Working with different Market Regimes Part II

Based on a refined, trustworthy investment philosophy, I return to the practical implementation of a key question:
How is successfully adapting an asset allocation to market conditions possible and can it yield higher risk-adjusted returns than a static portfolio?


I concentrate on the third basic building block of my investment philosophy – systematic downside protection through dynamic asset allocation – and analyze why market regimes have the tendency to stick around long enough to make it worthwhile adapting your portfolio to them – even if it is impossible to forecast changes accurately. My aim is to build an integrated framework of different concepts and measures for market regimes, to dynamically allocate to different assets, factors and systematic approaches in such a way, that my portfolio produces superior returns over the long term with tolerable volatility and drawdowns in all market environments. I favor simplicity and robustness over complexity and sophistication – I believe adaptive allocation should be a fairly straightforward process, that makes sense intuitively.


Because the financial markets are inherently uncertain, I don´t think it makes sense to be overly exact – this is not science, but aims to give a practical rule of thumb to adjust general allocation percentages. I use a mix of fundamental and technical data in my process as I think that paints the most insightful picture while providing unambiguous signals for practical implementation.
To make this easy, without the need for extensive programming skills, I look for tools that are convincing, easy to use and either free or very cheaply available.
The time horizon employed is a crucial factor. In my global asset allocation take a long-term view and try to capture multi-year market shifts to avoid drawdowns higher than approximately -15% in most asset classes.
Additionally I use more sensitive warning signals to adjust exposure to the systematic approaches that work on a shorter time horizon (e.g. short volatility or medium-term trend strategies).


Definition of six market regimes
The types of markets I categorize are:
  • Moving up
  • Moving sideways
  • Moving down
and combined with that
  • High volatility
  • Low volatility


In addition to these technical market definitions, fundamentals inform the long-term outlook for different assets and make it possible to enhance technical signals by long-term probabilities.

Expected returns vary with time
Various asset classes, factors and systematic approaches do better or worse in different market types. Regimes can be diverse and change at different times for each asset class – as a rule of thumb correlations of risky assets rise in a volatile down market when risk- off behavior kicks in. As my portfolio is mostly exposed to the long-term positive risk premia the market has to offer, it automatically has an upward bias and is therefore concentrated on the high-probability long side. In a negative market regime (volatile down movement) exposure is reduced and more cash held; in a positive environment (quiet up movement) moderate leverage of about two times is employed.
The fundamental idea is to take different risks at different times. When expected returns are high and corresponding risk is lower, I want to start investing aggressively and when expensive valuations or indications for a shrinking economy are supported by negative technical indications, I want to switch to capital preservation mode.


Are market regimes persistent?
The basic premise for me is that predictions generally are not reliable enough to be useful in practice. Therefore the most promising way to attempt to time the market successfully is by adjusting your portfolio allocation to the current state of the market. This can only work if market regimes last long enough for the current allocation to become profitable before the regime changes.
I think a variety of studies of market anomalies point to that conclusion:


1 Momentum effects – these point to prices exhibiting trend behavior for an extended period of time, before mean reverting to long term averages – exemplified in John Maynard Keynes´ adage: “The market can remain irrational longer than you can remain solvent.“
  • Short-Term Momentum (1-month) – exhibits a reversal in returns
  • Intermediate-Term Momentum (3-15 months) – exhibits a continuation in returns
  • Very Long-Term Momentum (1 to 5 years) – exhibits a reversal in returns: Trends start to break down as mean reversion takes over and the market returns to fundamental values and often overshoots.
2 Volatility clusters – all the way back in 1963 Benoit Mandelbrot observed a certain pattern in volatility variation, which he summarized as follows: ”Large changes tend to be followed by large changes – of either sign – and small changes tend to be followed by small changes”. This is especially pronounced in low volatility environments.
As low volatility regimes are usually coincident with up-trending markets and high volatility regimes with markets moving down, a clustering of volatility points to a persistence of up- and down-trends.


3 Fat tails and return asymmetry in financial price series – returns, especially of stocks, exhibit fat tails (meaning non-Gaussian, sharp-peaked and heavy-tailed distributions – with a higher probability of extreme returns) and negative skewness (an increased probability of big losses, because of a fatter tail on the negative left side).
This implies, that a method that protects by cutting losses before the fat left tail (e.g. a stop loss or a trend method) can earn above average returns over time.


