Flexibile Asset Allocation With C(r)ash Protection

The "Conceptual sketch" posting presented a survey for designing a portfolio that generates stable profits during every type of economic environment the investor is faced with. Stimulating as well as  challenging comments were made providing food for thought on the building blocks for such a model. During our research we came across a paper published in late 2012 by Keller and Van Putten: "Generalized Momentum and Flexible Asset Allocation (FAA), An Heuristic Approach". The interested reader is encouraged to get acquainted with the elements of FAA.

Common asset allocation strategies (like the TAA strategy) are based on the so-called "momentum anomaly", which is known for centuries. The gist of the momentum anomaly is that assets often continue their price momentum, defined as the change in price over a given lookback period. Therefore one should buy assets with the highest momentum and sell assets with the lowest momentum.

FAA incorporates new momentum factors into risk regime determination. Next to the traditional momentum factor (R) based on the Relative returns among assets, Keller and Van Putten introduced Generalized Momentum by adding these new factors: Absolute momentum (A), Volatility momentum (V) and Correlation momentum (C). In their paper Keller and Van Putten demonstrated that by expanding the traditional momentum approach, portfolio performance increases compared to the buy and hold benchmark, both in terms of return as well as risk.

Summarizing FAA, Keller and Van Putten present their strategy with an example universe of 7 index funds. Applying a 4 month lookback, from this universe at the end of each month the top 3 assets are selected through a nested ranking process of these 7 assets based on relative momentum (higher is better), volatility (lower is better) and correlations (lower is better). Last, each of the top 3 assets chosen, has to pass the absolute momentum test: if their absolute momentum is negative, just go into cash. Capital is equally allocated over the top 3 assets or if applicable into cash.

FAA was scrutinized by Empiritrage. In their full report following findings are reached:
FAA has significantly higher risk-adjusted return than an equal weight portfolio. FAA decreases maximum drawdown dramatically. FAA is robust when adjusting look-back periods. Absolute momentum can directly add value on identifying down side risk regimes and decrease maximum drawdown.

Source: Empiritrage

Conceptual sketch for an "All-Weather-Portfolio" by deploying Adaptive Risk Parity

As stated in the title, the nature of this posting is conceptual. It is a survey for designing a portfolio that generates stable profits during every type of economic environment the investor is faced with. What is about to follow is partly a compilation of the information found in several sources* along with suggestions and idea's put forward by co-researcher "Ram". This post is also an open invitation to anybody who is willing to share valuable insights for improving the model.

The goal of this quest is not about generating the highest possible returns. Instead it is about creating a portfolio with a risk profile as close as possible to cash, but with yields much higher than cash. The idea is to reduce the portfolio's overall volatility by investing in assets that naturally move in opposite directions. One of the largest hedge funds in the world, Ray Dalio's Bridgewater, applies this investment philosophy in their "All-Weather-Portfolio".

AWP is based on four premises:
  • The future is unknown and impossible to predict.
  • Assets are responding to two drivers: economic cycle (expansion/contraction) and inflation (high/low).
  • Asset classes thrive not equally during each of the four scenario's.
  • Risk is distributed equally over the four scenario's.
Instead of allocating equal amounts of capital to each asset class, the AWP-model assigns equal buckets of risk to the distinctive asset classes. Thus "risk parity" for each quadrant is accomplished to match equal odds for the next economical and inflationary condition. AWP assumes the economy transitions randomly from regime to regime and thus the AWP needs to have one or more asset classes working well in each regime. So AWP seeks for one or more asset classes generating decent returns in each quadrant. For instance:
  • Stocks for economic expansion and low inflation.
  • Bonds for economic contraction and low inflation.
  • Commodities for economic expansion and high inflation.
  • Inflation linked bonds for economic contraction and high inflation

Economic contraction

Economic expansion


25% of risk:
- inflation linked bonds
- precious metals

25% of risk:
- stocks
- commodities
- real estate


25% of risk
- government bonds
- inflation linked bonds

25% of risk:
- stocks
- real estate
- corporate bonds
    Table I  

How To Time The Market? Stochastics Rulez!

What if there was an indicator signaling a market turn ahead of time instead after the fact?
How about "Sell into strength and buy into weakness"?
Sounds impossible and a fools play?

Actually, applying an extra ordinary market filter is the way to go. Utilizing a stochastic crossover strategy in an unconventional setup yields a 60+% probability of a reversal signal. In a "long only" play, the win rate even jumps close to 85%. What is to follow is all about compounding and risk management.

