Question about bonds AA

Have a question about your personal investments? No matter how simple or complex, you can ask it here.
Post Reply
Topic Author
Almond
Posts: 87
Joined: Wed May 01, 2019 3:11 pm

Question about bonds AA

Post by Almond »

I know past doesn’t guarantee future. Backtested a AA — 100% stock, 50/50, and 60/40 over 40 years. Seems 60/40 performed best and 50/50 was not too bad as even the worse year would not have had you freaking out. Am I missing something?

https://www.portfoliovisualizer.com/bac ... teTreasury
User avatar
Third Son
Posts: 388
Joined: Tue Dec 03, 2019 6:05 pm
Location: North of Nowhere

Re: Question about bonds AA

Post by Third Son »

This is a good article to read... Ala Rick Ferri

https://www.etf.com/sections/index-inve ... nopaging=1
"A part of all you earn is yours to keep" | | -The Richest Man in Babylon
User avatar
Third Son
Posts: 388
Joined: Tue Dec 03, 2019 6:05 pm
Location: North of Nowhere

Re: Question about bonds AA

Post by Third Son »

This is a good article to read... ala Rick Ferri. Note chart 2

https://www.etf.com/sections/index-inve ... nopaging=1
Last edited by Third Son on Mon Mar 02, 2020 5:14 pm, edited 1 time in total.
"A part of all you earn is yours to keep" | | -The Richest Man in Babylon
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Question about bonds AA

Post by Thesaints »

In general, 100% stocks performed best in backtests (if "best" means "higher return"), unless your backtest is on some special timeframe..
Topic Author
Almond
Posts: 87
Joined: Wed May 01, 2019 3:11 pm

Re: Question about bonds AA

Post by Almond »

Thesaints wrote: Mon Mar 02, 2020 5:04 pm In general, 100% stocks performed best in backtests (if "best" means "higher return"), unless your backtest is on some special timeframe..
Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.

I thought 100% stocks over time performed significantly better, so now I am confused and maybe I am not using the right data.
Topic Author
Almond
Posts: 87
Joined: Wed May 01, 2019 3:11 pm

Re: Question about bonds AA

Post by Almond »

Third Son wrote: Mon Mar 02, 2020 5:01 pm This is a good article to read... Ala Rick Ferri

https://www.etf.com/sections/index-inve ... nopaging=1

Thank you. Will check it out
User avatar
Brianmcg321
Posts: 1875
Joined: Mon Jul 15, 2019 8:23 am

Re: Question about bonds AA

Post by Brianmcg321 »

Almond wrote: Mon Mar 02, 2020 5:18 pm
Thesaints wrote: Mon Mar 02, 2020 5:04 pm In general, 100% stocks performed best in backtests (if "best" means "higher return"), unless your backtest is on some special timeframe..
Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.

I thought 100% stocks over time performed significantly better, so now I am confused and maybe I am not using the right data.

$500,000 invested
1980 -2020 :
60/40 portfolio = $8,068,271
50/50 portfolio = $7,153,100
100% Total US Market = $11,805,841

https://www.portfoliovisualizer.com/bac ... allocation
Last edited by Brianmcg321 on Mon Mar 02, 2020 5:32 pm, edited 1 time in total.
Rules to investing: | 1. Don't lose money. | 2. Don't forget rule number 1.
dbr
Posts: 46181
Joined: Sun Mar 04, 2007 8:50 am

Re: Question about bonds AA

Post by dbr »

Almond wrote: Mon Mar 02, 2020 5:18 pm
Thesaints wrote: Mon Mar 02, 2020 5:04 pm In general, 100% stocks performed best in backtests (if "best" means "higher return"), unless your backtest is on some special timeframe..
Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.

I thought 100% stocks over time performed significantly better, so now I am confused and maybe I am not using the right data.
PV results are highly sensitive to timeframe. Buying long bonds at the unprecedented and probably never to be repeated peak of high interest rates in 1980 is going to produce returns similar to stocks under those very unique conditions, thus the results you get. There are certainly intervals of significant length when bonds outperform stocks. Backtesting is not a very good tool for planning asset allocation though it is a good tool for illustrating all kinds of things that have happened.

