The Trading Opportunities that Uncertainty Brings

Diversified Systematic Trend Following is well known for its ability to perform strongly during market uncertainty.

In fact, from the extensive research conducted by Alex Greyserman and Kathryn Kaminski in their landmark book “Trend Following with Managed Futures – The Search for Crisis Alpha” the term ‘Crisis Alpha’ was coined for Trend Following based on the techniques notable outperformance when alternative investment styles and asset classes were underperforming. This was not just a recent phenomenon. Rather it has been an enduring one spanning decades.

So now let’s have a closer look and observe the relative performance of the Diversified Systematic Trend Followers against the S&P500 Total Return Index and against a 60% allocation towards equities and a 40% allocation towards bonds. The 60/40 portfolio has been a popular portfolio allocation method used by wealth managers for the last 2 decades.

For reference purposes I have showcased the performance of the S&P500 TR Index (including dividends) since 2000 to September 2022. Notice the two very large drawdowns for 2002/2003 and 2008/2009 of approximately 50%. These poor performance periods coincide with the Tech bubble collapse and the GFC.

Also notice the small correction in 2018 associated with the Trump trade wars and increase in interest rates, the minor correction associated with Covid in March 2020, and the poor performance and building drawdowns that have been experienced since January 2022 which is associated with rapidly increasing inflation, slowing growth in China and geo-political uncertainty.

In Figure 2 below I have included the performance of the TTU TF Index which is a composite Index reflecting the performance of a large basket of Trend Following Programs (approximately 55 as at September 2022) with a long-term track record that exceeds 15 years.

I have also included the performance of the Vanguard VBIAX ETF which benchmarks the performance of the US Stock market and the US bond Index. This ETF provides a comparative benchmark to monitor the performance of the very popular traditional 60/40 portfolio.

The first point to notice in Figure 2 is that the historic negatively correlated relationship between the 60/40 portfolio has now broken down. In this current high inflationary environment both equities and bonds are strongly positively correlated and while equities are posting negative returns, so are bonds.

Bonds are no longer providing protective cushioning for the equity concentrated portfolio during periods of adversity.

But also notice in Figure 2 how historically, there is also a negatively correlated relationship between equities and the TTU TF Index during periods of equity crisis. Trend Following has demonstrated how it can come to the rescue when it is most needed.

While there is no guarantee that this relationship will endure, it certainly appears to be a historically stable relationship during periods of major adversity over the last 22 years which clearly persists to the current day. In 4 out of 5 equity downturns showcased in Figure 3 below, the TTU TF Index has been significantly outperforming. Notably all the major crisis periods of the Tech Bubble Collapse, the GFC and the current inflationary regime where equities have underperformed, Trend Followers have flourished.

The only exception to this relationship was the minor correction in 2018 whereby the short-term correction quickly reverted back into the prior regime. Short sharp corrections in general are unfavourable to trend following as it takes time for Trend Following Programs to exit from prior positions and lock into new emerging trends.

There is clearly something going on during periods of adversity where the nature of the market responds very favourably to trend following models. This brief article explores the possible reason why this relationship exists.

The second point to notice in Figure 2 is the relative outperformance of the TTU TF Index compared to alternative benchmarks. Over the long-term, Trend Following (TF) has demonstrated how its performance produces superior geometric returns.

We can see this relative outperformance when we observe some constituents of the TTU TF Index. Below in Figure 4 we showcase the performance returns of 10 TF Programs against Warren Buffet’s Berkshire Hathaway in yellow. You will notice that most of these TF Programs are currently sitting at their high watermarks and have enjoyed a spectacular ‘explosive’ ride over the past few years during a period where value investors and the US Equity Indices have been languishing.

Now while we have been focussing on how Trend Following (TF) has historically flourished during major equity market declines in the last 22 years, it is important to note that Trend Followers can also perform strongly during favourable equity regimes. For example, if we refer to Figure 5 we see that Trend Following outperformed in 2011, 2014/2015 and also 2020/2021 at the same time that equity markets were also performing strongly.

We find that the ability to not only perform well during uncertain regimes, but also during market bubbles and more stable market regimes is a major reason for our outperformance over the long term.

Let’s now refer to the overall correlation between the TTU TF Index, the S&P500TR Index and the 60/40 Portfolio. We can see the correlation statistics displayed on the chart of the TTU TF Index.

What we observe is that the TTU TF Index is uncorrelated with both the S&P500TR Index and the 60/40 Portfolio with values approaching 0 correlation. It is well known that Diversified Systematic Trend Following has a very low correlation to other popular investment methods and to any single asset class. This is not surprising considering the extensive market diversification of Trend Following which spans numerous asset classes. It is very unlikely that Trend Following will be highly correlated to any singular asset class.

