Trend Following Primer Series – Characteristics of Complex Adaptive Markets – Part 5
Primer Series Contents
- An Introduction- Part 1
- Care Less about Trend Form and More about the Bias within it- Part 2
- Divergence, Convergence and Noise – Part 3
- Revealing Non-Randomness through the Market Distribution of Returns – Part 4
- Characteristics of Complex Adaptive Markets – Part 5
- The Search for Sustainable Trading Models – Part 6
- The Need for an Enduring Edge – Part 7
- Compounding, Path Dependence and Positive Skew – Part 8
- A Risk Adjusted Approach to Maximise Geometric Returns – Part 9
- Diversification is Never Enough…for Trend Followers – Part 10
- Correlation Between Return Streams – Where all the Wiggling Matters – Part 11
Characteristics of Complex Adaptive Markets
Throughout this Primer Series we have been repeatably referring to the financial markets as being complex adaptive systems, and in this Primer, we would like to briefly understand this ‘complex and adaptive’ aspect of financial markets to demonstrate the important role it should play in shaping our philosophical approach to trading.
Now before we start, I need to stress that complexity science is a huge subject, so doing justice to the cause in this brief Primer is simply ‘the tip of the iceberg’.
Furthermore our ability as traders to adopt useful tools or statistical metrics arising from complexity science is a vain endeavor as the terms “complex and adaptive” mean that we are dealing with shifting goal posts all the time… There is no single metric that adequately describes it.
But the reason for our interest in examining complex systems is to demonstrate that powers of prediction within this arena have significant limitations. As a result we need to reign in our hubris and beat our swelling ‘all knowing’ ego into submission. It is in our interests as traders to become humble observers in the markets who ‘go with the flow’ as opposed to self-assured predictive thinkers that feel they know how to beat these markets.
The reality is that Financial Markets do not actually care what we think. It beats to the tune of its own drum and we are only privy to part of that beat.
Unfortunately, our brains believe that we can crack this riddle with useful workarounds and simplified predictive methodologies, yet by understanding how complex systems behave, our brains might want to think again. We need to think outside the ‘predictive box’ and in doing so we see that the standard toolbox we have been using to address these markets are a disaster waiting to happen. By understanding the limitations of statistical science, we become better traders as our appreciation of risk and uncertainty rises. Simply by appreciating that financial markets are complex systems is enough to make us better ‘risk managers’.
One of the central take-outs from this Primer is that being reductionists, humans like to be able to reduce complex problems into their constituent parts to better understand them, but in doing so, apply simplifying assumptions leading to conclusions that are ‘Almost Right’ but miss the fine print that actually make them ‘Not-Even Wrong’.
The complexity surrounding complex systems lies in their relational connectedness seen through their non-linear and emergent nature. They are scale independent and grow (typically fractally) from a seed into emergent form, where the emergent features themselves are not enduring in nature. These ‘ephemeral’ emergent features then act as scaffolds for future emergent form.
While offering a false allure of stability for indeterminate periods of time, the fine detail (frequently assigned by the statistician to rounding errors or having no material consequence) conspire to make complex systems continuously adapt over time. They cannot be faithfully described by any permanent description and are fundamentally non-stationery in nature. This essence of complex systems is not captured by conventional simple statistical treatment where we like to view systems as ‘being constructed’ from assembled bits and pieces of permanent and enduring nature. There is no scale in complex systems as they have no rulers or internal clocks. The method that complex systems grow is simply by non-linear fractal development using extremely simply rules. We do not necessarily see all these fractals, but rather see the emergent form arising from their rules-based construction.
You see, we find complex systems all around us wherever we look. In fact, you would be hard-pressed to find any ‘entity’ or ‘thing’ at all that is not actually a complex system at its heart. A brain, an atom, an elephant, a ship, a rainforest, a cloud, a city, a financial market, are all examples of complex systems, so it is worthwhile digging in to understand the central features of these amazing domains.