Note PV has a Monte-Carlo simulation tool that is probably a better approach to estimating what might happen in the future, though MC has its own set of issues and cautions. Another tool if you want a report characterized by past history is to examine the output from FireCalc, which can show you a complete set of outcomes for investing started in any year from 1872 or so onward up to the latest year that allows you to complete a history of the length you want.
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Question about bonds AA

Post by Thesaints »

Almond wrote: Mon Mar 02, 2020 5:18 pm Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.
It might be true over the particular 40 years you picked.
As you know, average return of stock market has been over 10%. Average return of bond market has been a little more than 5%
https://personal.vanguard.com/us/insigh ... llocations

It means that, on average, all stocks performs better than 60/40 on any timeframe.
Topic Author
Almond
Posts: 87
Joined: Wed May 01, 2019 3:11 pm

Re: Question about bonds AA

Post by Almond »

dbr wrote: Mon Mar 02, 2020 5:32 pm
Almond wrote: Mon Mar 02, 2020 5:18 pm
Thesaints wrote: Mon Mar 02, 2020 5:04 pm In general, 100% stocks performed best in backtests (if "best" means "higher return"), unless your backtest is on some special timeframe..
Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.

I thought 100% stocks over time performed significantly better, so now I am confused and maybe I am not using the right data.
PV results are highly sensitive to timeframe. Buying long bonds at the unprecedented and probably never to be repeated peak of high interest rates in 1980 is going to produce returns similar to stocks under those very unique conditions, thus the results you get. There are certainly intervals of significant length when bonds outperform stocks. Backtesting is not a very good tool for planning asset allocation though it is a good tool for illustrating all kinds of things that have happened.

Note PV has a Monte-Carlo simulation tool that is probably a better approach to estimating what might happen in the future, though MC has its own set of issues and cautions. Another tool if you want a report characterized by past history is to examine the output from FireCalc, which can show you a complete set of outcomes for investing started in any year from 1872 or so onward up to the latest year that allows you to complete a history of the length you want.
I changed it to 1990, so 30 years and 60/40 still performed better with less stress

https://www.portfoliovisualizer.com/bac ... rtTreasury
Topic Author
Almond
Posts: 87
Joined: Wed May 01, 2019 3:11 pm

Re: Question about bonds AA

Post by Almond »

Almond wrote: Mon Mar 02, 2020 5:43 pm
dbr wrote: Mon Mar 02, 2020 5:32 pm
Almond wrote: Mon Mar 02, 2020 5:18 pm
Thesaints wrote: Mon Mar 02, 2020 5:04 pm In general, 100% stocks performed best in backtests (if "best" means "higher return"), unless your backtest is on some special timeframe..
Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.

I thought 100% stocks over time performed significantly better, so now I am confused and maybe I am not using the right data.
PV results are highly sensitive to timeframe. Buying long bonds at the unprecedented and probably never to be repeated peak of high interest rates in 1980 is going to produce returns similar to stocks under those very unique conditions, thus the results you get. There are certainly intervals of significant length when bonds outperform stocks. Backtesting is not a very good tool for planning asset allocation though it is a good tool for illustrating all kinds of things that have happened.

Note PV has a Monte-Carlo simulation tool that is probably a better approach to estimating what might happen in the future, though MC has its own set of issues and cautions. Another tool if you want a report characterized by past history is to examine the output from FireCalc, which can show you a complete set of outcomes for investing started in any year from 1872 or so onward up to the latest year that allows you to complete a history of the length you want.
I changed it to 1990, so 30 years and 60/40 still performed better with less stress

https://www.portfoliovisualizer.com/bac ... rtTreasury
And over 20 years from 2000 both 50/50 and 60/40 do much better. I just always thought 100% stocks over time performed better so I am a bit confused

https://www.portfoliovisualizer.com/bac ... rtTreasury
dbr
Posts: 46181
Joined: Sun Mar 04, 2007 8:50 am

Re: Question about bonds AA

Post by dbr »

Almond wrote: Mon Mar 02, 2020 5:43 pm
dbr wrote: Mon Mar 02, 2020 5:32 pm
Almond wrote: Mon Mar 02, 2020 5:18 pm
Thesaints wrote: Mon Mar 02, 2020 5:04 pm In general, 100% stocks performed best in backtests (if "best" means "higher return"), unless your backtest is on some special timeframe..
Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.

