The Chaology of Markets (a Multifractal trading model)

There are tons of books, articles, and views on chaos and

fractals none of which has directly assisted the trader to It requires an understanding of a map called the logistic
trade the markets by chaos theory and fractal geometry or in map and how it outputs logistic systems.
fact shown that those ideas are in any way different to the
prospects of traders than current methods. In this article, we
end all of that and outline a Multifractal trading methodology
that shows why and how a method based those sciences is By noting the behavior of iterates over different rates of
superior to “Technical” and “Fundamental” Analysis and is change (systems), and the specific character of their
intellectually accessible to traders. fluctuations (using web diagrams), a set of classifications
emerge that specify sequences of iterations by their
By Samm Ikwue HDM, PGD, MBA – Market Chaotist
period-doubling behavior, i.e. the number of iterations

F
or the concept of a fractal market structure to be of any needed for some seed x to return to some specified
use at all, we need to know what a price fractal is, where marker or point in range. It is this period-doubling
it comes from, what it looks like and how it behaves in behavior in iterated sequences that leads to chaos.
terms of the abstractions of chaos theory and the
concreteness of traded markets. Now the link between chaos
and fractals is rather simple because chaos is the result of an
iterative process and fractals are defined by iterates. But until
now if you asked to see what a price fractal looks like you will
most probably draw a blank. We found and defined what a price
or market fractal is and this means we have articulated the
fractal footprint of markets (incontrovertibly).1 However, in
order that the reader gains the import of this, and
understands the basis for it, it is important to provide some
background understanding of the issues involved. In this
article a mathematical background is not assumed.
Figure 1: Logistic Chaos: Web Diagram
What exactly is Chaos?
The underlying operation of the logistic map is one akin to
Chaos is a mathematical concept and any trader can the repeated folding and stretching of the space to which
understand the mathematics of chaos and how it relates to it maps, which leads to exponential divergence in the
Stock markets, Forex markets, etc. In fact, it can be intuitive, sequence of iterates. It is by this exponential divergence
where a trader trades with a need to understand exactly how of sequences that we measure for chaos. So, bifurcation
price moves in real-time. As such, it is a powerful sense by rates explain the relationship between chaos and
which to understand market dynamics. unpredictability. This is clear if we consider that at high
rates of exponential change, small errors multiply at
1
Benoit Mandelbrot in his book “The Mis (Behavior) of Markets,” exponential speeds. The dynamics of logistic chaotic
showed how the Iterated Function Systems (IFS) formalism of
fractal geometry, may be applied to the market and what a fractal systems is summarized by a structural diagram called a
generator is in terms of a model that simulates the market. Our bifurcation diagram.
context is different in that we do not speak to a model that
simulates or projects the market but one that allows the reading
and trading of markets in real-time using actual price charts. As
such, the need was to find that fractal footprint in situ that defines
the overall structure of markets in ways anyone can reasonably
conclude to be deterministic of market price.
The Chaology of Markets (a Multifractal trading model)
is that the absence of nonlinearity in a model describing the
market is problematical because chaos needs nonlinearity.
Nonlinearity is really what helps to make a chaotic system
meaningful because that is what constrains its dynamics to be
within the limits they express and as such explain how the
parts of the system relate in order to be. In direct terms
therefore, the basic reason why TA and FA remain widely
employed (and the market wickedly impossible to
“master” by them), is that those disciplines (and their
variants in OFT, etc) reflect the limits of understanding
generally available to market participants (with respect to the
Figure 2: Logistic Bifurcation Diagram
dynamical structure of markets).
The bifurcation diagram plots the end behaviors of different
systems (i.e. the end behaviors of iterated sequences at Chaotic systems are linearly unstable but nonlinearly
different intrinsic rates of change) against a measure of stable (the conundrum for traders)
change in the dependent variable. It shows the period doubling
In order to trade a chaotic system and be consistently
route to chaos and thus the structure of a chaotic system.
successful, it must be obvious to the trader what the local
This diagram tells us important things that help us better
limits are in a global frame that suggests the largest
understand market dynamics on a structural level (i.e. given
immediate objective of the market (each trade). Without this
that markets are deterministic, chaotic and nonlinear). One
kind of knowledge (structure) in a linearly unstable system, it
such important piece of information it provides is that chaotic