Having this understanding allows us to see the forest for the trees and able to recognise that statistical treatment is rarely a precise tool. The errors we find between theory and reality are sufficient to overturn entire theories, so in understanding complex systems better, we can build our powers in skepticism which is a very useful trait for an ‘uncertain’ trader seeking an enduring edge. Being uncertain is a gift that leads to enquiry, whereas being certain is a ticking time bomb waiting to explode.
Understanding complex systems allows us to possibly remove the blinkers and see some of the heinous errors we have made through our simplistic rendition of human theoretical constructs such as:
- treating quantified risk measures as viable tools to assess uncertainty;
- where we infer system mechanics from probability sampling;
- where we assume system stability;
- where we adopt the baggage associated with the Law of Large Numbers to assume the nature of real distributions, and
- in our conventional risk metrics applied to risk management….to name a few.
The complexity surrounding ‘complex systems’ leaves simplistic predictive models tailored for a static domain in the dust. We need to recognise the significant limitations of standard statistical treatment in addressing them. In fact, having an understanding of the limitations of predictive modelling creates an ‘enduring edge’ for the wiser divergent trader accustomed to ‘uncertainty’. Our edge is the fallout that occurs when predictive models go wrong.
Despite our awareness that almost everywhere we look, we observe a system at work, our understanding of complex systems has unfortunately been left wanting, partly due to the confounding intractability of interrogating complexity but largely attributed to the simplistic reductionist methods of enquiry we have traditionally used in attempting to understand them. It is this reductionist process that simply fails to account for the reasons for system complexity, namely the inter-relationships and dependencies created between constituent parts as opposed to the constituent parts themselves.
Common to all complex systems is a multiplicity of many parts in which there is an absence of a central control element, either internal or external. Rather it is the system itself that progressively becomes more self-sustaining and efficient through the exchange of internal resources via interactions between system participants, allowing for nested sub-systems and system structure to emerge that enhance system stability and robustness.
A key property that arises from complex systems is emergent structure and behaviour and a property of a system is ‘emergent’ only if we genuinely have a new feature that cannot be explained through a more detailed fundamental description of its constituent parts. This is where a new ‘thing’ in naive terms is constructed within a system context that can only be explained by the relationships that exist in the entire system as opposed to its constituent parts. A ship, an elephant, an artwork, a building, an organisation, a thought. The range of potential emergent properties of a system are endless.
There are never actually any things when you look hard enough….there are only processes. ‘Things’ arise from a limited biased mindset that is unaware of the deeper relationships that exist and their importance. Western Cultures like to use nouns to create permanence to ‘a thing’ as their approach to understanding things has been through reduction and simplification but if you dig deep enough into a complex system, there actually no things at all, only processes. In fact, you find that it is nested processes all the way down to create processes within processes…within processes…… all the way down the rabbit hole. Systems thrive from their complexity. Simplification fails to understand their devilish character.
Also common to all complex systems is that the re-use of system elements for multi-functional purposes allows non-linear power laws to come into effect where relationships between things evolve from a one-to-one to a one-to many relationship. The result allows for progressive system efficiency and durability making the composite more resilient to perturbations…….however due to the strong coupling that arises between system elements, a failure in one or more relationships can lead to cascading failures making the entire system progressively more vulnerable to catastrophic risk. The reason for the italics….is because this is what excites the price follower in these financial markets. These rare moments of decoupling when markets decide to transition.
Price Followers have a strong respect for complex systems and recognise the inherent levels of uncertainty that exist in them given their complex nature. They never assume they know everything but rather go with the flow of it all and do not try to fight it. Predictors however like to be certain about certainty….but price followers are very uncertain about certainty…….the two philosophies lie on either end of a spectrum from certain to uncertain about how a complex system behaves. The difference in opinion therefore shapes very different trading techniques in tackling any edge present in these markets.
Markets are complex systems and we as participants are only privy to part of that complexity. We need to look from inside out when trading on the right edge as opposed to Gods who look from outside in, or from a perspective of hind-sight.