I thought 100% stocks over time performed significantly better, so now I am confused and maybe I am not using the right data.
PV results are highly sensitive to timeframe. Buying long bonds at the unprecedented and probably never to be repeated peak of high interest rates in 1980 is going to produce returns similar to stocks under those very unique conditions, thus the results you get. There are certainly intervals of significant length when bonds outperform stocks. Backtesting is not a very good tool for planning asset allocation though it is a good tool for illustrating all kinds of things that have happened.

Note PV has a Monte-Carlo simulation tool that is probably a better approach to estimating what might happen in the future, though MC has its own set of issues and cautions. Another tool if you want a report characterized by past history is to examine the output from FireCalc, which can show you a complete set of outcomes for investing started in any year from 1872 or so onward up to the latest year that allows you to complete a history of the length you want.
I changed it to 1990, so 30 years and 60/40 still performed better with less stress

https://www.portfoliovisualizer.com/bac ... rtTreasury
That could happen. People shouldn't be so surprised. Did you notice that interest rates are still very high in 1990 and that the period is characterized by a 40 year continuous decline from 1980 to 2020. The duration of Vanguard long term bond funds is on the order of 15-20 years

What would you like the message to be?
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Question about bonds AA

Post by Thesaints »

Almond wrote: Mon Mar 02, 2020 5:45 pm And over 20 years from 2000 both 50/50 and 60/40 do much better. I just always thought 100% stocks over time performed better so I am a bit confused
They do, on average.
dbr
Posts: 46181
Joined: Sun Mar 04, 2007 8:50 am

Re: Question about bonds AA

Post by dbr »

Thesaints wrote: Mon Mar 02, 2020 5:51 pm
Almond wrote: Mon Mar 02, 2020 5:45 pm And over 20 years from 2000 both 50/50 and 60/40 do much better. I just always thought 100% stocks over time performed better so I am a bit confused
They do, on average.
Yep, and the message is that you can't expect the expected return. There is still a big caveat at looking at long bonds during any part of the last forty years. Heck, the trailing one year return of VUSTX is 31.58% compared to 6.78% for VTSAX. You had to be an idiot to be in stocks rather than long bonds the past year. :shock: My VIPSX even returned 10.01% trailing year to now.
User avatar
vineviz
Posts: 14921
Joined: Tue May 15, 2018 1:55 pm
Location: Baltimore, MD

Re: Question about bonds AA

Post by vineviz »

Almond wrote: Mon Mar 02, 2020 5:45 pm I changed it to 1990, so 30 years and 60/40 still performed better with less stress

And over 20 years from 2000 both 50/50 and 60/40 do much better. I just always thought 100% stocks over time performed better so I am a bit confused
On average, 100% stocks DO have higher expected returns than a portfolio that is partly in bonds. The backtests you're looking at are unusual and/or not comparable to today in two important related) ways: they include a period of time with steadily decreasing interest rates; and they include a period of time when the starting yields were significantly higher than now.

Predicting the future path of interest rate changes is a fool's errand, so unless the investor has a functioning crystal ball the default assumption should be that future interest rates will not change from today's levels.

And since the expected return on bonds is more or less the starting yield to maturity, the investor should expect the total return for intermediate term Treasury bonds to be equivalent to their current nominal yield or about 1.0 to 1.5%.

In 1990 similar bonds were yielding over 8% and in 2000 they were yielding around 6%, so obviously past returns from bonds were higher than current expectations.

Making a case that FUTURE bond returns will mimic past bond returns would be extremely difficult.


Expected returns for global equities are a whole other can of worms, but I personally wouldn't count one the future returns of a 50/50 portfolio to match or exceed that of an equity-only portfolio.
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch
User avatar
puc_ytpme
Posts: 52
Joined: Fri Aug 02, 2019 2:04 pm
Location: USA

Re: Question about bonds AA

Post by puc_ytpme »

vineviz wrote: Tue Mar 03, 2020 3:54 pm
Almond wrote: Mon Mar 02, 2020 5:45 pm I changed it to 1990, so 30 years and 60/40 still performed better with less stress

And over 20 years from 2000 both 50/50 and 60/40 do much better. I just always thought 100% stocks over time performed better so I am a bit confused
On average, 100% stocks DO have higher expected returns than a portfolio that is partly in bonds. The backtests you're looking at are unusual and/or not comparable to today in two important related) ways: they include a period of time with steadily decreasing interest rates; and they include a period of time when the starting yields were significantly higher than now.