Ways to identify and use market regimes
I propose a simple point system using three inputs (for each I found simple online tools): volatility regime, trend and fundamentally expected return. Each input generates a positive (+1) neutral (0) or negative (-1) value, which add up to a score between -3 and +3 that can be translated into a gradual portfolio exposure and allocation adjustment for each asset class separately. Such a non- binary approach is much easier to implement than a complete switch in and out of asset classes. Here are the practical example exposures for different scores that I use (intermediate scores change gradually between extremes):
  • -3: negative – exposure is reduced to a predetermined minimum buy and hold allocation (for example 15% for equities and 10% for real assets and 10% for fixed income) and a higher cash percentage held.
  • 0: neutral – exposure is 100% of the basic portfolio allocation for each asset class (e.g. 33% for each respective asset class when using an equal weight allocation as the neutral base allocation).
  • +3: positive – exposure is leveraged to the predetermined maximum (2x to 3x leverage in my case, but of course you can just set it to 1x or any other value) – leverage is predominantly set by the volatility regime and the number of asset classes in an uptrend.


1 Volatility Regimes:
Using volatility as a market timing tool is still quite uncommon, but becoming more popular and a growing number of academic papers on the subject are available. Most studies use complex math (which I don´t have a sufficient background for), but to extract the essential information a simpler approach using the S&P 500 volatility index VIX is also possible.
Volatility as a market timing tool functions like a risk- on / risk- off switch between risky assets (e.g. equities, volatility strategies) and safe haven assets (cash, bonds, gold, hedges).
We want to be invested at full leverage in risky assets in a low volatility regime and hedged or moving into safe assets in high volatility regimes.


To quantify the timing switch, we take advantage of certain properties displayed by volatility:


  • Volatility tends to spike briefly (usually when the stock market slumps), followed by lengthier downward trends.
  • Volatility returns to its long-term mean (mean reversion effect), which is reflected in different prices for volatility futures with different expiration dates: the Volatility Term Structure.
  • Volatility forms volatility clusters (regimes): the best predictor of future volatility is current volatility.


Compared to absolute VIX levels, the VIX Futures Term Structure incorporates additional useful information as it takes not only the absolute levels but also future expectations for the VIX into account and, because volatility futures can be used as a hedging tool, it also incorporates levels of investor´s demand for insurance against market declines. It provides a distinct quantitative value whereas absolute VIX levels are not as useful (e.g. it makes a big difference if volatility shoots up from a low level to a VIX of 14 or reverts down to it from a higher level).


We can use the slope of the term structure as a timing indicator. Around 75% of the time the futures with a longer time to expiration are more expensive, the term structure slopes upward, which is called contango. When the futures closest to expiration are more expensive, the structure is in backwardation, sloping downward.


  • Upward sloping Term Structure points to a bull or flat market
  • Downward sloping Term Structure points to a bear market


A great tool to put these insights into practice can be found at The main page shows the current term structure and the tab „contango“ a historical chart of the price difference between the second and first VIX future. Whenever this value is positive our risk-on switch gets a value of +1. To avoid some whipsaws around the zero line we go risk-off (score -1) at a contango value below -1,5%.


Additional volatility ideas:
  1. Volatility can also be viewed as an alternative asset class with its own independent source of return – the volatility risk premium – this is the basis for the short volatility strategies in my portfolio. The same timing mechanism can be used.
  2. Volatility timing deals primarily with the issue of how much to invest in equities. For bonds and real assets it makes sense to stress fundamental information (see below) to decide on allocation size.
  3. The historical normal situation of positive correlations between stocks and bonds changes in volatile times. The correlation even switches signs from barely positive to significant negative correlations, which means that bonds are a great safe haven asset.
  4. In high volatility regimes trends are less reliable and harder to trade because of wild market swings. As they only occur about 25% of the time the probability for a switch from high to low volatility is much higher and so it is better to radically reduce exposure and wait for high volatility to pass before investing preserved cash once again. Drawdown reduction is the main goal.


2 Momentum
I use a trend indicator – also called absolute or time series momentum – as a simple portfolio filter. It is intertwined with volatility, but on its own it carries the greatest weight of all timing indicators. In a recent post I describe in detail the validity of a trend approach including the different methodologies one can use and the performance profile and difficulties one can expect.
As a tool any website that displays charts with rudimentary customization (e.g. Yahoo Finance) is perfectly sufficient. A commercial platform that is affordably priced is TradeStops, which uses a volatility based trailing stop loss strategy, which gives similar signals as the moving average strategy described below and has the added advantage of sending out automated email alerts.
My filter is long-term and used on all assets in my portfolio: a 275-day moving average of price (as the 200 day SMA is very widely used and prone to whipsaws, I distance myself from it a little bit).
I want to hold an asset when it is in an uptrend with a price above the moving average (+1 for the point system) and reduce my allocation when its price falls below its moving average (score -1).
That´s it – extremely simple, but very powerful as it effectively cuts the left tail of the return distribution. It works, because financial assets display a non-normal distribution of returns with pronounced extremes that is skewed to the left – more large losses happen than in a normal distribution. While keeping absolute returns at the same or a slightly higher level, cutting these infrequent large losses substantially reduces volatility and drawdowns leading to higher risk- adjusted returns – the holy grail in investing.