While we understand that TF is uncorrelated with Equities, it is important to note that this does not imply that TF is not causally correlated during particular market regimes as Figures 3 and Figures 5 imply. It just means that over the long term the relationship produces an uncorrelated result.

So from what we understand so far, clearly there is something going on with these markets where Trend Following Programs are rewarded during periods of uncertainty, and if the last 22 years is any guide, we anticipate that this relationship will continue.

While Trend Followers are clearly biased and would conclude that 100% trend following is the ultimate portfolio for the long term, we are strong advocates for the need for traditional portfolio managers to consider including a material allocation of Trend Following to their investment portfolios to provide a degree of portfolio protection to their primary holding.

But now let’s dig into the reasons for why this relationship exists. Why does Diversified Systematic Trend Following offer powerful performance returns during periods of market adversity? There is some relationship that exists between the propensity of financial markets to trend with periods of market uncertainty.

The first thing we need to understand is why Trend Following exhibits these unusual properties when compared to other forms of investment. This can be summarised by two broad features that are common to all diversified systematic trend following programs:

  1. Simple Trading Rules that follow Price and Cut Losses Short and Let Profits Run;
  2. The Extensive Market, Systems and Timeframe Diversification that is deployed by Trend Followers.

Simple System Designs that Cut Losses Short and Let Profits Run

Trend Following techniques adopt a simple premise of cutting losses short and letting profits run and trade in both the long or short direction.

Such a simple premise means that when markets offer protracted periods of trending condition either long or short, trend followers flourish. Of course, when market regimes do not offer trending regimes, trend followers stagnate or enter periods of building drawdown.

Because trend followers cut losses short, we need protracted periods of unfavourable regime comprising many small losses to lead to large drawdowns, but when favourable trending regimes emerge, we can quickly get back to our high watermarks with a few good trend trades that pay for many of the small losses incurred along the way.

The simple trend following rules apply a trade entry using a very small bet with an initial stop and a trailing stop that progressively moves in the direction of the trend either long or short with favourable price moves (refer to Figure 6).

The small bet size, initial stop and trailing stop are used to mitigate adverse price risk between entry and exit. We never let a small loss become a large one.

The profit potential of the trending model is unlimited. No profit targets are used which allows for the potential for massive gains during favourable trending regimes.

By cutting losses short and letting profits run, the histogram of trades, when viewed over the long term, has the following typical positively skewed signature. Many small losses and small wins with the occasional very large win that skews the distribution to the right.

About 10% of our trade distribution accounts for the profitable equity curves over the long term. The vast majority of the trades we undertake do not materially contribute to performance. We can visualise this in the following histogram that is typically produced by Trend Following Programs (refer to Figure 7).

The dominant profit contribution of the major winners is why we refer to our technique as a process that  ‘hunts for Outliers’. It is the massive anomaly such as fat tailed events that are frequently found in uncertain market regimes which is the raison d’etre for why Trend Following produces long term wealth.

Extensive Diversification across markets, timeframes and systems

Any individual market may only have 2-3 Outliers over a 22-year history. Some have more and others have less, however by definition Outliers (aka massive trends) are unpredictable in timing and occurrence. With only 2 to 3 Outliers in a single market return stream over a long history, you can see why Trend Followers need to diversify to increase their trade frequency.

Given their infrequent nature but massive impact on the PL of a trader, we need to diversify as far and wide as we possibly can. Increased diversification in terms of market diversification, system diversification and timeframe diversification allow us to increase the relative frequency of Outliers in our portfolio distribution of returns.

It also allows the Outliers in our distribution to be more dispersed throughout the time series. Our PL’s are heavily impacted by Outliers. They provide the lifting power for the equity curve. By diversifying widely across markets, systems and timeframes we have a better chance of distributing Outliers across the time series rather than finding that all our Outlier trades occur at the same time. The more we can distribute the Outliers across the return series, the more we can smooth the volatility in the equity curve which assists principles of compounding wealth.

We can see the impact of Outlier trades in the following ensemble of return streams generated by an uncorrelated portfolio. Notice where the Outlier trades exist in the return series and how they are widely distributed (refer to Figure 8). Notice how each return stream is quite volatile and lumpy.

When we consolidate these distributed lumpy return streams together into a portfolio, our performance appears far smoother than the reality is (refer to Figure 9).

You can see in the following chart that increased levels of diversification improve our overall returns and also reduce our overall drawdowns. It is well known that diversification provides correlation benefits through return stream ‘drawdowns’ being offset by return stream ‘drawups’ in an uncorrelated portfolio, but less well known that a composite of positively skewed return streams produced by Trend Following Models increases the overall Compound Annual Growth Rate of a portfolio given the greater representation of Outliers in the portfolio distribution.

