The Practitioner's Discipline
Relational knowledge is built by remembering how a place changes
Previous: Episode 5, What the Eye Cannot See, the Pattern Reveals
Next: Episode 7, What the Quant Approach Cannot See
I stood beside a stream with a Gubbi Gubbi elder. He pointed to a flowering tree and told me where the fish would be downstream that season. I looked at the tree, then at the water. I could see no reason to connect them. I had arrived with the habit of treating a sign as a property of the thing in front of me: the tree had flowers; the stream had fish. The location he named seemed to require knowledge that the scene had withheld from me.
He explained the path between them. Blossom attracted an insect at that time of year. Where branches reached over water, insects fell into the stream. Fish gathered where food arrived. We were looking at the same visible place. He was reading a seasonal association learned through attention to that place, while I was looking at separate objects.
The explanation did not turn the blossom into a universal indicator of fish. It gave me a relationship with a place, season and sequence. The teaching belongs to the elder and a broader Gubbi Gubbi relationship with Country. I recount my encounter, not the tradition in full. What I can bring to my own work is the question it raised about how patient observation makes a connection legible.
More than one visible thing
The tree, insect, stream and fish could each be described separately. The answer lay in their sequence. Flowering brought insects; the overhanging branches brought some insects to the water; fish came to feed there. It was not a trick of perception. The elder knew an association at this place that I did not.
The blossom was a visible sign. Its meaning depended on a relationship the elder had learned, rather than on a property I could infer from the flower alone. Bohm’s distinction between visible form and underlying order helped me later describe my surprise, but the ecological explanation was concrete: food reached fish through a seasonal chain.
Nothing visible changed after he spoke. Yet I could now follow the route from flower to insect to stream to fish. Of course they gathered where the food fell. That moment of comprehension was mine; the knowledge that made it possible was his.
A transect through a changing forest
Consider an illustrative forester who walks the same route through a Victorian mountain ash forest over several years. These are towering Australian eucalypts. The forester notices fewer bandicoot diggings in the leaf litter, changes in ground vegetation and more frequent signs of introduced foxes. The observations suggest relationships worth investigating, but they do not yet establish what caused the changes.
The forester maps the diggings, plants, birds and ground conditions by place and date. They compare the pattern with rainfall, fire history and other possible causes. They do not need to assert that foxes caused every change to decide which relationship to investigate. The value of the repeated transect lies in a record that lets them ask whether the associations moved together and what happened next.
There is an important contrast between that work and a model trained on a convenient data table. The table may contain counts on separate columns but omit which animals used the same patch, when a plant flowered, or how a disturbance travelled through the place. Better instruments can add those observations. The decisive step is choosing to record the relationships in the first place. Precision applied to an impoverished set of variables cannot recover what was never observed.
Two records at the trading desk
The encounter shifted how I treat memory at the trading desk. I write an observation while its outcome remains unknown and let the later record test it. My first record is about implementation. On a day with a signal, I record the data used, the eligible contract, the order, position size, fill and exit. This lets me distinguish a rule’s result from an execution mistake. It also prevents a profitable breach from being quietly counted as a success of the strategy. The record is deliberately unromantic. It is where fidelity to a method becomes observable.
My second record is about research. I write down what I noticed, when I noticed it, what I think the configuration might imply, and what evidence would count against that interpretation. If several markets seem to respond differently to breakout attempts, that goes into a dated note rather than into an unannounced override of today’s signal. Later I can count the times the same appearance led nowhere, compare other markets and periods, and test whether a changed rule would have improved the complete trade distribution after costs.
These two records protect each other. The implementation record keeps my research from being based on a version of the strategy I did not actually trade. The research record keeps faithful execution from becoming an excuse to avoid questioning the rules. An apparent improvement selected from many historical variations can be especially deceptive; Bailey and colleagues examine how repeated strategy selection can overfit a backtest. [1] A new idea needs a separate, deliberate evaluation.