becomes a gamble to action reads. The trader is simply not
systems are fractal in structure, and so we can equate chaotic
sure of what is going on and everything soon begins to appear
behavior to fractal behavior (the bifurcation diagram is self-
random. Because the trader observes sequences in fast non-
similar, i.e. fractal). This means that an understanding of
monotonic evolution, the trader requires the specialized
fractals and fractal behavior enables measurement and
knowledge of chaotic dynamics and a setup that militates
control of the fundamental dynamics of the system. This is
against the “confusion” arising from chaotic properties of the
important because fractal behavior is less abstract in concept
variable price to read an emergent fractal structure per
and can be read by the pattern of point displacement in a given
period. Therefore, if a dynamical market system is
system. The chaos game for instance.
deterministic but generally unpredictable; and if in addition
Implications of Markets as Chaotic Systems it evolves by persistent cyclic trends (aperiodic cyclicality),
and is also known to have a fractal structure; then it is
There are several important (rational, even somewhat moral) possible to prescribe an interpolative model that is a
implications arising from the knowledge that markets are general model of the market and that will exploit it
nonlinear and chaotic. One crucial implication is that a key consistently. This means that based on the knowledge of how
premise on which “technical” analysis (TA) is based (history such a market is dynamically ordered, it is possible to read
repeats itself) is fallacious. Another key implication is that no and trade such a market with a consistency of result that
linear model of the market is suited to explaining price except demonstrably outperforms the market. As such, the market
in very partial terms (of which both so-called model can be shown, not just to be more effective than any
“fundamental” analysis (FA) and “technical” analysis (TA) predictive linear models of the market, but to be the correct
are examples). This is because the chaotic variable (price in general trading model of the market.
this case) evolves in nonlinear ways. But even more significant
The Chaology of Markets (a Multifractal trading model)
Comparing the “predictivity” of linear methods and dynamical systems in ways not possible before the science
the chaotic nonlinear method was formalized by Benoit Mandelbrot 41 years ago.
Multifractals are a generalization of fractals not
The “predictability” of any chaotic system depends on (1) characterized by a single dimension. Rather, they express a
how much error or uncertainty we are willing to tolerate in a continuous spectrum of dimensions reflecting complex
given forecast or future estimate of the chaotic variable (2) dynamical forces in play. Multifractals are often found in
how accurately we are able to measure a system's current experience. Coastlines, clouds, lightning, the human heart and
state, and (3) Lyapunov time (a time scale reflecting the time electronically traded markets are examples
from initial conditions to the point when a chaotic system
becomes unpredictable). Therefore, these provide a basis by “Fractal geometry is not just a chapter of
which to judge the “predictivity” of different approaches mathematics, but one that helps Everyman to see
on a comparative scale. The claim here is that it is possible to the same world differently.” - Benoit Mandelbrot.
improve current “predictability” of outcome from say 1/2
to say 1/20 in terms of 1 and 2 above, and as such, greatly The Multifractal Trading Model
clarify the empirical sense of Lyapunov time per interval in the
case of 3. This model relies on applying the tenets of Fractal Market Microstructure
Geometry and Chaos theory to reading and trading markets, Regardless of how orders flow through (the different dialects
i.e. it directly employs real-time Fractal Analysis of markets. of) an electronic trading system, in the end, there must be a
matching protocol that simplifies the interface between bids
Understanding the theoretical basis for the
and offers. For this model, the lowest offers frontend an
Multifractal Trading Model array of sell orders to match an array of buy orders – with
A fractal is a never-ending pattern. Fractals are infinitely the highest bids at the front end. And this applies across
complex patterns that are self-similar processes across variable chunks of orders reaching the market between
different scales. They are created (i.e. computer generated session open and close (and combining with existing orders).
models of fractals) by repeating a simple process indefinitely In processing chunks, elements either side of the market
in an ongoing feedback loop. Mathematically, any real system (bids/offers) are spaced by point value (and order size) and
that describes the same kinds of functions is a fractal. We can therefore dynamically define “demand and supply
say that fractal geometry is to chaos theory what geometry schedules”. So, these are vigorously populating ranges in
is to algebra in expressing the mathematics of chaos. A market time that generate many changing variables and as a
power of Fractal Geometry is the ability to model (explain) result, we have a myriad of sources per interval of not so
