Dual Momentum Permanent Portfolio (DMPP)

[ochotona]

Actually, I am posting the signals for the simpler, Absolute Momentum way to modify the PP in the thread "Trendfollowing signals for the PP". Cash and bond buy and hold, only turning gold and stocks off and on. Next update on 2/25/16.

[Pet Hog]

Thanks for the quick response.  It seems like my understanding of the process was close.  You didn't address the GEM concept.  Would you advise ever owning international stocks or a total bond market fund in place of US stocks?  That seems to go against the HBPP principles.

Using Absolute Momentum to trigger rebalancing is another application altogether. Let's say you didn't want to take the risk of going all-in, all-out on gold and stocks. You could rebalance when gold and stock momentum triggers go off (when they change to above or below the yield on cash), to maybe take advantage of a possibly propitious moment to buy or sell.

Just to be clear, here you are talking about rebalancing a vanilla PP back to 4x25 -- you are never swapping stocks or gold for cash.  Is that right?  I don't usually follow the discussions about momentum and other forms of market timing; excuse my ignorance, but has this approach been modeled?

[ochotona]

Thanks for the quick response.  It seems like my understanding of the process was close.  You didn't address the GEM concept.  Would you advise ever owning international stocks or a total bond market fund in place of US stocks?  That seems to go against the HBPP principles.

Just to be clear, here you are talking about rebalancing a vanilla PP back to 4x25 -- you are never swapping stocks or gold for cash.  Is that right?  I don't usually follow the discussions about momentum and other forms of market timing; excuse my ignorance, but has this approach been modeled?

CraigR did address owning foreign stocks in the PP, for up to 5-10% of the portfolio (p. 82) depending on user preference.

The GEM is discussed here.

No, I am actually talking about swapping gold to cash and stocks to cash if conditions get bad! Dr. Mebane Faber did it already, shown here. The results are pretty good! You get S&P500 returns with Permanent Portfolio volatility. The chart y-axis is logarithmic, so the wealth difference between PP and "timed" PP is actually very large.



The idea behind trend-following the PP is... yes, you need gold, equities, cash, and bond... but you don't need them all at the same exact time, all the time.

[Pet Hog]

Ah, it's getting clearer: you are using GEM merely as an indicator for timing your change in assets between stocks and cash.

Faber's chart appears at first glance to suggest that a momentum PP is vastly superior to a vanilla PP, but on second look I don't see much difference during the 1970s or from about the mid-80s onward.  Its glory period was from 1980 to 1982, where it really excelled.  That was a volatile time for all the components, so I guess the momentum approach avoided some big losses (e.g., holding gold in 1981).  Another thing to note is that he is considering monthly rebalancing also for the vanilla PP; I would like to see a chart for his momentum strategy against that for a vanilla 35/15 PP before concluding that it is indeed a better strategy.  Perhaps I'll work on that myself!

Thanks for your help, ochotono!

[ochotona]

This is the Dual Momentum GEM.

[rickb]

Faber has neither a PhD nor an MD or other professional degree, so I'm not sure why you refer to him as Dr. Faber.

I suspect he's thinking about Marc Faber (who does have a PhD).

[MachineGhost]

After spending all day and night yesterday hacking Antonacci's Global Equities Momentum (GEM) model, you would be forgiven in thinking that the Dual Momentum concept adds a huge edge to any portfolio.  That is not the case.

To make a long story short, GEM is an example of the post hoc fallacy or in more layman terms, data mining.  The same exact criticisms I've applied elsewhere to the so-called value and size effects also applies here.

What drives the outsized gains in the GEM is the inclusion of essentially emerging markets exposure.  In the 1970's when the dollar was depreciating like mad, foreign equities were the emerging markets of its time and acted in an inverse relationship to the falling USD.  In fact, if you replace foreign equities with gold, there is essentially no performance difference (but you do wind up with a -43% maximum drawdown off a gold's peak.)

Around 1990, gold ceased to play its former role as a USD hedge and the torch shifted to the US for its dot.com bubble.  It is not until the early 2000's that the then emerging markets once again plays an outsized role, this time for another massive bubble it still has yet to recover from.  While gold did go up in this timeframe as well, it was no longer the main driver of returns vs emerging markets.

There's also a smaller effect from the bond bull market but it only adds a couple percent of CAGR's.

Bottom line: Curve fitting the symbol group to only those securities that had outsized returns in the past is pure data mining.  The past does not represent the future.  In fact, GEM absolutely sucks when run on other symbol groups.  Relative momentum alone is not enough to guarantee you will lasso the unicorns.  If only it were that easy, we'd all be super-rich.