10 Year Treasury in ETFs

[ochotona]

IEF seems to outperform the 10 year Treasury benchmark at Portfoliovisualizer.com. Therefore, I would be inclined to just be happy with IEF and not over-science it.

[jalanlong]

I personally have a strong preference for the cash & long term treasury (barbell) combination, over the 10 year ladder or intermediate fund, because it's so much more flexible, and there are many good things about having a big chunk of cash. Assuming the barbell really is equivalent to the 10 year bond, which I guess might vary.

To do this you use a proportion of 40% (of the 60%) long treasuries (25 year duration) and 60% cash. That would put the total portfolio as 36% cash, 30% stocks, 24% long bonds, and 10% gold.

So according to my backtest the regular Desert Portfolio (using IEF in place of 10 year bonds) beats that portfolio and the PP both in pure terms and in volatility adjusted terms since 2007.

I know that backtesting is dependent on the timeframe you use. But aside from a spike in gold where i would want 25% instead of 10%, in what other scenarios would I want to have the PP instead of the Desert?

[sophie]

That's interesting, jalanlong. Can you post some numbers?

I'd assumed the barbell is equivalent to IEF, but it is probably not simply a linear combination. I would expect it to be heavily dependent on the shape of the yield curve, and that difference in performance may not hold up in other time periods (or going forward).

[jalanlong]

That's interesting, jalanlong. Can you post some numbers?

I'd assumed the barbell is equivalent to IEF, but it is probably not simply a linear combination. I would expect it to be heavily dependent on the shape of the yield curve, and that difference in performance may not hold up in other time periods (or going forward).

So using the Portfolio Visualizer from Jan 2008 until Sept 2019:

60% IEF Ishares 7-10 Year Treasury
30% VTI Total Stock Mkt
10% IAU Ishares Gold ETF

CAGR 6.58% Standard Dev 5.83% Max DD -10.40% Sharpe 1.03 Sortino 1.72


30% SHV Ishares 1 Year Treasury ETF
30% TLT Ishares 20+ Treasury
30% VTI Total Stock Mkt
10% IAU Ishares Gold ETF

CAGR 6.28% Standard Dev 5.92% Max DD -9.56% Sharpe 0.97 Sortino 1.57

Not a huge difference. Biggest difference was 2010 when the 10 Year made 13.80% and the Barbell made 10.92%

[sophie]

Ah, I see the problem. You split the barbell 50/50 between cash and long treasuries. To equate to a 10 year, you would use 40% long treasuries and 60% cash.

Also you used SHV which returned near zero during most of your tested time period. Keeping cash gives you a lot of options that can do much better, e.g. I Bonds, directly held T bills, online savings accounts, CDs etc.

[jalanlong]

Ah, I see the problem. You split the barbell 50/50 between cash and long treasuries. To equate to a 10 year, you would use 40% long treasuries and 60% cash.

Also you used SHV which returned near zero during most of your tested time period. Keeping cash gives you a lot of options that can do much better, e.g. I Bonds, directly held T bills, online savings accounts, CDs etc.

Changing it to 60/40 cash and TLT makes it worse actually. That makes a CAGR of 5.67% during the same period. But your point about SHV not always being the best option is very valid I am not sure how much that would juice up your returns to get an extra 1% out of 36% of your portfolio. But it would certainly help.

[jhogue]

I personally have a strong preference for the cash & long term treasury (barbell) combination, over the 10 year ladder or intermediate fund, because it's so much more flexible, and there are many good things about having a big chunk of cash. Assuming the barbell really is equivalent to the 10 year bond, which I guess might vary.

To do this you use a proportion of 40% (of the 60%) long treasuries (25 year duration) and 60% cash. That would put the total portfolio as 36% cash, 30% stocks, 24% long bonds, and 10% gold.

So according to my backtest the regular Desert Portfolio (using IEF in place of 10 year bonds) beats that portfolio and the PP both in pure terms and in volatility adjusted terms since 2007.

I know that backtesting is dependent on the timeframe you use. But aside from a spike in gold where i would want 25% instead of 10%, in what other scenarios would I want to have the PP instead of the Desert?

I think that the scenario your chosen backtesting timeframe is missing is something like the "Volker recession" of 1980-1981, in which the Fed chair drove up short term interest rates to kill inflation and returns from the other three assets all declined. If you had cash in a money market fund, the rates hit 20-21% and you felt like a genius. If you were holding intermediate or long term Treasurys, you got burned when bond prices collapsed. Maybe those conditions won't return any time soon, but I would not bet on it.

Like sophie, I am a fan of the traditional HBPP barbell because I think it gives investors much more flexibility to react to changing or unusual conditions in the money markets.

[jalanlong]

https://www.portfoliovisualizer.com/bac ... 0&total3=0

According to my backtest, in 1980 the Desert Portfolio lost 1.63% and the standard PP lost 6.13.

[jhogue]

Hmmm… I tried (unsuccessfully) to run a comparison between the Desert and HBPP portfolios for longer periods (5 or 10 years) ending in 1981.

Portfolio Visualizer shortened the period to Jan. 1978 – Dec 1981, with a note that: “The time period was constrained by the available data for Long Term Treasury (Jan 1978-Sep 2019).” According to TreasuryDirect.gov the 30 year T bond replaced the 25 year T bond in 1977 and Portfolio Visualizer is not equipped to compensate for the multiple changes that have occurred in the term (and sometimes the offering) of the bellwether long bond.

From Jan 1978 to Dec 1981, the Desert Portfolio had a CAGR of 7.96%, and the HBPP had a CAGR of 12.96%. It seems likely to me that you were correct and that the latter outperformed the former due to the gold spike of 1980. But this exercise is also perhaps another cautionary tale that the utility of backtesting is limited when the definitions of apples and oranges have changed over time.

[jalanlong]

The same things I like about the Desert Portfolio are the same things that worry me about it.

One the one hand, it is so simple to have my entire portfolio in 3 funds (IEF, VTI, IAU) that seem to cover most economic scenarios. But on the other hand I look at my brokerage statement and think oh my gosh all of my money is in 3 funds. I have none of the various assets I am told I need to diversify. No corp bonds. No intl. No small cap or factor funds. You gotta have strength in your convictions I guess.

[sophie]

You need at least 20% gold to do well in a 1970s-like era - check the year by year real returns on Portfoliocharts.com and you'll see what I mean.

I thought the whole point of the Desert Portfolio was to appeal to conservative investors who are skittish about gold. It fills that niche very well. It certainly would have done better in the 1970s than any purely stock/bond portfolio.

[jalanlong]

You need at least 20% gold to do well in a 1970s-like era - check the year by year real returns on Portfoliocharts.com and you'll see what I mean.

I thought the whole point of the Desert Portfolio was to appeal to conservative investors who are skittish about gold. It fills that niche very well. It certainly would have done better in the 1970s than any purely stock/bond portfolio.

You are correct that if I start a backtest at 1980 then 60/30/10 (10Y Treasury, Stocks, Gold) beats 50/30/20 but if I change the starting date to 1972 then that 8 years makes all the difference and the 50/30/20 wins easily. How much of that came from the decoupling from the gold standard which will never be repeated again?