Predicting the future path of interest rate changes is a fool's errand, so unless the investor has a functioning crystal ball the default assumption should be that future interest rates will not change from today's levels.

And since the expected return on bonds is more or less the starting yield to maturity, the investor should expect the total return for intermediate term Treasury bonds to be equivalent to their current nominal yield or about 1.0 to 1.5%.

In 1990 similar bonds were yielding over 8% and in 2000 they were yielding around 6%, so obviously past returns from bonds were higher than current expectations.

Making a case that FUTURE bond returns will mimic past bond returns would be extremely difficult.


Expected returns for global equities are a whole other can of worms, but I personally wouldn't count one the future returns of a 50/50 portfolio to match or exceed that of an equity-only portfolio.
Vineviz,

Welcome back :D
No person ever steps in the same river twice, for it’s not the same river & they’re not the same person
Thesaints
Posts: 5108
Joined: Tue Jun 20, 2017 12:25 am

Re: Question about bonds AA

Post by Thesaints »

Let's say that asset A outperforms B yearly with a 90% probability by 5% and there is a 10% probability B outperforms A by 200%.
Under the above hypothesis, turns out that B has an expected outperformance over A of about 2.3%/yr.
You don't really invest in B expecting to beat A yearly, though...
dcabler
Posts: 4543
Joined: Wed Feb 19, 2014 10:30 am
Location: TX

Re: Question about bonds AA

Post by dcabler »

vineviz wrote: Tue Mar 03, 2020 3:54 pm
Almond wrote: Mon Mar 02, 2020 5:45 pm I changed it to 1990, so 30 years and 60/40 still performed better with less stress

And over 20 years from 2000 both 50/50 and 60/40 do much better. I just always thought 100% stocks over time performed better so I am a bit confused
On average, 100% stocks DO have higher expected returns than a portfolio that is partly in bonds. The backtests you're looking at are unusual and/or not comparable to today in two important related) ways: they include a period of time with steadily decreasing interest rates; and they include a period of time when the starting yields were significantly higher than now.

Predicting the future path of interest rate changes is a fool's errand, so unless the investor has a functioning crystal ball the default assumption should be that future interest rates will not change from today's levels.

And since the expected return on bonds is more or less the starting yield to maturity, the investor should expect the total return for intermediate term Treasury bonds to be equivalent to their current nominal yield or about 1.0 to 1.5%.

In 1990 similar bonds were yielding over 8% and in 2000 they were yielding around 6%, so obviously past returns from bonds were higher than current expectations.

Making a case that FUTURE bond returns will mimic past bond returns would be extremely difficult.


Expected returns for global equities are a whole other can of worms, but I personally wouldn't count one the future returns of a 50/50 portfolio to match or exceed that of an equity-only portfolio.
Welcome back!
User avatar
ruralavalon
Posts: 26351
Joined: Sat Feb 02, 2008 9:29 am
Location: Illinois

Re: Question about bonds AA

Post by ruralavalon »

Almond wrote: Mon Mar 02, 2020 5:45 pm
Almond wrote: Mon Mar 02, 2020 5:43 pm
dbr wrote: Mon Mar 02, 2020 5:32 pm
Almond wrote: Mon Mar 02, 2020 5:18 pm
Thesaints wrote: Mon Mar 02, 2020 5:04 pm In general, 100% stocks performed best in backtests (if "best" means "higher return"), unless your backtest is on some special timeframe..
Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.

I thought 100% stocks over time performed significantly better, so now I am confused and maybe I am not using the right data.
PV results are highly sensitive to timeframe. Buying long bonds at the unprecedented and probably never to be repeated peak of high interest rates in 1980 is going to produce returns similar to stocks under those very unique conditions, thus the results you get. There are certainly intervals of significant length when bonds outperform stocks. Backtesting is not a very good tool for planning asset allocation though it is a good tool for illustrating all kinds of things that have happened.

