Going with the Flow – A Practical Example to Diversified Systematic Trend Following

Let’s say you have read some posts on this website with an open mind and can see the process in it all. Namely how to approach trading from the viewpoint of an observer embedded in a complex system who seeks to ‘go with the flow of it all’ as opposed to one who seeks to predict a future outcome with certainty.

Rather than speaking in riddles for those that find it hard to separate out that which can be practically applied versus mere foundational philosophical viewpoint. I thought it may be better to climb down from the monastery and demonstrate how we can put rhetoric into action without any use of complex equations or statistics to predict with pinpoint accuracy……..I promise.

All we will be using are past observations of return distributions arising from system application using visualisation techniques and what happens when we compile a vast array of system return distributions into a sum of histories statement about the past arising from those applied systems.

Now in hind-site, there is always ‘an optimal path’, but selecting a ‘curve fit’ optimal path for the future requires comprehending unknown unknowns…and that is a fictional statement. But there are ways we can reduce the set of all possible paths into simply those which are favourable and those which are unfavourable to the applied system. That is by ensuring an asymmetrical positively skewed ‘divergent’ design in the applied system that simply constrains unfavourable paths while leaving all future ‘advantageous paths’ open to future possibility. This is the quite the opposite of a negatively skewed ‘convergent design’ which assumes the past is a reliable guide to the future state. In other words a ‘predictive statement about the future replete with ideas such as ‘reversion to a historic mean, buying the dips and selling the tips, or future convergence to intrinsic value’.

What we are doing here is replacing the notion that an optimal ‘curve fit’ path of the past that is projected into the future into a statement of acceptance of all possible advantageous paths into the future from this moment of now (or an initial state of now).  A future of many advantageous possible histories as opposed to a single optimal (aka ‘optimised’) path for the future. The single optimised solution for the future necessitates the application of tight design constraints to steer it along the optimal path which significantly reduces the freedom in its choice of paths. On the other hand, by using many simple systems each with a small set of design constraints, we allow for many more advantageous paths to the future than that allowed for in an optimised model of predictive certainty.

Of course there is a sacrifice to be made by diversifying into a suite of ‘many systems’ that navigate many possible favorable paths, if the past condition extends for a short time into the future. In this instance those optimised single systems of predictive certainty head towards nirvana while the ‘many system approach’ of many possible advantageous paths is left with sub optimal results ….however we are prepared to sacrifice certainty of the short term with robustness for the long term and accumulate those sub optimal fruits along the way.

Having this approach in mind we use the visual histories of the past (the paths of all favorable  historic return distributions arising from ‘many applied systems’) to observe weaknesses arising from the total result from what is already known (predictable risk of the past) and plug these weakness through simple methods such as stop placement, with the remaining unknown unknowns of future adverse paths being restricted in impact through applied position sizing allowed for through diversification.

What I mean by this is that stops are really no secure form of insurance guarantee for the adverse outlier. They are good for restricting predictable risk, but have limited impact when **it hits the fan. Your only method of surviving the adverse outlier of unknown extent or timing is through limiting your exposure to these impacts through position sizing. You do this through diversifying into many uncorrelated return streams where your best of any one return stream are very small in the scheme of things.

So here is the practical example.

Above is an array of divergent strategies each with a favorable outcome over a 20 or so year history. The blank charts are charts of negative skewed paths which we remove. Those seeking predictive certainty in the remaining array would be tempted to use some form of filter to reduce this set to the ‘optimised’ solution for the future. The divergent trader however embraces the fact that this array represents a ‘many path’ approach of favorable outcomes in the past and would prefer to use all of them as a basis to project into an uncertain future.

We recognise that each path comprises different design variables used to manage risk. All of them are designed with stops in place which are established at ‘predicted’ levels of adverse past volatility and have an unbounded profit condition to asymmetrically position each strategy towards an outcome of profitability. By consolidating all these strategies into a single portfolio we ensure that the position size allocated towards each strategy is low. This is counter-intuitive to the predictive trader that has an optimal path in mind and uses leverage as his preferred weapon to ‘turn up the gas’.

So now let’s consolidate them all and see how this compares against any singular ‘predicted outcome’.

Now what we have in the chart above is a consolidated statement of the past. It is not a single possible path of the past, but a plural statement of past histories.

Now don’t assume that this is a representative statement of the future as future conditions are likely to be different to the past, however let’s see what is embedded in this consolidated solution in terms of managing risk within an uncertain future.

  • It contains some return streams that when projected into the near future will outperform provided that recent returns are a reliable guide to a possible short term  future;
  • It contains some return streams of a past state that will allow for considerable upside if those market conditions return in the future;
  • All of the past solutions have risk release valves baked into their design signature to protect against the unknown condition;
  • The composite of systems in the collective have ‘self regulating features’ built into the composite system (portfolio). In other words, the uncorrelated nature of this suite allows for the weakness of some systems to be addressed by the strengths of others….The collective position itself helps to plug the deficiencies of each singular return stream;
  • There is no evidence of selection bias in this collection as we have incorporated all systems into the collective. (aka it is not a curve fit solution).
  • The suite of systems have demonstrated that they are sufficiently robust to stand up against an array of past market conditions and furthermore if future conditions are sufficiently adverse when compared to past market conditions, these systems will not participate in that uncertainty…..but if future uncertain conditions are more favorable than past market conditions, then all systems will correlate with a favorable outcome……and Kaboooooooommmmm!!!!!!!!….that’s when fortunes are made.

Now the key to utilising this approach is to collect a vast array of systems and performance data and be in a position to quickly organise this morass of data into a coherent narrative.

It is not about the efficacy of a single system or predictive stance. It is about how you compile simple systems, each with a slight potential edge from the past into a far more powerful narrative to survive an uncertain future.

We have provided these tools to use in our shop and if you want to get down into the dirty details I advise you to listen to the two videos below.



Trade well and prosper

Rich B

You must be logged in to post a comment.