the explicit dynamics of chaotic systems. This allows two obvious fluctuations defining the price curve in real-time (in
equivalent senses of deterministic chaos: (A) a system that addition to the more obvious one we describe further on). This
“appears” to have “random” arrangement in space state reflects something called intermittency (or aperiodic
cyclicality - the signature sign of chaos). In other words, this
and or (B) “random” progression in time. This is extremely
algorithmically driven matching engine defines an intensely
consequential since graphical concepts and insights tend to be
unstable scheme in dynamically processing latent “demand
much easier to grasp. Fractals are infinitely complex (that is
and supply curves” in market time. Importantly (and as we
to say detailed). This means fractal phenomena can be
show next), the output of this process per interval, is simply
explained (modeled) to infinitesimal detail. Fractal dimension
market clearing; which information then feeds trader
is the measure of such complexity - i.e. the ratio of the change
reaction over the same intervals.
in Scale to that in Detail. The important point that is made here
is that all of this analytic power allows insights into complex
The Chaology of Markets (a Multifractal trading model)
Clearly, not only can we visualize the movements implied, etc 1. So we deduce from ALL of this - that for price to
from this basic order matching scheme, but we can infer from “move” in market time, based on order flow, we
this what a pivot is. A pivot then, is that clearing price point need a three point structure for any aggregation of
(order book level), that exhausts two matched (oppositely orders on any scale, large or small (the initiator).
signed) arrays, where there are no further matches ahead in
2. That is to say, that there must be, for uprising price,
the current interval, or in such proximity, as to sustain an
two buy pivots and one sell pivot and for downfalling
initial direction. We of course abstract from the fact that
price, two sell pivots and one buy pivot to generate
limit orders provide liquidity and market orders consume
liquidity. But as long as a current range is actively populating directed movement or flow. The Iterated Function
with orders (i.e. orders are queued either side of the market), Systems (IFS) formalism of fractal geometry would
the range is not cleared and a pivot is not established to call each instance (a fractal generator).
enable a reversal in such a range. Therefore, a pivot occurs, if 3. As such, a fractal primitive (the fractal generator)
the market exhausts oppositely signed arrays at a price point in this model, simply consists of 3 consecutive range
(order book level), with “momentum” (active queue) still on clearing price points (order book levels) in the same
one side to reach a “new” range of oppositely signed orders. flow. These are price points in micro space (and
As such, spot pivots pervade the entire trading space therefore macro space) at each of which price
including the shortest possible interval. This system does not reverses direction in sequence to define a price
exhibit stable equilibrium over any term, equilibrium is cycle. This underlying transactional framework
everywhere unstable. repeating (iterated) indefinitely across all scales
(the rules of recursion).
Simplicity and Recursion
In effect, fractal primitives define as volatility shaped (self-
Traded markets function at least 24/5 all year round (and
affine) linear inequalities in market space (regardless of
stock exchanges for significant periods of each weekday).
scale). Therefore, primitives are the best predictors of
During this time, all that goes on (per session) in terms of
direction immediately following in the same transaction space
microstructure dynamics (barring one or two abstractions),
or flow. As linear inequalities, there can be only 6 different
is what we have described, regardless of the variety of; player
definitions in any flow (combining in a diverse number of
types, order sizes, and investment horizons, intentions and
ways). In other words, fractal primitives affirm the (exact)
influences, news and market events or what have you. As such,
extent to which predictivity anywhere exists in market space.
all that happens in the market all of the time is buying and
Fractals allow (among other things) the trader to evaluate the state
selling. Therefore, the market at its most basic defines an
of the chaotic system per interval, and therefore across time
incredibly simple form of existence: It admits all incoming scales. The stress here being seamlessness; from market
buy/sell orders per interval, matches them and signs the flow. microstructure to market macrostructure, and what is more,
Mr. Market is mechanical. As such, the apparent complexity evidence of the feedback loop in the cyclical flow of markets.
of traded markets comes from the incessant and recursive
but even actions of market participants across different “Fractals should be the default, the
scales, and not from any other direct measures of their approximation, the framework.” - Nassim
humanity. This is to say, a simple pattern repeated over and Nicholas Taleb.
over again (i.e. indefinitely) on different scales within a fixed
structure --> intimates a fractal.