Note PV has a Monte-Carlo simulation tool that is probably a better approach to estimating what might happen in the future, though MC has its own set of issues and cautions. Another tool if you want a report characterized by past history is to examine the output from FireCalc, which can show you a complete set of outcomes for investing started in any year from 1872 or so onward up to the latest year that allows you to complete a history of the length you want.
I changed it to 1990, so 30 years and 60/40 still performed better with less stress

https://www.portfoliovisualizer.com/bac ... rtTreasury
And over 20 years from 2000 both 50/50 and 60/40 do much better. I just always thought 100% stocks over time performed better so I am a bit confused

https://www.portfoliovisualizer.com/bac ... rtTreasury
Using actual investable funds rather than indexes, and the more common total bond fund rather than long treasury fund, gives different results. Portfolio Visualizer, 1997-2020.
100% = $113k
60/40 = $84k
50/50 = $76k
With the expected higher volatility for higher stock allocations.

Using actual investable funds with intermediate-term treasury bond fund. Portfolio Visualizer, 1993-2020.
100% = $113k
60/40 = $87k
50/50 = $79k
With the expected higher volatility for the higher stock allocations.

Using actual investable funds with long-term treasury fund. Portfolio Visualizer, 1993-2020.
100% = $113k
60/40 = $114k
50/50 = $109k
With the expected higher volatility for the higher stock allocations.
"Everything should be as simple as it is, but not simpler." - Albert Einstein | Wiki article link: Bogleheads® investment philosophy
Topic Author
Almond
Posts: 87
Joined: Wed May 01, 2019 3:11 pm

Re: Question about bonds AA

Post by Almond »

ruralavalon wrote: Thu Mar 05, 2020 11:17 am
Almond wrote: Mon Mar 02, 2020 5:45 pm
Almond wrote: Mon Mar 02, 2020 5:43 pm
dbr wrote: Mon Mar 02, 2020 5:32 pm
Almond wrote: Mon Mar 02, 2020 5:18 pm

Did it over 40 years and 60/40 did best, and while 50/50 came third the difference was small and max drawdowns were sig smaller. Making it easy to sleep at night.

I thought 100% stocks over time performed significantly better, so now I am confused and maybe I am not using the right data.
PV results are highly sensitive to timeframe. Buying long bonds at the unprecedented and probably never to be repeated peak of high interest rates in 1980 is going to produce returns similar to stocks under those very unique conditions, thus the results you get. There are certainly intervals of significant length when bonds outperform stocks. Backtesting is not a very good tool for planning asset allocation though it is a good tool for illustrating all kinds of things that have happened.

Note PV has a Monte-Carlo simulation tool that is probably a better approach to estimating what might happen in the future, though MC has its own set of issues and cautions. Another tool if you want a report characterized by past history is to examine the output from FireCalc, which can show you a complete set of outcomes for investing started in any year from 1872 or so onward up to the latest year that allows you to complete a history of the length you want.
I changed it to 1990, so 30 years and 60/40 still performed better with less stress

https://www.portfoliovisualizer.com/bac ... rtTreasury
And over 20 years from 2000 both 50/50 and 60/40 do much better. I just always thought 100% stocks over time performed better so I am a bit confused

https://www.portfoliovisualizer.com/bac ... rtTreasury
Using actual investable funds rather than indexes, and the more common total bond fund rather than long treasury fund, gives different results. Portfolio Visualizer, 1997-2020.
100% = $113k
60/40 = $84k
50/50 = $76k
With the expected higher volatility for higher stock allocations.

Using actual investable funds with intermediate-term treasury bond fund. Portfolio Visualizer, 1993-2020.
100% = $113k
60/40 = $87k
50/50 = $79k
With the expected higher volatility for the higher stock allocations.

Using actual investable funds with long-term treasury fund. Portfolio Visualizer, 1993-2020.
100% = $113k
60/40 = $114k
50/50 = $109k
With the expected higher volatility for the higher stock allocations.
Thank you
Post Reply