So we have now reached the point where we have outlined the core design principles that we need to ensure are now embedded in all our different varieties of Price Following Technique.

Core Design Features:

Must have positive skew – Our asymmetrical design method impregnates every one of our systems with this core feature. An initial stop combined with a trailing stop condition and open ended profit potential manages downside risk exposure at all times from entry to exit and our open ended profit condition with no profit target helps to bias our trade results towards this outcome over the Law of Large numbers. This design feature is the ultimate reason why our systems are biased towards higher R:R

Must Trade in the Direction of the Long Term Price Bias – Outliers can take any direction they want to…however you need to take a 50% bet and pick a direction. With an asymmetrical trap, one side must have a very finite lower bound and we avoid any attempt at hedging our bets as this compromises the open ended profit part of this riddle. If we took an effective straddle with our outliers, with a short oriented trap plus a long oriented trap, we would then need to predict when to turn off the hedge and let the successful side run. We avoid the temptation to predict in our technique…..so pick a direction.

What we have found in terms of our understanding of the auto correlation that exists in a price series from time to time is that over the long term, what gives rise to the  extended anomaly is the extent and nature of it’s discreet but serially correlated momentum moves which give rise to the ultimate form and rate of change of the trend itself. There is more linear extension when trading in the overall direction of the primary trend direction (long term price bias) than there is in the counter trend move that occurs from time to time.

We therefore attack any form of outlier from a single preferred direction whether that be a single momentum surge or a composite of momentum surges over a long time interval that collectively account for a massive outlier over time. We want more outliers and more extended outliers…hence we adopt a preferred direction in our guess work when defining how to orientate our anomaly traps.

Must have filters that define where anomalies are more likely to occur. It is not sufficient these days to simply let loose a trend following system into a single market and timeframe and expect this single system to achieve positive expectancy over the long term. The markets emergent complexity has evolved and has been responsible for style drift in the trend following FM camp with many looking towards longer timeframes in their effort to avoid the noise of an efficient market. Gone are the days of the 50 SMA or the Turtle Trading System in it’s capacity to offer lucrative risk adjusted returns. With more complex ‘efficient’ market conditions, the same broad principles still apply but the recipes of applying these principles now differ.

There is now just too much noise and mean reversion going on in today’s efficient market. The composition of the market today is different to what it was in the past less mature market environment. Today and particularly given the the market intervention methods imposed by the Central Banks in suppressing market volatility, systematic trend following processes have had to adapt or perish with excessive drawdowns. Mean reversion has been a dominant factor of the market condition since 2010 and the predictive rhythm has been facilitated by the vast explosion of ‘predictive algo’s’ focused towards front running, pattern recognition and mean reversion.

We have spent considerable time mulling over all of this and undertaking tests to challenge our assumptions, much of which has has been shaped by Bill Dreiss from Dreiss Research supported by his stunning long term track record. He sort of plots in the camp of the Diversified Systematic Trend Followers….but he sees things slightly different than the traditional trend following mindset.

He bases his methods of trading the markets using the same fundamental underlying principles that we do. In a complex system such as the financial markets, price appears to stretch in a unidirectional way longer than what a normal distribution would imply.

He uses a fractal wave algorithm as the basis for trading trending conditions ….and the underpinnings of this philosophy combined with our testing give us hope in our models as being more representative of the underlying reality. We concur with his philosophy but just state there are particular zones in the market condition where this ‘stretch’ in price movement is more prevalent.

Here are some of the issues you need to contend with in your backtesting to get a feel for what we are saying.

1. You may find that the naive application of a simple trend following system works in only one or a few sets of market conditions (eg. a single market) but fails in 10 other different liquid market conditions (multi-markets). Why? Aside from the very rationale explanation of the different volatility that exists between different instruments in term of price movement which can be normalised by comparing price volatility in percentage terms, there is a deeper factor going on. Even after volatility adjusting to allow for a more direct comparison, the same issue exists.