You see, when you look from inside out as opposed to outside in, you immediately understand that you are just a curious emergent form residing in a domain of knowledge that always expands. The ability of complex systems to use their emergent structure to act as scaffolds for future form mean that our quest to understand them is simply never-ending. Once we understand an emergent feature which we describe within a particular domain, we then find that this domain operates within a larger domain of enquiry….and so on deeper down the rabbit hole we go. This is symptomatic of an embedded observer seeking to understand the system they reside in. It is a continuous quest of knowledge in an uncertain domain whose knowledge frontier expands.
This frontier between our knowledge of the market state and the uncertainty of the state that resides outside our bounds of competence is what distinguishes risk versus uncertainty. Our assumed knowledge regarding the state of the domain we understand leads us to many false conclusions and makes us blind to ‘fat tailed events’. Only in hindsight do we understand that the domain we trusted was just too narrow in extent. Unfortunately, as humans limited to our knowledge, we assign certainty to the known knows and wrap this certainty up with financial metrics that define ‘risk’.
So given our penchant for ‘certainty and predictability’ we like to encapsulate our risk metrics under Gaussian assumptions that theoretically take priority over the reality of ‘complex systems’ and use metrics such as the Sharpe ratio or standard deviation as a basis to assess trading risk. All the time, the divergent price follower who simply observes how complex systems actually behave shout “NO!, This is the wrong way to treat these complex systems. They deserve more respect”.
What we fail to address however is the uncertainty that lies outside our knowledge bubble that sends us to the graveyard early due to the inevitable inter-dependencies that exist between our limited domain of knowledge and the relationships that extend into this uncertain ‘uncharted domain’.
The multiplicity of agents in a financial system which is characteristic of complex systems, are the system participants. These include bankers, brokers, traders, hedgers, speculators, and gamblers. The way they interact create system complexity and the ‘gross expression’ of this myriad of interactions is price action. These participants have different opinions, but they transfer that opinion into the market through their entry and exit. Now the thoughts that shroud these opinions extend well beyond the financial markets themselves into a broader domain called ‘life’ and this is where the uncertainty residing in the left and right tails of the distribution of market returns germinates. The real reason for why the hedge fund decided to offload its risk may forever be shrouded by ‘uncertainty’.
Markets effectively represent huge computers that efficiently compile this opinion into a gross statement represented by price. Now as efficient as the market is in processing opinion represented by trading behaviour, we have seen in preceding Primers that arbitrage opportunities reside in this price action.
The innumerable transaction events of different scales (of entries and exits) arising from participant behaviour consolidate into either random features with no collective impact on overall price action (aka noise) or non-random emergent features whose collective impact place a bias on overall form. Under this interpretation we treat price action as an emergent signature arising from collective participant trading behaviour.
We can only interpret the very broad characteristics of this complex interacting behaviour through our reliance on gross statistical measures arising from price behaviour…just the same as we can interpret the gross features of a cloud by its temperature and pressure arising from the mechanical behaviour of its composite water molecules. By summating these interactions, the overall behaviour of the system itself can be summarized through the mechanics of price action and this behaviour can either be abrupt and fleeting in in nature or sustained over a period of time.
Now having an understanding that price action is representative of an emergent form arising from participant interactions, we can use metaphors that might help us interpret these systems better. Unfortunately, the complexity surrounding ‘complex systems’ renders traditional statistical treatment hopelessly deficient so using metaphors is often a very useful way to interpret these vast riddles.
It is helpful to consider price action as an emergent statement arising from the myriad behaviours of participants who exert an influence in overall form. Under such an analogy, we could group participants by common behaviours and consider the mix of participant grouping and their overall influence on price action. Recognising that participants change over time; we can there clearly see how the mechanics of price action is never stationery. It really is a moving feast. The ever-changing dynamics of participant mix insist that we cannot predict the future market state with fidelity.
There is no certainty in our financial markets. There are simply periods of stability in an otherwise uncertain state. It is uncertainty that is ubiquitous in a complex system and it is certainty that should be viewed as the anomaly. While we as price followers nod our heads to the statisticians and treat a trend as an anomaly, we secretly ‘flip the bird’ to the statisticians out there with our understanding where the anomaly really resides.