Updated chart

Made in Switzerland: a Global Market Rotation Strategy for MDY, IEV, EEM, ILF, EPP and TLT

[Update added: See Postscript]

On Seeking Alpha Frank Grossmann published his Global Market Rotation Strategy (GMR). The goal of the GMR strategy is to achieve above average returns while avoiding big losses during market corrections. To that extent TLT is added as a "safe haven" during roaring bear markets. The GMR Strategy switches between 6 globally distributed ETF's on a monthly basis:
  • US Market (MDY- S&P MidCap 400 SPDRs)
  • Europe (IEV- iShares S&P Europe 350 Index Fund
  • Emerging Markets (EEM- iShares MSCI Emerging Markets)
  • Latin America (ILF- iShares S&P Latin America)
  • Pacific region (EPP - iShares MSCI Pacific ex-Japan)
  • US Treasury Bonds (TLT- iShares 20+ Year Treasury Bond ETF)

The algorithm of the GMR model is quite similar to Kevin McGrath's TAA model posted in Februari 2013, except that the GMR model offers the possibility to allocate assets based on weighting not only past performance, but volatility too.

"Vixenator II" or how to turn "risk" into "opportunity"

What happens when "risk" is turned into a mimetic poly-alloy? We get Vixenator II and "risk" turns into "opportunity"! While working on a risk ratio, a statistical approach towards VIX and SPX turned out to be rather promissing.

SPX weekly chart
Back in January a statistical framework was provided for DMI, MACD and RSI. For creating Vixenator II the same approach is used. For the sake of brevity please consult the mentioned posting for the theoretical background.

Bell curve of "normal distribution"
By design Vixenator (I) was limited to daily charts. For the sequel a new composition is used resulting in an indicator way more mouldable than its predecessor: because of its "liquid elements" Vixenator II allows for application on all timeframes.

So, The Governator Is Back, Huh? Meet the Vixenator!

"I'll be back" and guess what? Yes, Arnold is back. Again, I might add. Anyway, the nickname "Governator" came into my head when I was introduced to the following neat indicator: Vixenator (thanks mr. T!).

But before introducing the Vixenator, I'd like to thank all my visitors for 50,000+ pageviews and their continuous support!

The original indicator was scripted some 4 years ago by "Osikani" and it compares the value of the VIX to the Implied Volatility of the SPX. What is Implied Volatility (IV)? In TOS speach:
Implied volatility is an estimate of the volatility of the underlying stock that is derived from the market value of an option.  Implied volatility is the volatility number that, if plugged into a theoretical pricing model along with all the other inputs, would yield a theoretical value of an option equal to the market price of the same option.
As "Osikani" put it: "Put it on a daily chart (the only chart on which it will plot), and see how it calls the market direction quite well: VIX above IV of SPX is a falling market; and the converse."

After 40,000 pageviews: HLC_Trend

Today the pageview counter hit the 40,000 mark. A big thank you to all my readers!

And now for something completely different after devoting a considerable amount of time to the TAA strategies. Let us take a look at an indicator based on the differences between a smoothed version of the close against averages of the high respectively the low: HLC_Trend.

Binary Market Timing based on XLI/XLU for EoM, EoW or EoD traders

Since the release of the previous post regarding TAA strategies, Aurelia and I have been working on the implementation of another market timing filter Kevin McGrath developed on the Stockfetcher forum. By using the XLI and XLU etf pair as proxies for bullishness and bearishness, the allocation strategy is simplified down to a binary process.

To SPY or not to SPY?

Thanks to Aurelia's continuous support in the development of the TAA strategy I have the opportunity to present a variation of the rotation system. Let's postulate that the stock market has two states: greed against fear or "risk on" opposite to "risk off" mode. When risk is in vogue stocks tend to rise and bonds drop, while in corrections, the opposite happens: investors flee from stocks into bonds.

Aurelia observed that Kevin McGrath's basic TAA model (see prior postings) with EFA, IWM, SPY and TLT, reflects these two risk stages. When risk is "on", the strategy allocates into IWM (or EFA) and when risk is "off" funds flow into TLT. However when the strategy indicates to rotate to SPY, the market predominantly seems to be in the middle of a risk transition. Meaning the selection of SPY is indicative for a change in risk acceptance:
- risk shifts to "on": TLT > SPY > IWM or
- risk shifts to "off": IWM > SPY > TLT.
See the upper sub pane window on the chart. Notice the (negative) profit contribution for SPY too (9x: selection of SPY for 9 months, code snippet courtesy of ST. Thanks!)

New milestone: 30,000 pageviews

Not even two months have passed since the site stats of this blog showed 20,000 pageviews and now the 30,000 mark is hit already. Thank you for dropping in!