You would therefore be tempted to conclude that certain markets trend better than others. We believe that is an incorrect assumption. It is the impacts of the participants in the market concerned that causes the altered behaviour. Change the participant mix…then change the market behaviour. It is the impacts of the plethora of participant competition in normal market conditions  that cause the noise or predictable oscillation in the normal zones of day to day price variation. This is exactly what we have been referring to in the past posts in this introduction. Every single liquid market has it’s unique behaviour or degree of predictable rhythm defined by participant behaviour, however within that same context, every single market has the potential to exhibit trending anomalies based on it’s complex system structure and the nature of markets from time to time in wandering to new market equilibria during unpredictable transition events….so we need to keep a watchlist on all of them.

The way around this dilemma is to…. step away from the noise and predictive tendency of that market behaviour. In stepping into the anomalous zone where predictors don’t dare to tread, the participants that dominate in that zone are the Price Followers. Those predictors that tread into that zone where their predictive models break down will unhappily transfer their net wealth to you.

You still need to deploy similar design principles of all times of cutting losses short and letting profits run 24/7….but in today’s more competitive market, you need to now be more specific in the ‘where you decide to trade’. 

The impacts of the applications of a filter are significant to your PL.

1. Your trade frequency on a particular market and timeframe takes a nose dive. This is counter to the notion that a high trade frequency is desirous in terms of being representative of statistical significance. The way around this common criticism placed on our methods is that… ‘after performing price normalisation so we can directly compare price volatility in % terms’ …..we treat all markets as simply being data that represents how price responds to varying market conditions. For a given data stream, we are very selective in when we trade, but with intense diversification across many data streams applying that same selective system, we up the trade frequency at the portfolio level to give us the rigour of statistical significance in our method. Being selective in the trades you undertake is a major plus for the trend follower in reducing the adverse impacts of over-trading. By focusing only in the outlier zone, the price excursions that enter that zone literally “scream out at you to take them”. There is very little uncertainty in the decision. You must take all of them….and having them “scream” at you helps in pulling the trigger. Once our traps are sprung we like to put these beasties out of their agony with a bullet given the kind chaps we are.  ?
2. Your Pwin% lifts to above what you might expect from a trend following Model without significantly sacrificing your R:R. More traditional Diversified Systematic Trend Following FM’s that are less specific in where they trade…but rather trade as many trending events under very wide market diversification……. tend to cluster statistically around a 20%-30% Pwin rate and compensate this feature in the need for a very high R:R to ensure positive expectancy long term.  The Pwin% of these traditional methods has been persistently declining forcing them to spread their wings across asset classes in their need to supplement their techniques with outlier trade impacts. The low Pwin% is therefore resulting in progressively increasing volatility in terms of their drawdown exposure. This is having a significant psychological impact on the investment community that place their faith in these successful long term approaches……however our models attempt to directly address this symptom and suggest that perhaps there is a way that we can offer an alternative solution that is more appealing to investors seeking lower volatility but still offering the superb ‘bonanza’ effect that the outliers will bring. The selective placements of ‘trend following traps’ in outlier zones significantly reduces the impacts of whipsaw from simple noise and mean reverting tendency. Price moves in the outlier zone are quantitatively more extended in nature than price moves in normal zones. This price extension is favourable to our win rate. Of course there is a 50% chance an outlier will go the other way…but the bias of ensuring we trade in the direction of overall longer term price bias combined with the extended price move lifts our Pwin% up to between 40%-55% with very little sacrifice to our R:R. In fact, trading in the noise dilutes your average R:R., so we can get the best of all worlds. A relatively high Pwin% plus a very high R:R. Now just imagine what this achieves in the overall expectancy equation.

The end result of these design considerations is expressed by this general ‘trap signature’ which is placed at all points in our outlier zones. This provides general design guidelines around which we embed in every system. Now within this general design guide we then deploy hundreds of different system variations to catch a vast range of different types of segment of the anomaly that have favourable impacts to our PL. It is not important to catch the whole anomalous move, but rather those favourable positive price extensions within the anomaly itself. We view this as the profitable segments of the anomalous condition.

A more traditional trend following mindset would say….’add to winning trades’….where we say as an alternative conclusion, to diversify with your systems to catch every possible profitable segment. The effect is similar but nuanced in terms of our opinion that outlier price extensions are more variable in nature than classic trends but have more overall price extension.

We then place a diverse range of different types of trap that obey these core principles into the outlier zone at all points in this outlier spatial map and keep them turned on 24/7.