Improving on the Permanent Portfolio

[craigr]

Are these actual performance numbers you achieved yourself with real money beforehand or is it only after-analysis?

[Stefan]

MG: 17% MaxDD is a lot of pain for 6.7% CAGR. Not to mention 20% MaxDD for a 9% CAGR. I am not sure you want to trade that. You are way better off using the standard HBPP. PRPFX had 11% CAGR with 21% MaxDD over the same period.

I think you can achieve a 7-8% MaxDD for 7% gain by using a risk parity allocation, with well known TF methods (nothing proprietary) as I mentioned in my prev posts - if you really want to use timing and write a program for TF and position sizing.

[MachineGhost]

These are simulated, with commissions deducted.

Are these actual performance numbers you achieved yourself with real money beforehand or is it only after-analysis?

[MachineGhost]

Those were annual returns, not CAGR.  PRPFX is weighted more to commodities and foreign resource stocks.  The standard "adolescent" PP was what earned 9.01% yearly with 20.50% drawdown.

Replacing domestic with 12.5% foreign stocks will increase the "adolescent" PP to 34% maximum drawdown, but it will certainly increase returns a bit in both portfolios.

It looks like I will have to revisit the strategic allocation issue.  To maintain at least a 3.59% CAGR long term gain, the nominal cannot decrease below 6.79% CAGR.  7% CAGR isn't much slack to allow flunctuations.

MG

MG: 17% MaxDD is a lot of pain for 6.7% CAGR. Not to mention 20% MaxDD for a 9% CAGR. I am not sure you want to trade that. You are way better off using the standard HBPP. PRPFX had 11% CAGR with 21% MaxDD over the same period.

I think you can achieve a 7-8% MaxDD for 7% gain by using a risk parity allocation, with well known TF methods (nothing proprietary) as I mentioned in my prev posts - if you really want to use timing and write a program for TF and position sizing.

[MachineGhost]

This is mainly for archival purposes in this thread.

To minimize tracking error vs the CPI back to 1928, the following portfolio weights were optimal:

16.57% Stocks
38.94% T-Bills
14.59% T-Bonds
10.99% Real Estate
19.91% Gold

2.65% Real CAGR, 21.48% Real MaxDD (1948), 2.61% Inflation.

UPDATE: This portfolio is invalid.  The T-Bonds were actually using AAA 20-year Corporate Bonds.  I forgot I had done that.  Oops.

[D1984]

This is mainly for archival purposes in this thread.

To minimize tracking error vs the CPI back to 1928, the following portfolio weights were optimal:

16.57% Stocks
38.94% T-Bills
14.59% T-Bonds
10.99% Real Estate
19.91% Gold

2.65% Real CAGR, 21.48% Real MaxDD (1948), 2.61% Inflation.

MG, where did you find returns for real estate going back to 1928? NAREIT only goes back to 1970.

[moda0306]

I can't help but think that for most Americans that hold a good chunk of their wealth in real estate for a good chunk of their lives (often financed by the equivalent of a negative bond), that investing in even more real estate outside of that is not a good idea.

I know that homes work in a slightly different wavelength from rentals, but I think in the medium-to-long-term (5-20 years) they're going to move about the same, as the cost of housing defines how much people are willing to rent for, and vice versa.  As interest rates and home prices drop, the rent/own decision tree could become such that todays rental rates are way too high... A relatively small drop in rental rates could bring the earnings yield of these things crashing down considerably, considering the current profit margins (or lack thereof) they're experiencing.

I definitely think renters should be open to REITS in their retirement accounts, but as a homeowner I wouldn't touch them with a 40-foot pole.  Just my 2 cents for those planning on "diversifying" into REITS while sitting on a $200k home an a negative bond of $160k.

[MachineGhost]

MG, where did you find returns for real estate going back to 1928? NAREIT only goes back to 1970.

S&P/Shiller has the data going back to at least the late 1800's.

The weight was for real (non-equity) and unleveraged; obviously any use of leverage would reduce the weight proportionately.  I was surprised real estate had a role to play considering the real CAGR was only like 0.58%, the worst of all asset classes.  Perhaps it acted as a substitute when gold was illegal to own or export?

MG

[MachineGhost]

Going back to page 4 of this thread you discounted Decision Moose and Small Cap Value holdings - but staying with that theme my figures indicate that since 1926 something around 30% exposure to SCV alone would have been sufficient to uplift 100% of total fund value with inflation. For the PP the figure is around 44%. DM has limited history, but assuming 26% exposure to DM achieved inflationary uplift of the whole then a 30 SCV/44 PP/26 DM has three separate streams that each potentially uplifts the whole by inflation. If all three achieve that objective then total gains would be 3x inflation.

