Portfolio, TAA Investing

IVY timing model extensions – weightings & momentum

Today I’ll cover a few other extensions to the IVY timing model beyond those I covered in my last post. The source of these extensions is the same, Mebane Faber’s IVY update outlining these approaches. The extensions I’ll cover are varying the weighting of the asset class mix and taking advantage of momentum in the asset classes to increase performance. Lets jump right in.

The first extension is to change the asset class weighting of the GTAA 13 portfolio covered in the last post to include more bonds (40% vs the 20% in GTAA 13) and change the default cash investment from t-bills to 10 yr govt bonds. This portfolio is called GTAA Conservative and now the original GTAA 13 allocation is called GTAA Moderate. The GTAA Conservative allocation is below.

GTAA Conservative allocation july 2013

I’ll cover the performance results of all these extensions together in a bit but lets cover the last extensions first. The last extensions use the recent performance (momentum) of the GTAA Moderate asset classes to vary the weighting of the asset classes. The original background paper outlining the basics of these momentum strategies can be found here. This extension takes GTAA 13 moderate asset classes and ranks then by the average of the return over the last 1, 3, 6, and 12 months. Then an equal weighted portfolio of the top 6 ranked asset classes is formed. Also, the asset classes are only included if they are above their 200-day SMA.  Otherwise, that portion of the portfolio stays in cash. This is called the GTAA Aggressive Top 6. Also, an even more aggressive portfolio is formed with the top 3 asset classes. This is called GTAA Aggressive Top 3. OK. Now lets take all these extensions of the IVY timing model and see how they perform. See table below.

GTAA extensions con mod agg performance july 2013


As a reference point, the GTAA 13 MOD portfolio in the table above is the same portfolio as the GTAA 13 portfolio in the table in my previous post. The conservative portfolio is not much different than the moderate portfolio and doesn’t seem to offer any advantages over the GTAA 13 MOD portfolio. The aggressive portfolios that take advantage of momentum have markedly higher performance, 5% to 7% extra returns per year. The downside is higher volatility and about double the drawdown of the GTAA 13 MOD. Also, the downside is the increased difficulty in implementing these portfolios by keeping track of the all the signals.

In short, the aggressive extensions to the already extended 13 asset class version of the basic IVY timing model offer increased performance but with higher drawdowns. And also with increased complexity. But, hey, for 5% to 7% a year over the already great GTAA 13 MOD portfolio which already offers 1.56% more per year than the original IVY Timing (GTAA 5) portfolio it may be well worth the effort for some. My only caution would be to really think about if you can handle the larger drawdowns. From my experience most investors can’t. They say they can, particularly during rising markets, but when the rubber hits the road they run for the hills and abandon their investment approach.

Full Disclaimer - Nothing on this site should ever be considered advice, research or the invitation to buy or sell securities. These are my personal opinions only.

About paul.novell@gmail.com

19 thoughts on “IVY timing model extensions – weightings & momentum

  1. So, since the default cash investment is now 10 yr govt bonds instead of t-bills, I guess you ignore any sell signal (when the monthly close drops below the 10-mo SMA) on the 10 yr govt bond portion of your portfolio?

    Also, I assume that using the 10 yr govt bonds for cash also applies to the Aggressive portfolio’s?


  2. Just got news letter from Mr. Faber (I invested in his new etf). Amazon has his new book on sale – free. Our new book, Shareholder Yield, is available for free download on Amazon for the next five days.  If you haven’t read it, pick up a copy this weekend (you can read on iPad, Kindle, computer, etc).

  3. Paul,
    I’m trying to replicate your results and am having trouble accessing index data. GFD wants $30K for a subscription and I’d rather invest the money. I am willing to pay for data, but not that much! Where do you get your data from?

    1. Adam, yeah the data providers charge an armload for their data. My data is pulled from a variety of free sources. All the IVY data comes from Mebane Faber’s paper, VBINX data from Vanguard, Permanent Portfolio data from the Crawling Road blog, SP500 and 60/40 data from various web sites – Professors Shiller, Damodaran, French. It was a pain to collect. In my next major update on the IVY returns at year end I plan to make my data available on Google drive. Maybe sooner if I find the time to get it more organized and presentable to the public.


      1. There’s no magic bullet then? I’m looking into accessing databases through the library and my former university. If I find anything I’ll let you know. a.

        1. Nope. No magic bullet. Pure slave labor I’m afraid….. I’ll try and get my files cleaned up soon and make them available.


      2. Hi Paul,

        Thanks for another very insightful post.

        Wondering if you had a chance to clean / share the data used for your backtest?

        Glad to help with cleaning the data if you need an extra hand.


          1. I did – I thought you had access to the monthly timeseries to reproduce Faber’s models.

            However have you tested the ETF’s in your spreadsheet vs. the indexes used in the Faber backtest?

            Or do you think they are a close enough approximation of the Fama-French portfolios?

            On Tue, Nov 25, 2014 at 5:34 PM, Investing For A Living wrote:

            > libertatemamo commented: “Trevor, are you referring to the IVY > portfolios? Paul”

          2. No I don’t have the monthly time series. I have the data he presented in the paper and that’s all.

            I have tested the ETFs as far back as their history goes, which is not very long. But they match well with the indices used in the paper.
            The are a close enough approximation of the indices Faber uses. Fama French data series is quite different and is for different puroposes.


  4. hi Paul,
    would like to know how you weight the monthly rtrns for the AAG versions (EW?) and would the results be the same using the 10m-ma rather than the 200d- ma?
    great blog,

    1. Andre, yes all the AGG portfolios are EW and the results are practically the same if you use the 10mo SMA vs the 200 day SMA. Me Faber has answers to all typed of nuances in the FAQ on this site.


Comments are closed.