We can observe this phenomenon in the scatter plot below. Notice how with increased diversification, there is less dispersion in results in terms of the CAGR and Maximum Drawdown and that a more diversified portfolio trends towards higher CAGR with the same level of drawdown (refer to Figure 10). Portfolios comprising only 5 of a possible 44 market universe produce a diverse scatter plot. As we increase the level of diversification to 20 portfolios drawn from a possible 44 universe and then to 40 possible portfolios drawn from the same possible universe we notice that the dispersion reduces and trends towards higher CAGR.

The key to this outperformance under diversification relates to the Outliers that we hunt using our simple price following models. Under extensive diversification, our aim is to capture as many of these massive unpredictable directional anomalies that we can across a portfolio return series.

Below is an example of a diversified suite of trend following systems attacking an Outlier (in this case USDJPY – Refer to Figure 11). Notice that we are not applying a single trend following system to this extreme price move. We deploy many uniquely configured trend following systems using short term, medium term and long-term models. This approach magnifies the impact of Outliers in our trade distribution of returns.

So now that we have an understanding of how our Trend Following processes work, and the beneficial impacts we receive from extensive diversification, it’s now time to understand what is so special about these directional market moves we call ‘Outliers’.

You see there is something special about these directional anomalies that significantly contribute to our long-term performance. They are such important contributors to our success story, and we only need 5-10% of these massive anomalies in our performance returns to generate significant wealth over the long run.

But what we find when we dig into the weeds is that these massive directional anomalies are found lurking at the edge of chaos. This is the zone of the ‘fat tails of the market distribution of returns’ where markets are known to diverge strongly away from prior periods of stability and equilibrium.

Liquid Markets are Not Normally Distributed at all Times

When we plot the market distribution of returns for any liquid market over a large time series such as the last 30 years we find that the distribution exhibits a leptokurtic tendency. (Refer to Figure 12).

The distribution of market returns does not fit neatly within the Normal Distribution. This is fortuitous for investors and traders, as there is no edge to be derived from a normal distribution. A normal distribution implies that all prices are independent to each other and there is no time correlation in a price series where a price at an earlier time impacts price at a later time.

Now in Figure 12 we clearly see that there are three zones where price possesses correlated behaviour. Towards the peak of the distribution around the mean of the distribution and in the left and right tails of the distribution which are located at extreme points away from the mean of the distribution.

This leptokurtic signature implies that the market wanders between different regime states. At times the market exhibits behaviour where prices oscillate about an equilibrium. This behaviour provides the peaks in the distribution around the mean and is indicative of market behaviour associated with market stability and predictability. This behaviour is sought after by investors and traders that adopt ‘convergent behaviour’. At other times the market behaviour extends far into the tail regions of the distribution of market returns.  This behaviour is indicative for markets in a state of transition and are highly unpredictable in nature.  This is the region that the Trend followers are known to lurk.

Now it must be noted that the bulk of the market distribution of returns does indeed lie within the Normal Distribution. This is a symptom of a market’s efficient nature. However the fact that we find that the market can at times display correlated behaviour which allows the distribution to extend beyond the envelope of the Normal Distribution, this is sufficient for traders to harvest an edge from this serially correlated opportunity. But there are two broad forms of serial correlation that are important to a trader. Negative correlation found in convergent regimes and positive serial correlation found in divergent regimes.

Convergent behaviour is associated with methods that assume market stability and market predictability. The behaviour is associated with a markets desire to achieve an equilibrium state. Such methods as pattern recognition, high frequency trading, value investment and mean reversion adopt these principles on the basis that their investment strategies assume the market’s desire to want to revert back to a stable equilibrium. Convergent markets exhibit negatively serially correlated behaviour. This means that a high in price is followed by a low in price and vice versa that sets in train the markets propensity to oscillate about an equilibrium.

When markets possess convergent behaviour and oscillate about an equilibrium, there are a plethora of techniques that seek to exploit the repetitive conditions associated with this stability and predictability. In this environment, the impacts associated with this multiplicity of behaviours dampens the volatility in this regime and edges are quickly exploited. As a result, the behavioural impacts exert a negative feedback which ‘stabilises the market’. In this regime, economics follows the Laws of Diminishing Returns. Most investors and market participants adopt convergent behaviour as market ‘predictability’ is sought by this style of investor. The ‘convergent regime is information rich where we have a stable mean, and a stable variation about this mean in terms of standard deviation.

We also see Central bankers adopting convergent techniques to stabilise the market and suppress volatility through buying the dips and selling the tips. This massive influence from the Central Bankers can lead to protracted periods of convergent behaviour and the spawning of a plethora of convergent styles that seek to harvest this predictability.