Fractal primitives defined by the mathematics of

Fractal geometry
The Chaology of Markets (a Multifractal trading model)
From Micro to Macrostructure In the simplified, primitive driven “trending” example, each cell
shown is a time frame, so we have a 1-minute frame, a 1-hour
Fractals Scale across Market Space frame and a 1-day frame inscribed with 1.5 cycles or
oscillations in each case. We are saying that the 1-minute
frame contains at least one and a half cycles of sub-minute
bars just to make its 1-2 leg up and at least one and a half
cycles to make its 2-3 leg down and at least one and a half
more to make up the last phase or half cycle shown. The same
logic runs through the one hour frame and the one day frame
in terms of their makeup in sequential flow (in practice the
nested (iterated) subcycle count per feeder frame could be
several more to the same effect but not less than 1.5 cycles
per feeder frame). Of course, this implies (as in
demonstrates) that the market is a singularity (a point in
time). It means that the market is deterministic at the margin
in dynamically scaling to a single form or price point across
Figure 3: Fractals Scale across the entire Market Space
all intervals. As such there is no sense in which price does not
So we have that, market space is space where price fractals move as a unit at all times and what is more, the marginal
define aperiodic cycles across time in simple (but nested) 1,2,3 move is NEVER a fair game. Everything happens (scales) in
legs - from sub-minute time compression through M1 – MN and (persistent) sequence following a fundamental order (iterated
in fact at (> MN). Fractal behavior whether over very small or hierarchy) from left to right (until a point) and then reverts
very large intervals is the same. Importantly, the smaller ones to origin (finite loop equilibrium) to resume (once again)
are nested in the bigger ones (i.e. Fractals nested within left to right. This order does not change because the order
fractals or equivalently nested within phases of larger fractal expresses in the rules of recursion intrinsic to market
cycles) all with the same form, and hence the deterministic mechanism.
(i.e. moves to only one other next state depending on the
current state), dynamically scaling, Multifractal structure of
markets.

Figure 5: A Fractal Profile (single resolution)

So to cut a long story short, the fractal price structure is

defined by nested (iterated), aperiodic (cyclical), price point
displacement and the market can go up or down in range,
Figure 4: "Trending" Example
over ALL time scales in downfalling or uprising runs.
The Chaology of Markets (a Multifractal trading model)
Never fine lines - rough lines - fractals spiral in and out of a dynamic is somewhat similar to these points on the Mandelbrot
nested (iterated) structure of finite loop equilibriums or set attracted to a singularity. Measuring such dynamic for
“origin” (i.e. mean-reverting series), to resume scaling in trend is almost meaningless and referencing a specific time
direction or counter direction. Clearly, the setup is all the frame for trade control pointless. Much of what we have
while driven by a nested (iterated) system of spot pivots described to this point is exactly reflective of this kind of
defining fractal primitives across the entirety of market space. dynamic.
In fractal geometry, we hear mathematicians say that the
orbit of some seed, s is attracted to S. S being a complete
mapping of all s (the complete fractal form). Sometimes S is
called a strange attractor. So yes, cycles, but not sine or
cosine wave like. Instead, much rougher, aperiodic, non-
monotonic, nested, infinitely scaled, i.e. fractal and therefore,
illusory, when read outside of that structure, especially if
focusing on a single resolution.

