It is now time for a bit of myth busting in our introductory journey to describe the basic tenets of a Price Following Approach.
Here are some randomly constructed price charts derived using Excel’s random number generator at the tick or M1 level and then constructed as a longer term price series.
Why so many random examples you may be asking?
It is to drive home a point and hopefully make you never look at visual clues alone as a method to determine future price direction and a reason for why so few predictive methods that use visual clues actually work long term.
What is clear from these examples is the inability to visually discriminate between a completely random data series and a price series with directional bias (serial correlation present).
The two types of auto-correlation required to either converge a random price series or diverge a random price series cannot be determined by visual method. It starts to test the validity of a number of Technical approaches to trading the market and brings home that important notion that our brains like to see causation in a complex system’s patterns of behaviour…..where no such causal link may exist.
Does this mean that all forms of technical analysis are as effective as reading tea leaves? No….it just means that using visual clues alone is far more subjective than other methods that are actually backed up by a long term substantiated track record. We have already stated that an edge lies in both Predictive technique such as applied by value investors and those quantitative mean reverting techniques (in harvesting any positive kurtosis of the market distribution of returns)….and the edge that lies in Price Following techniques (in harvesting any negative kurtosis of the market distribution of returns). We place greater trust in the verifiable track record and the quantitative proof than the subjective assessment that our fickle brains like to suggest amounts to causality.
We further hate the frequent statement that every trading strategy works and that what is more important is that ‘it suits your personality’. What a load of pyscho mumbo jumbo. Defining where the edge resides is a tough nut to crack. Most systems actually fail to exploit an edge and simply trade randomness. Let’s be a bit more objective first before we start levitating and chanting.
Statements such as this make you warm and fuzzy but detract from the importance of the game in determining the better models that are more representative of the underlying reality.
In a prior post “Don’t be fooled by Randomness”, we explore this notion of randomness further.
For example here is another random series which we would like to think contains causative form. A simple trending condition. Given it’s random construction, there is no room for the notion that something will drive this trend further into the future on the same trajectory. Your notion that this pattern holds meaning is therefore false….however your system rules would probably make you trade it.
The discontinuation of the trend in the series in this example would not be ’caused’ by a counter-trend real move, but rather a simple random variation to this price series.
But below is the same random price series with a slight 2% directional bias imposed across the entire random data set where the average price extension across the time series is slightly extended.
If you had fallen for the first example’s trap….you might now be more cautious and wait for a better and more enduring pattern to determine your entry signal…..but once again you actually miss the point of where the alpha resides in this example as it lies in the drift (the price extension itself) over the entire data set as opposed to a few single bars or pattern of meaning. Autocorrelation is rarely persistent however and tends to be discontinuous but repetitive in nature. The more likely story is one of general randomness punctuated at times by autocorrelated momentum impulse.
Now after taking account of the different scaled resolution between these two examples, you simply cannot see any difference between them using visual clues alone. You can only determine the bias in the price series through quantitative methods….but that is only in hindsight.
So here is the actual result when comparing the two price series in hind-site. Now you can clearly see where the exploitable opportunity resides. Not in the random pattern, but rather the separation state between the random pattern and the same random pattern with price drift applied.
Without the ability to use quantitative tools as we trade on the right edge of the chart, how do we know where to take our price following entry?
The bottom line is that you never know but you use a backtest to test your assumptive model so that in a live environment you have a better ‘chance’ of trading a feature of enduring substance.
We use a long term backtest to test our assumptions using quantitative as opposed to subjective methods to assess the overall positive expectancy of the systems we deploy.
Furthermore we need to map the results of our equity curve generated from the backtest against market data to see whether our system takes entries when there is a greater probability that we are attacking an enduring feature (an outlier) with a greater chance of having causation backing it (aka autocorrelation), as opposed to an ephemeral random pattern of no causative substance.
So this is why we avoid trading the ‘normal’ zone of market activity where price activity is ‘noisy’.
The noise expressed by randomness arises from the complex interactions created through mass participant behaviour of different sizes and agendas……. and may or may not comprise a hint of causative bias.
