Trend Following Primer Series – The Need for an Enduring Edge – Part 7

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
• The Pain Arbitrage of Trend Following – Part 12
• Building a Diversified, Systematic, Trend Following Model – Part 13
• A Systematic Workflow Process Applied to Data Mining – Part 14
• Put Your Helmets On, It’s Time to Go Mining – Part 15
• The Robustness Phase – T’is But a Scratch – Part 16
• There is no Permanence, Only Change – Part 17
• Compiling a Sub Portfolio: A First Glimpse of our Creation – Part 18 
• The Court Verdict: A Lesson In Hubris – Part 19
• Conclusion: All Things Come to an End, Even Trends – Part 20

The Need for an Enduring Edge

In our previous Primer we pointed out how we, as trend followers, are very uncertain about the degree of confidence placed by the current state of the industry in ‘understanding these financial markets’. This makes us very sceptical traders who view all traditional methods of interpreting them as possibly suspect. Our reticence to trust traditional doctrine makes it appear that we have no guiding quantitative tools to work with in navigating these markets as we are unlikely to trust any of them, however that is not the case. We don’t need these elaborate tools that seek to simplify or act as proxies for interpreting the reality. We simply need to apply critical thinking and powers of observation to develop our own tailored tools that we use in that dreaded zone that lies in the tails of the market distribution.

Nearly all trading metrics and tools you see are sculpted with ‘predictability’ in mind and are targeted towards the peak of the distribution of returns. Trend Followers however are targeting the tails of the distribution and require a totally different set of metrics and tools to deal with them.

To demonstrate that trend followers can perform or even outperform alternative trading philosophies, we then took you on a brief introduction to some of the best performing funds of the world to demonstrate that trend following is perhaps one of the most effective processes we can apply to trading these markets without requiring us to invest in an army of quants, mathematicians and physicists to achieve our wealth ambitions.

We then discussed how trend followers use the term ‘robustness’ seen from the ‘warts and all story’ of an entire trading record as a basis to assess whether or not their systems have an edge. Once again we prefer the entire story provided by a back-test or validated track record that spans across a broad array of different market regimes before we award that system with the phrase ‘to possess a demonstrated ‘edge’. We prefer to understand this entire story of performance as opposed to using more traditional ‘shortcut’ measures such as the term Positive Expectancy which in their summarised efficiency of a single result, frequently require the use of assumptions that make their application suspect.

But nevertheless ‘an edge’ is an essential term that all speculators (as opposed to gamblers) need to digest, as without it, you are left floating down a river ‘without a paddle’.

A system with an edge is defined in the literature as a system with overall positive expectancy over its trade history…..and the greater the trade sample size, the greater confidence we have that the system displays an actual edge.

We use a proxy called ‘Expectancy’ to quantify this edge which is defined by how much money, on average, we can expect to make for every dollar we risk and can be calculated using the Expectancy formula E = (Pwin x Avg$ Win) – (PLoss x Avg$ Loss) over the trade history of that system.

• E = Expectancy or Expected Return
• Pwin = Probability of a win
• Avg$ Win = Average win in dollars over the trade history
• PLoss = Probability of a loss (or 1-Pwin)
• Avg$ Loss = Average Loss in dollars over the trade history.

But be sceptical as soon as you see a static formula applied to adaptive markets. The formula above leaves the impression that Expectancy can be used as a basis to compare different systems, however the devil is always in the details.

As soon as we apply a probability in a formula to achieve a ‘single summarised result’, we immediately turn a non-stationery background environment into a stationery one. We have a single value to represent an entire trade history. Given that markets are adaptive, then trading performance, being a derivative also needs to be variable. But a single statistic implies constancy across the time series. Many robust systems experience lower expectancy during unfavourable conditions and higher expectancy during favourable conditions, so a single statistic is not going to tell us this information and is therefore not a very useful statement about the performance of the  overall system or portfolio. Treat expectancy as a general guide to assess an edge but do not treat it in absolute terms.

Furthermore, no mention is made in the formula above about the sample size required to make this assessment. Can an edge be calculated on 10 trades or 10,000 trades, and what does this infer? Is an edge calculated on 10,000 trades in a high frequency trading environment that might span a few weeks, the same as the edge calculated on 10,000 trades in a long term trend trading portfolio over 20 years?

You see, by inferring that the statistical formula has meaning, we automatically include associated biases of reasoning into it.

Ideally what we want to infer in our assessment of a system with an edge is whether there is a profitable bias in the trade results of a system that is beyond what a random distribution of trades would imply. A large sample size of trades is therefore preferred to a small sample size of trades in our use of the expectancy formula….but perhaps more important than sample size are the number of various market conditions that the strategy has performed under.