A Knowable Symmetry

As a result, (and this is key) – markets do not “trend” per

time frame in any useful (tradable) sense. Markets scale as a
unit, i.e. price spirals out of “origin” as a unit, going from left
Figure 6: Points attracted to a Singularity
to right, across scales (up or down in range) to a single point.
At such a point, price is back in “origin,” fluctuating (i.e. The trading objective, therefore, is to isolate and time
spiraling out of “origin”) to resume (up or down in range) as a tradable intervals (persistent series). We do this by
unit, and from a single point - making time frames dynamically accounting for the iterated sequencing,
entirely arbitrary. This is knowable symmetry; as such price proportions, magnitudes and temporal periodicity of Hurst
can be consistently timed with regard to the points in and out types as they combine (from left to right) and define a
of any sequence of target “origins” in a given flow. Price “discrete” flow. Fundamentally, this is a process that
movement is in no sense random. The mathematics of its “unscrambles” time (i.e., defines the dynamic outside the
dynamic is fixed (for all its incredible volatility). Markets are notion of time frames). This means, we completely do away
“predictable,” i.e. within the limits of fractal behavior because with trend analysis, focusing entirely on the fractal dynamics
electronically traded markets conform to a fundamental presenting within a calibrated fractal trading interface. Using
submission of Fractal Geometry; which is that, there is no the IFS sense of the market, we seek to follow the market as it
qualitative change when the scale of a fractal object changes hops from singularized point to singularized point. There is no
– a property known in mathematics as “scale invariance.” place for second-guessing the market by recourse to events
The old school 50/50 risk/reward paradigm is sourced in the exogenous to the scaling rules under interpretation.
idea that the market is random (with the large baggage of
emotions that entails for participants).

Trading by Chaos (Multifractal analysis and trading)

Trend is NOT your friend when the system is chaotic: consider

the dynamic here where the points appear to be orbiting
around and spiraling into a single point or singularity. Price
The Chaology of Markets (a Multifractal trading model)
Conclusion: “Unscramble time” Edgar E. Peters (1994) “Fractal Market Analysis – Applying Chaos Theory to
Investment and Economics”

First, it is clear that the dynamical fractal structure defines James Gleick (1987) “Chaos Making a New Science”
market opportunities - in terms of timing and range. To be Lori Gardi (http://www.butterflyeffect.ca): Close to the Edge – Event Horizons,
effective in such a space the trader must find a means to Black Holes and the Mandelbrot Set

accurately read the local limits in a global frame that defines

the largest immediate objective of the market. Among other
nuances of number behavior, price is in non-monotonic
evolution all of the time and the setup must account for the
flow orientation implied by the emergent fractal structure.
Second, there are at least three types of dynamics
acknowledged by fractal geometry (Brownian series, Mean-
reverting series and Persistent series), and it is the iterated
proportions and sequencing of each type combining in real-
time “discrete” flows (i.e. movements from singularized point
to singularized point), that multifractal analysis must account
for, to allow the trader time tradable intervals to near
exactitude and to scale (Cf. Mandelbrot’s clock time versus
trading time). But third and even more profound is that - to
read those Hurst types into validity, you “unscramble” time -
my (Mis) Understanding of Mandelbrot.

Figure 7: "Unscramble Time"

References:
Edward N. Lorenz (1993) “THE ESSENCE OF CHAOS”

Julien C. Sprott (2000) “Strange Attractors: Creating Patterns in Chaos”

Benoit Mandelbrot and Richard L. Hudson (2004) “The (Mis) Behavior of Markets”

Benoit Mandelbrot (With M. L. Frame) (2002) “Fractals, Graphics, and Mathematics

Education”

Benoit Mandelbrot (2004) Fractals and Chaos: “The Mandelbrot Set and Beyond”