3 Fundamentals
Fundamental analysis is the mainstream methodology used by market timers – for example by value or growth investors. It is hard to use successfully, because it is a very competitive environment and a myriad of complex, detailed and varied methods exist and can be exploited. I pick out the ideas that I find especially rewarding in combination with the technical indications of volatility and trend. Most successful seems to be the powerful concept of mean reversion, while extrapolating strong fundamentals into the future is often a losing game of performance chasing.
“From financial history and from my own experience, I long ago concluded that regression to the mean is the most powerful law in financial physics: Periods of above-average performance are inevitably followed by below-average returns, and bad times inevitably set the stage for surprisingly good performance.” – Jason Zweig


In general fundamental analysis is a very blunt timing tool as extreme fundamentals can stay extreme for long periods of time. It is best suited to set long term return expectations and influence the tilt of a portfolio´s asset allocation.


Valuation is the main fundamental criterion used by many investors, but it tends to play out over rather long time frames above three years.
There are a variety of ways to measure it in different asset classes – it boils down to the common sense principle “buy low and sell high”. Good tools blend multiple expected return models: Research Affiliates have recently launched a interactive asset allocation platform which shows the expected long term return for each asset class depending on valuation measures like Shiller´s CAPE ratio. For value based international stock market expectations going country by country Star Capital publishes an up-to-date list.
For a one click ETF portfolio including overall exposure {I}Cycle Engine uses a variety of fundamentals to weight assets dynamically anticipating developments of the market cycle.
A robust neutral allocation is to equal weight asset classes: 33% equities, 33% bonds, 33% real assets & cash. Then tilt towards the assets (and sub-assets, e.g. countries) with the highest current long-term return expectations and reduce exposure to overvalued areas.
When using return factors (e.g. value and momentum) as a component of asset selection it is good practice to pay attention to valuations of different factors in comparison to historical means and avoid overpaying for it. The value factor itself, for example, can be relatively expensive or cheap. Research Affiliates provides good information and background on that as well.

Equities – Leading indicators of economic growth
For equities leading indicators of economic growth can be used to judge the probability that a correction may turn into a bear market. The stock market moves in the general direction of economic growth over the long run and its upward bias is very pronounced. Corrections are most likely to be temporary, unless they coincide with a shrinking economy. Severe bear markets (of -40% or more) occur in the context of economic recessions and, even though we witnessed two of them in the last two decades, they are actually very rare and only happened five times since the 1920s.
Most useful are shifting trends in leading economic indicators as, for example, unemployment rates or the treasury yield curve. Changes in GDP itself are a lagging number and quite useless as asset prices will have already reacted to changes in economic conditions while they occur.


Fixed Income and Real Assets – Interest Rates and Inflation
For bonds there is a straightforward way to get a relatively reliable future return profile: It is simply the current bond yield. This can be used to set a basic allocation percentage – for example at the present time extremely low interest rates make for poor future prospects and I set my allocation at a minimum.
As bond returns are inversely related to inflation, poor future returns may be compounded and safe haven characteristics lost in an inflationary environment. This is the time when real assets, like commodities, usually show the best returns and should be overweighted.
Low commodity valuations can be judged by looking for multi- year bear markets. An interesting value proposition appears when commodity prices fall below production value and a rebalancing of supply and demand is triggered as production is reduced – which will cause prices to eventually rise again, but this process may take a very long time.  Roll yield / Carry is an important commodity specific return component which should be taken into account when making individual commodity investments – a good methodology would be to select commodities with low valuations, in an up-trend and with a positive roll yield.


Integration of fundamentals with the point system
The complexity and amount of information in fundamental data makes it much harder to derive a simple point score. I use a combination of two approaches: valuation will yield a score of -1, 0 or +1 (high, neutral, low). All additional information is blended into the basic weight of portfolio allocation to asset classes and to shift the probabilities of technical signals to be correct.
Apart from fundamentals many other criteria (e.g. personal investment goals and risk tolerance, age, available capital etc.) play a role.

As a current practical example, I describe my portfolio allocation in detail here.


Leave a Reply