Strategy 1 – Weekly Tactical Asset Allocation (WTAA)


Strategy1 has been derived from the article “A Quantitative Approach to Tactical Asset Allocation” by Mebane Faber. Check his website and you will find the link to download the paper.  The system trades a universe of 6 ETF’s representing various asset categories (listed below). It will trade once a week by allocating 100% of its equity to the fund that has the highest momentum (combination of 3 ROC settings). To filter out turbulent market conditions the RSI of the VIX index must be below a certain level. Trading signals will trade the Friday Market Close (MOC) and exit on Thursday’s Market Close (MOC).

The picture below summarises the performance of System 1. The format has been inspired by the setup used by David Varadi over at CSS Analytics.

  • This system has been tested for the 7 ETF’s over the period 1 Jan 2003 – 18 June 2010. The start date of 1 Jan 2003 was chosen as most of ETF’s had some trading history by that date.
  • The stats for the system are based on a fixed positionsize of $10.000. The results do take into account transaction costs based on Interactive Brokers fee structure, however slippage costs were excluded.

As of the last week of June 2010 the signals generated through this system will be provided live before Friday morning.
One can sign-up to this blog to receive the signals.



















  • EEM – Emerging Markets
  • EFA – Foreign Large Caps
  • GSG – Commodities
  • IEF – US Treasury Bonds 7-10 YRs
  • SPY – SP500
  • VNQ – Real Estate (US)
  • GLD – Gold
  • 16 thoughts on “Strategy 1 – Weekly Tactical Asset Allocation (WTAA)

    1. My compliments for the blog, Quanting Dutchman! I really appreciate your openness about the strategies. It really helps (me) to see things from a different perspective. Currently I’m in the process of developing intraday arbitrage strategies, but have also some concepts for lower frequency trading. If you are interested, we could exchange some thoughts and ideas, just take a look at my blog and drop me an e-mail (could be in Dutch ;-).

    2. Hi Quanting Dutchman,

      Can you/will you publish the calculations used for the signals WTAA generates (i.e. open-up the black box)?


      – another quanting dutchman 😉

      • Hi Mark,
        At the moment I have no plans to post further on the WTAA system (too many other projects cooking).
        Maybe in some time I will come back to the system and do some further posting.
        ps. drop me an email if you are interested in exhanging ideas in Dutch 🙂

    3. Hi Quanting,

      congrats for the high quality blog, really appreciate the analysis. One question here though, as VIX could be very unstable and volatile, when you say “To filter out turbulent market conditions the RSI of the VIX index must be below a certain level”, do you take i) a fixed period on the RSI or a dynamic one ? , ii) a fixed threshold on the RSI above which you go 100% cash or a dynamic one ? and iii) do you only trade on Thursday MOC and Friday MOC or do you exit position during the week when the VIX Filter says so ?
      Keep up the good work !

        • Hi there,
          very interesting post. I tried to replicate this on my own, with a bit of success, however using different set of 3 x different ROC, i can achieve good returns (between 15% and 20%) but can not get rid of a 15% drawdown. Any insight about your way of calculating momentum criteria ?
          A humble curious!

    4. Hi,
      from what I saw here, I’ve had the feeling that the ROCs were kind of the following type : MA(returns, period1)/Stdev(MA(returns,period1),period1) * MA(returns, period2)/Stdev(MA(returns,period2),period2) * …
      nope ?

    5. The strategies posted by quantingD. are fantastic! But I have a question, in which I am lately thinking.
      What about walk forward studies?
      I’ve seen some trader-guru-blogger that says he does not make walk forward studies, prefer to understand the algorithm.
      But Howard Bandy, the Amibroker books writer, says that backtest optimized are useless, only walk forward studies are predictive.
      What do you thing, Dutchman? Have you used the walk forwards in your strategies?
      Best regards!

      • Hi Gonzaga,
        I have read somewhere that strategies will have on average 25% worse performance after “go live”. The backtest is usually too good as result of optimization and/or data snooping. I must admit that when I designed the WTAA strategy I was less aware of the importance of these pitfalls.
        Now on the walk-forward theory, I have read the books of Howard Bandy and can see the benefits of the WF approach. I can do an analysis and apply it to this strategy. Any suggestions what parameters to optimize with a WF?
        – QD –

        • Hi QD!
          In my experience, systems profits fall too much when using and in-sample period of few months (talking about slow systems). At least, I have to use an in-sample period of 1 year, and usually I use periods of three or four months out of sample.
          But at this moment I don’t know if the WF studies are always a smart study..
          The parameters to study, well I think should be those more relevant to the system. Perhaps one of the ROC’s, and another walk forward study with the level of the RSI of the VIX.
          And perhaps is good to try one walk forward 5 years-insample, 5 years out-of-sample.. to see what happens..?

    Leave a Reply to Mark Cancel reply