Let’s imagine a particular trading style that adopts a particular set of rules. It may be a method deployed by institutions, hedgers, or speculators. That particular set of rules dictate terms in how price unfolds through the interaction with that system in the market. It defines a subset of price behaviour that characterises a behavioural impact of a subset of it’s participants on overall price. The size and impact of that subset will exert an impact on other sub groupings of different behavioural type.
Now in a highly competitive market during normal market conditions we can define the ways participants impact on overall price by their behavioural impact. Some dominate over others causing a price bias….but when markets get really efficient it is simply characterised by noise. This is a statement that reflect the notion that overall price movement has no other causal impact associated with it’s future direction apart from simply the way this collection of participants and their behaviours interact. In highly competitive zones the dominant causal agent affecting price trajectory is simply randomness. It is only when this price behaviour becomes more coordinated that we can assign a dominant causal factor to it….such as a particular behaviour of a dominant participant, or when the range of participants that interact during particular exotic zones start to exhibit a coordinated behaviour.
In our trading philosophy, that avoids prediction like the plague, we simply adopt the following assumption. It is in the anomaly zone or the zone of extreme price movement where participants are ‘more likely’ to display a coordinated behaviour, or the nature of participant that resides there are dominated by a particular class of behavioural impact. This is where we find price bias thriving and where noise has less of a say in the overall price trajectory outcome.
We focus on the ‘anomaly zone’ that lies a few standard deviations away from the ‘noisy’ normal zone, where we assume that for price to have actually reached this zone….something more special than a random chance event may be the causative factor driving prices to that zone…..and within that zone, competition is likely to be far less prevalent….. as only a few types of speculator actually participate in that zone namely the Price Followers and a swathe of failed Predictors.
Most of the traders that focus their models in capturing alpha in the noisy zone, when taking an excursion into the anomalous zone, are now typically exhibiting coordinated behaviour as their models are now failing and the intrinsic risk in their systems are building…..and as a result in a concerted behavioural move….the price extension occurs that we have been waiting for…… the predictors are now being ‘squeezed’ and the Price followers are on the other side of their trades predating on this capitulation behaviour.
Below is a neat way to understand what we are talking about. A complex system’s method of expressing different emergent outcomes. Noise versus collective behaviour. Think of each individual metronome as a predictive sub grouping or class that focus on different exploitable opportunities….and look what happens when prediction changes to coordinated behaviour.
Listen to the tune of price. What causes the collective behaviour which dictates coordinated price direction in this specific complex system at certain times is the tiny relationship bias that exists between the metronomes expressed by the vibration on the table. This is a representative way of describing how alpha is eaten by each sub grouping and then predictions fail when there is no predictive alpha to eat. The initial vast array of different behavioural impacts now becomes coordinated as different predictive models of behaviour fail. Where complex predictive behaviour alters to simple human behaviour. The sub groupings then consolidate into a massive coordinated single grouping expressed by a single coordinated behaviour where human behaviour of fear or greed starts to dominate the prior predictive behaviour of many types…..and then the Price Follower attacks with his entry in the direction of the rhythm. The Price Follower waits until that predictive rhythm manifests.
So listen to music more to appreciate the rhythm 🙂
Trade well and prosper
The ATS mob
Want to know more? Then join us at ATS Trading Solutions. We offer unique services for the retail trader wishing to learn more about the techniques applied by the systematic diversified trend followers to generate wealth building sustainable returns. Membership includes access to all Programs and Trading Tools available in our shop, systematic development and portfolio construction education and practice and participation in our Members Forum.
For a short time only we are offering free access to the exclusive areas of our website reserved for paying members with 100 free vouchers on offer for the first month, so you can have a good look around and get a feel for what we offer. Just go to the membership page of our website and choose the Premium Membership Plan and proceed to checkout. On the Membership Checkout page just click on the discount code and apply the code VIP100 to receive your first month free. So long as you cancel your membership in your first month then you won’t be charged a penny. If you don’t cancel in the first month it will automatically go to $9.99 AUD per month following your first free month.
Looking forward to seeing you over there. So come and join us and participate in our forum with other like minded systematic diversified trend following folk.