Now at the opposite end of the spectrum in zones well away from market equilibrium and on the verge of market chaos, we see another form of regime which is the opposite to the convergent regime. We refer to this regime as the ‘divergent regime’. This is where markets adopt more chaotic behaviour and is associated with those times when markets transition between periods of equilibrium. The market loses its predictability during transitions and behaviour becomes chaotic in nature. In this zone lies the Outliers which have open ended profit potential.

If we draw small samples from the tail regions of the distribution of market returns we find complex distributions with a pathological character such as the ‘cauchy distribution’. No longer is there a single mean and a standard deviation to work with. This is an information poor environment for traders where anomalies happen which are well beyond the predictive abilities associated with more stable distributions. The divergent regime is typically synonymous with market uncertainty.

Now many may think that divergent regimes are few and far between but when we examine any liquid market data set, we find that massive Outliers of >5 Standard Deviations occur with significant frequency and far more than what is implied using assumptions drawn from the Normal distribution.

So now that we understand that liquid markets can shift between different market states, lets understand how these massive Outliers can be produced. What causes prices to make massive directional moves?

If we assume that the markets are a zero-sum game and that the price is moved by the buying/ selling behaviour of its agents such as the Central Banks, the Investors, the Hedgers, the Traders and the Gamblers, then we can imagine the markets as comprising an ecosystem of participants which we can group into convergent and divergent behaviours. The behaviour of a participant in buying and selling securities exerts an appreciable force to move prices. Collective behaviours from similar behaviours make this force non-linear in nature.

The relative weighting of convergent and divergent behaviours therefore produces a spectrum of market behaviours (or regimes) that range between pure convergence to pure divergence. Noisy markets which are the dominant form of market expression is when divergent forces counterbalance with convergent markets.

Now trends can be found in all market regimes but the serial correlation that drives the trend is important when it comes to explaining the ‘persistence of trends’.

For example, under convergent regimes we still find trending price series however they typically represent segments of broader mean reverting cycles. The trends are not enduring, and they tend to oscillate about an equilibrium given their negatively serially correlated nature. This form of trend is ideal for traders that adopt mean reverting techniques.

We can also find trends in a noisy price series where there is no serial correlation present in the time series. Despite looking exactly like other forms of trend, they have no enduring persistence. Their trend form is purely a result of how a random price series can be constructed (Refer to Figure 15).

Under divergent regimes, trends display directional persistence. Serially correlated clusters of price data are causally linked and drive the directional momentum in a trend. It is within this regime that the Diversified Systematic Trend Trader is found ‘hunting their Outliers’.

In this zero sum game we find the past ecology of participant behaviour changes under uncertainty. Convergent traders are forced to become divergent traders as their models, which were designed to exploit market predictability, no longer work. Convergent traders find that they very quickly enter serious drawdowns. The symptoms of negative skew lurking in their models is fully exposed.

But this transition of behaviour from convergence to divergence is not an instantaneous change. The impacts associated with these shifting behaviours of participants which grow over time exert non linear power laws on price movements. Prices tend to be ‘sticky’ and these transitions take time to evolve and come into effect, but inevitably lead to powerful extended price moves over time under the driving force of positive serial correlation.

As a result, during major market transitions we find that convergent traders abandon their predictable models and then without a roadmap to guide them revert to a behavioural tendency of fear and greed. They either exit their trades in haste or worse still, they average down into their positions to forestall the inevitable adding leverage to their demise. However, during protracted periods of divergence, such methods meet risk of ruin head on and margin calls force them out of their positions with high leverage. These behavioural impacts actually serve to reinforce the trending condition from growing collective behaviour.

This mass transfer of investor and trading behaviour from one of convergence to divergence in a zero sum game sets in place a mass transference in wealth over time from convergent participants to divergent participants.

Given that there is only a handful of trading techniques that exploit opportunities under divergence such as Diversified Systematic Trend Following and some forms of option trading such as the buying of calls and puts, we can see how there are periods of time when divergent methods experience massive wealth inflows in the form of enduring trends.

These are the reasons for the explosion in wealth during periods of uncertainty for the trend following community and why we see this during the tech collapse, the GFC and the current high inflationary regime. The seeds of divergence are not only confined to a single market, but in a complex system with nested dependencies, this wave of divergence typically extends across asset classes in periods of market instability.  In our hunt for Outliers, we are exploiting the opportunities far and wide associated with convergent traders whose models no longer work as market uncertainty spreads across the financial markets.

Of course, all trends ultimately end as these extreme price moves cannot continue forever. Markets need to become stable again to allow for the wheel of economics to turn. Usually extreme trends require massive intervention by Central bankers and financial intermediaries in attempting to restore order and stability to the market, but left to their own devices, this propensity of markets to shift towards chaotic behaviour is a sober reminder to why prediction has its shortfalls and why cutting losses short and letting profits run reigns supreme in the long-term success of a wealth manager.

Trade well and Prosper

The ATS mob

You must be logged in to post a comment.