While sample size is a useful proxy, and large sample sizes are more likely to have encountered a broader range of different market conditions, perhaps an even better measure of the term ‘edge’ is the relationship between system profitability which includes the number of different market conditions encountered by the system. While this would be nice information to have, our ability to derive such a formula is problematic due to market complexity. So we are forced to use generalizations.

We can of course use the expectancy formula as an idealized generalization which has some degree of merit in application to assess whether a strategy is profitable but watch out for the assumptions embedded within it.

Now many retail traders are under the misapprehension that an edge relates to the Probability of a Win (Pwin%) alone and therefore seek systems that can generate winning percentages that exceed 50%, however it is quite easy to guarantee this outcome by adopting profit targets that are half the size of the Initial Stop loss or by developing systems that deploy averaging down techniques or Martingale progressions that hold onto losers and never realise the loss until the total loss is recovered or in many cases the account blows up. Such manipulative tricks are commonly applied by charlatans that prey on the retail community to make it appear that they have found the ‘Holy Grail’ of trading systems that never incurs a loss.

While an illusion ‘of safety’ is created with a strategy that offers a 90% win rate, such illusions only offer psychological comfort. Warehoused risk lies in these solutions as it does in any trading strategy we generate. There is no way that we can avoid the risk inherent in a trading system through clever tricks. If we want to strive for higher returns, then we need to accommodate greater risk in our strategies to achieve it. What we will find in a later Primer is that trend followers use diversification of non-correlated and co-integrated return streams as a far more powerful method to achieve this ambition.

What is forgotten by many Retail traders seeking a high win rate is that Expectancy needs to include both the win rate and the reward to risk relationship for each trade. That is why the formula includes measures of the win rate and also the average win and the average loss.

For trend followers that are more often wrong in their trade decision, the Pwin% is usually in the order of between 20%-40%. Where we turn this game into one of positive expectancy is through our reward to risk relationship. By always cutting losses short and always letting profits run our average wins far exceed our average losses by multiples.

For example, let us apply the following results based on a trade history sample to our Positive Expectancy equation.

• Pwin% = 25%
• Avge Win$ = $3,250.00
• Avge Loss $ = $500.00

Therefore:

E = (0.25 x $3,250) – (0.75 x $500) = $437.50

As we can see from the example above, despite the poor Win% of only 25%, the system produces a positive expectancy of $437.50 per trade over the data history. If we have a large trade history (sample size) then we can be confident that our system displays a real demonstrable edge, subject of course to the limitations of  the equations assumptions.

But just having an edge does not guarantee a winning result. An edge plays out over an extended trade sample size. In every trade undertaken, there is a degree of luck incorporated into that trade event and as seen in our previous Primers dealing with random chance, luck alone may have a significant say in a small sample of trades undertaken.

A real edge as the well-known Australian trend trader ‘Nick Radge’ states only shows up over the next few thousand trades.

For example, chart 16 below demonstrates a comparison between a random set of equity curves and an equity curve (in red) from a trend following system with a real trading edge.

Chart 16: Thirty Random Equity Curves Including a Trend Following System with an edge (in red) -500 trades

 

On close examination of the chart above, you simply cannot detect that the trend following system has any defined edge over the first 500 trades. In fact, there are many random equity curves that have outperformed this trend trading system over a small trade sample size. Starting out with $1,000 over the first 500 trades the system with an edge managed to barely break-even.

However, let’s see how this slight edge plays out over 10,000 trades. You might say….’C,mon, are you seriously saying we need 10,000 trades under our belt before we can conclude that a system has an edge?”. Well we have a neat trick up our sleeves that significantly increases our number of trades undertaken to more quickly assess whether a system has an edge or not. It is called a diversified portfolio. Under a systematic diversified portfolio which comprises hundreds or thousands or discreet trading systems, your trade frequency quickly reaches levels well beyond what a simple discretionary trader can manage.

Chart 17: Thirty Random Equity Curves Including a Trend Following System with an edge (in red) -10,000 trades

Now if you closely look at the chart above, you will be hard pressed to see any random equity curves at all. They have all deteriorated to a very small equity balance by about the 5000th trade. However, now look at the performance of the trend following system with an edge over the 10,000 trades. It is now exceedingly obvious that something was driving this performance apart from lady luck.

Starting out with $1,000 we can now see how the next 10,000 trades with compounding have magnified the result to $350,000.

This is what we mean when we refer to trend following as a wealth builder as opposed to a cashflow generator. There is simply no way that we can guarantee that our weak edge will be reflected in short term performance. Random chance has a significant say in the matter over the short term. We can however be far more confident in the long term that there will be a sufficient trade sample to allow the edge to play out and benefit from compounding.

Of course, to allow that to happen we need to strictly manage risk along the way. The large effect of luck in the short term can significantly impede our wealth pursuits if it is not strictly managed.

This is how an edge plays out. It can only be demonstrated with a large trade sample size. Therefore, you need to be careful when you hear people refer to a trading system having an edge. You need a large sample of trades to back up this statement.

