitschrift der Sowjetunion
and his continued support thereafter. Most of 3 (1933) 64.
all, I thank (and apologize to) my wife, Ann, [14] R. P. Feynman, Space-Time Approach to Non-
and our nine children who have endured inor- Relativistic Quantum Mechanics, Rev. Mod.
dinate neglect during the period this problem Phys. 20 (1948) 367.
has consumed their husband and father.
[15] N. N. Taleb, The Black Swan: The Impact of the
College Station, Texas. jmanhire@tamu.edu. Highly Improbable, 2nd ed., Random House,
New York, 2010.
References [16] M. Gallegati, S. Keen, T. Lux, P. Ormerod, Wor-
rying Trends in Econophysics, Physica A 370
[1] R. Landauer, Information Is Physical, Physics (2006) 1-4.
Today 44 (1990) 23.
[2] P. Mirowski, Physics and the ’Marginalist Rev-
olution,’ Cambridge J. Econ. 8 (1984) 361.
[3] M. F. Osborne, The Stock Market and Finance
from a Physicist’s Viewpoint, Crossgar Press,
1977.
[4] V. Gontis, S. Havlin, A. Kononovicius, B.
Podobnik, H. E. Stanley, Stochastic Model of
Financial Markets Reproducing Scaling and
Memory in Volatility Return Intervals, Physica
A 462 (2016) 1091.
[5] L. von Mises, Social Science and Natural Sci-
ence, J. Soc. Phil. & Jur. 7 (1942) 240-245.