Experience helps me decide what to investigate. The written rules decide the action due under the current strategy. If evidence warrants a change, I define, test and document the new rule before giving it authority over live trades. That separation is not an insult to practitioner judgment. It is how I make the judgment accountable.
For example, my implementation log might show a breakout entered in the correct market but with an unusually poor fill. The research log might contain a dated suspicion that fills have worsened during a certain time of day. One incident calls for checking the order and broker record. A series of comparable observations can support research into execution timing. Those are different jobs. If I change the entry rule after one poor fill, I may be treating an operational failure as a strategy insight.
Conversely, a research note about narrowing trend breadth may turn out to be meaningful across several markets and periods. The live implementation log shows whether I actually took the signals whose outcomes I later analyse. Without it, an apparently successful filter could be built from trades I had already skipped for unrelated reasons. The two records make the separation between acting and learning visible, which is particularly valuable when the result tempts me to remember a better version of my decision.
Memory with a date on it
People often speak of the seasoned trader’s “feel”. Some of that feel may be a genuine compression of many earlier encounters. Some may be memory selecting the handful of vivid cases. A dated record can begin to separate them. I may notice deepening pullbacks in a trend and suspect a change. I can mark the observation before seeing the next three months, then examine all instances that fit the same description. If the impression proves useful, I can ask whether it survives later periods and whether acting on it is worth its costs.
The discipline can start before one has long experience. Choose a limited set of relationships, note their state at regular intervals, and record the interpretation made at the time. Record the surprising absences as well as the striking presences. Return to the same place, market or case and see what changed. A beginner’s notebook cannot equal knowledge transmitted through generations of life with Country. It can put the beginner on a path of learning from observed relationships rather than retrospective certainty.
The master does not see a different world. They have learned to remember more of its relationships. Such expertise still answers to the place and the next observation. In trading, my much narrower version of that discipline begins with dated records rather than an unexplained appeal to feel.
There are days when this discipline feels slow. Recording a suspicion while still taking the valid signal gives me none of the immediate relief that comes from acting on the suspicion. Months later, however, I can inspect the note alongside the full set of outcomes. The patience creates an asset my memory cannot provide: a statement of what I believed before the result could flatter or embarrass me.
At the desk, I can write down an association before knowing the result, then return to all its appearances rather than the memorable few. The elder showed me what trained attention to relationships can make visible. The flower had been visible to both of us. The relationship was what I had failed to see.
The stream kept running after our conversation. Episode 7 asks what a model loses when it treats a measured path as a fixed property rather than a trace of relationships still changing.
Previous: Episode 5, What the Eye Cannot See, the Pattern Reveals
Next: Episode 7, What the Quant Approach Cannot See
Sources and notes
[1] David H. Bailey, Jonathan M. Borwein, Marcos López de Prado and Qiji Jim Zhu, The Probability of Backtest Overfitting, Journal of Computational Finance 20, no. 4 (2017): 39–69.
Richard Brennan writes on systematic trading, complex adaptive markets, and the philosophical foundations of trend following at atstradingsolutions.com. His books include The Fractals of Finance, Complex Adaptive Markets, Carved by Impossibility and The Aussie Turtles Trend Following Guide.
Want to explore why structure exists at all?
Carved by Impossibility: What Remains When Everything Else Is Eliminated
The book explores the architecture of constraint, emergence, and reality itself, and what it means for how we understand markets, life, and the universe.
Available now on Amazon in paperback, hardcover, and Kindle.
Want the theoretical foundation for why trend following works?
The Fractals of Finance: Determinism, Adaptation and the Geometry of Markets
The book explores the full architecture of feedback, fat tails, and fractal structure in financial markets, and what it means for how we trade, invest, and understand risk.
Available now on Amazon in paperback, hardcover, and Kindle.
Want a practical field manual for trading trends and capturing outliers?
The Aussie Turtles Trend Following Guide: A Field Manual for Hunting Outliers adapts the timeless principles of the original Turtle traders into a systematic, rules-based approach for modern markets. Co-authored with Adam Havryliv.
Available now on Amazon in paperback, hardcover, and Kindle.