While visitors at first mainly came from the traditional English speaking coutries, since a few weeks the audience is listening in France, Germany, Sweden, The Netherlands and even Ukraine too, to mention only the higher traffic countries (next to Australia, Canada and the USA of course). See the Earth globe on the right hand side for total visitor distribution from all over the world. For some reason that yellow spot in the big blue Pacific always makes me smile :-)

La bienvenue Willkommen Välkommen Welkom Ласкаво просимо

Please feel free to post in the comment section. If you feel more comfortable, do not hesitate to write in French or German. In those languages I should menage, with some help from Google translate if needed, but especially my Українська мова is a bit rusty though ;-)

I would love to hear from you how this blog became so popular among you people.  Where did you find the bread crumbs that did lead you to this blog in the first place? Please do post links to referring forums and sites.

So thanks again and goodbye, au revoir, auf Wiedersehen, hej då, tot ziens, до побачення !

Monthly Tactical Asset Allocation System: Equity curve added

Writing the thinkscript code for an equity curve for the Tactical Asset Allocation System posed quite a challenge. Scripting a dynamical all in system seemed like solving the chicken and egg dilemma, but finally the study is ready to be published. The TAA-system is explained here.

Tactical Asset Allocation: 28, 29, 30. Rotate!

How about a long term trading system that needs only some 30 seconds each month to keep track? Amazing, but now by all means doable. Thanks to Kevin McGrath's Tactical Asset Allocation system. Kevin claims a combined capital growth of 70+% over the last five years and he has backtested the living daylights out of his system settings as well as the ETF selection.

In short: for this particular long only TAA system one determines at the end of each month the relative strength of four ETF's: EFA, IWM, SPY and AGG*. The ETF with the highest reading is considered strongest. That ETF is the pick for the next month and all assigned assets are then rotated accordingly. Even better: now thinkscript takes care of the monthly system evaluation. The only action for the user to do is pull up the chart and next enter one sell and one buy order. Nothing more, nothing less. Extremely simple, extremely effective!

Milestone: 20,000 pageviews

Based on current visitor traffic my pageview stats will hit the 20,000 mark during the upcoming weekend. Of course this site is small fry in the blogosphere, but to me this is a great encouragement to continue posting. At the turn of the year the count was 16,000 and now, only some six weeks later 4,000 more again. A big thank you to all my visitors!

Guestpost by Lar: KC_PercentK

TrendXplorer has been so generous in sharing his scripts, I thought I would share something in return I have been working on with him recently. It is a trend indicator based upon the Keltner Channel, with a Bollinger PercentB like concept, I have named KC_PercentK. However, it uses a short term (6,0.5) and a long term (20,1.5) PercentK to determine the trend.

Sigma levels for identifying bullish and bearish mode with z-scored momentum oscillators like RSI or TSI

As a sequel to my post of last Sunday about a statistical framework for technical indicators, this post is about a technique how momentum indicators help to identify a bull or a bear market. For this technique we'll assume that a momentum indicator can sustain a bullish or bearish mode for a prolonged time. This behavior was introduced to me by Zev Spiro CMT in his daily market letters. Zev distinguishes a bullish mode when the RSI breaks above the bearish resistance level of 65. When in bullish mode the 40 level acts as bullish support. So during a bull market the RSI may fall down to this 40 level as long as it doesn't cross below this level. When RSI breaks below the bullish support level of 40 a change of trend is assumed and RSI has shifted into it's bearish mode.

See the chart below for an example on a long term weekly chart of SPX. Instead of the usual RSI the z-scored version is used and please note the support and resistance sigma levels are manually calibrated, unlike Zev's fixed levels.

Providing a statistical framework for DMI, MACD and RSI

After 16,000+ page views in the last year, I'd like to take advantage of this opportunity to thank all of my visitors and followers for their continued interest and support. And while today is the Twelfth Night or Epiphany it's also just in time for wishing everyone a profitable, prosperous and healthy New Year.

Now over to the subject of this posting. In 2010 a former software engineer at Intel Corporation by the name of Michael Gutmann published an article on "A Statistical Approach to Technical Indicators". Gutmann also wrote a book titled "The Very Latest E-Mini Trading: Using Market Anticipation to Trade Electronic Futures."

In his article Gutmann illustrates a method of converting technical indicator outputs from asset-specific values to statistical measures of price extension and compression. He stated that the results of such indicators can be used across markets without modification.

Gutmann explains that standard deviation is the most common method of estimating the spread, or dispersion, of a data set:
The data’s mean, or average, referred to with the Greek letter μ (“mu”), is first calculated. The data’s standard deviation is then calculated as the average distance of the data set from its mean. The standard deviation, referred to with the Greek letter σ (“sigma”), is easy to work with because it takes values that are the same units as the original, underlying data.
The science of statistics has determined that for many naturally occurring populations (population height, weight, test score, etc.), data is “normally” distributed about its mean. This is the well-known bell curve of population distribution. Interestingly, bell curves are completely defined by their mean and standard deviation. This allows one to say that a normal distribution has approximately 70% of its data contained with one standard deviation of its mean and 95% within two standard deviations, regardless of the values computed for the data’s mean and standard deviation.