I have to discount DecisionMoose because it is a pure momentum model and not dependent on fundamental macroeconomics for deciding what asset classes to invest in.  Momentum is prone to failure and can go through extended periods of underperformance.  That kind of uncertainty is just too risky to invest large $$$'s on without a long history through different economic climates.

How do you have SV data back to 1926?  I don't have tilt data past Simba's 1972 to make such judgements.  But, some caveats: the Fama-French three-factor model completely disregarded a fourth factor -- momentum -- because they didn't know how to explain it and didn't believe in it anyway; also many studies and formulated indexes from that era considered "growth" stocks to be the inverse of value (i.e. overvalued), which is not the case in real-world applicability; so the historical "overperformance" of value vs "growth" is largely illusionary as true growth and value are neck and neck; one needs to pay very special attention to how indexes are composed to make sure they are not conflating the error.  Even when I quantitatively screen stocks for value, it is trivial to use the inverse of the rankings rather than setup a new spreadsheet specifically for growth factors; I posit that value that is going up in price is actually displaying growth characteristics, otherwise such would remain value traps.

On a separate note, it looks like IWM and SPY are overtaking BTTRX that DM is more recently holding and if things stay as they are currently I wouldn't be surprised to see DM switch into either one of those in the next week or two. You indicated DM had a -25% MaxDD in that earlier posting, but more generally its quite a low-down type method as I believe it uses a trailing stop loss and also jumps ship if relative strength falters (which is also a form of stop loss). With 26% in DM and 44% in PP you might consider that as a form of 70% low-down, leaving 30% SCV as the main drawdown risk/volatile holding.

DM doesn't use a trailing stop, AFAIK.  Stocks can lose a lot of money in just a week in-between switchings with no cash/bonds to counter balance, but thats par for the course.  The -25% MaxDD was from holding equities during 4Q 2009 to 2Q 2010.  It only hit a new high water mark (barely) in 2Q 2011 and its only up about 3% net to date since the peak in 2009.  That is going on 3 years of underperformance.  At what point do you decide to jump ship, now or when it has a MaxDD of -50%?  That is the problem with black box systems.

MG

[Reub]

Clive, I love that Dalai Lama quote!

[stone]

Clive, I love that Dalai Lama quote!

I second that. I also really liked Clive's previous Einstein quote too.

[MediumTex]

Clive, I love that Dalai Lama quote!

I second that. I also really liked Clive's previous Einstein quote too.

I saw this great news clip where the reporter had been interviewing the Dalai Lama and at the end of the interview he said he had a joke he wanted to tell him.

He said: "A man walks into a pizza shop and asks the guy at the counter, 'Can you make me one with everything?'"

The Dalai Lama looked at him stone faced.  No hint of a smile...nothing. 

The complete lack of response was much funnier than the joke.

[moda0306]

MT,

I saw that... I felt bad for everyone involved... but it was hilarious.

[MachineGhost]

I have differing figures for DM than you MG.

Thanks for pointing this out.  There are missing trades in the backtest I need to figure out why that is occuring.

MG

[MachineGhost]

Here's the latest update of my research.  I have settled on several broad portfolios to track going forward.  They are:

HBPP: The traditional Permanent Portfolio as envisioned by Harry Browne.  Return: 8.16% CAR, -15.01% MaxDD.


MTPP: Same as HBPP, but using market timing for stocks, bonds and gold for the entire duration starting Sept 1977.  Return: 9.79%, MaxDD: -7.76%.


Because the MT portfolios may have more transactions than annual rebalancing, I have deducted commissions and slippage from all portfolios.  

Cash is modeled as an average 6-month CD interest rate (virtually the same as a 5-year Treasury ladder) which will overstate the interest income earned in low rate periods such as currently.  I decided to do it this way instead of using a fund which are prone to negative capital gains and fees that cash doesn't suffer.

MG

[MachineGhost]

Clive, where can I learn more about the FTM portfolio?  Not topical for n00bs, but an in-depth expert analysis as to the whys and the hows.

MG

[MachineGhost]

The last time 10-year T-Bonds traded at 2%, a 45-year bear market followed between 1941 and 1986, losing 56% in real terms.  Wow.

MG

[MediumTex]

The last time 10-year T-Bonds traded at 2%, a 45-year bear market followed between 1941 and 1986, losing 56% in real terms.   Wow.

MG

The U.S. economy saw staggering increases in output and productivity during that period, which one would expect to be accompanied by periods of inflation and/or strong stock market performance, either of which could provide upward pressure on bond yields.

Maybe it's not as shocking as it appears.