To see an edge play out, you really need to focus on the next few thousand trades. A single trade result is inconsequential to systems with a real edge. Luck plays a very large role in the short-term These financial markets are truly efficient most of the time…however you only need a small enduring edge to take advantage of compounding and the ultimate wealth it can generate.

Now this leads us to a problem when assessing the edge in either convergent or divergent systems.

Under a convergent methodology that is predictive in nature and responds to a ‘current’ market condition in the now by attacking the edge that exists in the peak of the market distribution, we need to see the edge immediately imparted in the trade results. We therefore do not want to see our equity curve immediately deteriorate when we move from the In Sample environment where we developed our trading models to the Out of Sample environment where we apply these models in a live market environment.

However, for a divergent methodology that is non-predictive in nature and target the edge in the tails of the distribution, we need to allow for enduring patience before we can assume that our systems are not performing as per their design. Having designed our systems within the In Sample environment, it may take thousands of trades before we can infer anything about their performance efficacy.

The Need for an Adaptive Edge

For the vast majority of retail traders, most are very unlikely to survive more than a few years and even for those that manage to escape the lessons of the market with a track record greater than a few years, there is no guarantee that it was not luck alone that allowed them to survive the day. You see there is so much room for luck in an efficient market that it might take many years for a trader to realise that they never possessed any real trading edge in the first place.

Now given that liquid financial markets are non-stationery in nature, your edge also needs to adapt over time. The application of a simple trading strategy with no changes over an extended timeframe of say 10 years or so duration is unlikely to hold its edge. So, you must have a process embedded within your trading plan that allows you to adapt your technique over time to respond to changing market conditions.

Many readers will be familiar with the Turtle trading experiment of two successful professional traders Richard Dennis and Bill Eckhardt. Richard and Bill launched the ‘Turtle Traders Experiment’ where they advertised for traders in the Wall Street journal and inducted their trainees in a Trend Following breakout recipe called the ‘Turtle Strategy’. The strategy was wildly successful in the 1980’s generating a fortune for the group of traders. From the success of the program many of the Trainees then later started their own trend following funds and have been successful to this day….however the recipe has changed substantively.

What worked in the 1980’s no longer works in its prescribed form today. While the overall breakout technique is still deployed today, the parameter variables have needed to respond to adapting trending conditions.

Such is the adaptive nature of these markets.

An Edge is a Pre-requisite Required for Trading Success

Despite the limitations of the Expectancy Equation in assessing the profitability of a strategy in adaptive market conditions and the embedded assumption of stationarity in its method of calculation, a real edge or “profitability bias” in the trade history is an essential prerequisite that must be achieved before we can apply fancy money management tricks or compounding techniques to magnify the overall result.

For example, if your system does not have an edge, there is literally no way that money management techniques designed to optimise your profitability can take effect. Many retail traders get stuck in ‘the rut of pink unicorns’ where they believe a Martingale strategy or some complex progression method can come to their rescue and deliver a no-loss strategy over the long term without requiring a definitive edge in the underlying system. They spend their entire lives caught up in this fiction, and leave their photos on the fridges of their homes to remind their partner that they still exist as they devise their Masterplan in their trading rooms.

This is akin to believing in a perpetual motion machine or some arcane form of alchemy. If you want to progress down this path, then be warned….you will require a padded cell by the end of it.

Another popular trading myth derived from the ingenious George Costanza principle in the Seinfeld TV series declares that if you flip a strategy that loses money, we can turn a losing system into a strategy with an edge. Once again this is a ‘Fools errand’.

What many do not realise is that most of the time a losing strategy is simply a strategy with no edge at all. It produces a random series of trade results. If we impart trading costs into the equation then the negative drag of the costs result in a losing strategy over the longer term.

When you flip a strategy with no edge, once again you need to apply the negative drag of frictional costs and you then find that the flipped version is also a losing strategy.

You see a flipped strategy actually requires correlation in the time series of trades to offer any possibility of providing a ‘George Costanza solution’. So, say we have a losing strategy that has a clear autocorrelated bias in the trade series that sends it directly on the path to hell. When you flip this strategy, there may be sufficient autocorrelated bias in the series to exceed the negative drag of costs and turn into a strategy with an edge…..but like a strategy with an edge, a legitimate ‘Costanza Strategy’ is as rare as hens teeth.

Most retail traders simply trade random strategies with no edge at all and may or may not be successful dependent on how the chips fall. The only way you have a chance of turning trading into a sustainable career is by finding strategies with an elusive edge.

Without an edge, then psychology, money management, risk management and diversification methods won’t save you.

More importantly however, an edge is the first step towards something far more powerful. The principle of compounding which we will be visiting in our next Primer.

Stay tuned for our next instalment in this Primer Series.

Trade well and prosper

The ATS mob