Expand the Mind to Experience Wider Domains
Before we leave this brief introduction to Complex Systems, I will leave you with a few examples that will tease the mind and help to shed new light on the way that you view complex systems. Unfortunately, the study of complex systems is a huge topic and cannot be conveyed in a single Primer but at least these examples may get the reader to appreciate what complexity science is about and why us trend followers pay lots of attention to observing processes at work.
Example 1: Metronome Synchronisation
In Example 1 we undertake an experiment where we observe the behaviour of 32 metronomes supposedly operating independently to each other in a random chaotic manner. This is of course an assumption from simplifying the experiment through reductionist logic. What we actually find is that there is a slight bias imparted by the table itself that synchronises the metronomes to create emergent order throughout the entire domain. The entire setup needs to be viewed as one complex system.
It is the processes that matter. Not the things. Look for the relationships between things to identify processes at work. Sometimes the market will be noisy…….and sometimes the market will be coordinated. Don’t predict it, just follow it. The table with the metronomes holds the key to the connection between them. Observe the entire experimental setup, as without it, you will start trying to make sense of the things and lose sight of the process. Just having this understanding empowers you to see opportunity and risk and then simply make better decisions.
The reason that this toy seems mystical is the way you have approached the problem. A table, a metronome, many metronomes. All “nouns” of convenient human categorisation that take form in our language and then become permanent ‘objects’ of your thoughts leaving us blind to the underlying process . All based on a limited understanding of a system forgetting the importance of the relationships between different emergent aspects of the system. You need to approach this problem using verbs to find the processes at work and eradicate those nouns of perceived bias.
Example 2: Not random but rather Deterministically Complex
In classical systems there is order in every system state. Financial markets are not quantum systems and they obey classical rules. There is no ‘actual’ fundamental randomness in classical systems, there is just emergent complexity. We assign terms such as randomness or non-randomness to describe complex systems, but they are not the reality, only human constructs that allow us to understand them. Example 2 highlights this principle.
The physicist David Bohm coined the term the ‘Explicate versus the Implicate order’ to wrestle with this notion. What changes is the nature of the state. In complexity lies deep simplicity. Very simple things that are configured in different arrangements.
The grander system state moves to different states through gradual transition but the impact of this transition is not linear throughout. What makes the fat tails is that you are only observing a small sub section of the entire state.
Observation of the processes is the key as opposed to small sample prediction of the things. If you first observe the greater system then you can choose the optimal path by going with the ‘flow of it all’.
Methods of prediction without knowledge of the entire processes at work can be disruptive to the system state. When you think you know it all from your sample statistics, then backtrack and think again as markets are adaptive and constantly changing. Do not ever deal in certainties.
Knowledge is an adaptive process that allows you to ‘keep up’, not a cup that is either empty or full. There is great value in using statistics. The problem however lies in its naive application.
Example 3: Prediction versus Uncertainty
The two pendulums below are used as a metaphor to describe convergent behaviour and divergent behaviour seen within complex adaptive systems.
Convergent trading systems (as exemplified by the single pendulum with a single mean) are based on the principle that price will converge to an estimated value whether that be an estimate of intrinsic value, or an estimated historic mean. They are therefore predictive and linear in nature given this assumptive stance and are couched in terms of assumed stationery condition.
Divergent trading systems (as exemplified by the dual pendulum with multiple means) are based on the principle that price will diverge away from an estimated value. They are non-predictive and non-linear in nature and relate to market transitions as opposed to periods of market stability.
Markets like any complex system tend to exhibit quasi stability (false equilibria) punctuated by unpredictable periods of non-stationarity. They therefore exhibit two types of arbitrage opportunity. That arbitrage associated with predictable stationery conditions (eg. convergent trading styles) and that arbitrage associated with unpredictable moves away from stationarity (eg. divergent trading styles).
A single pendulum exhibits predictable oscillatory behaviour suited to convergent trading styles such as mean reversion or value investing and the double pendulum which exhibits rich dynamic behaviour exhibits divergent behaviour suited to price following.
Stay tuned for our next instalment in this Primer Series.
Trade well and prosper
The ATS mob