Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

Lifecycle Investing refers to the strategy popularized by Ayers and Nalebuff. The basic concept is to use leverage while early in the accumulation phase to maintain a relatively constant equity exposure, rather than a standard glidepath or constant allocation, which has much more equity exposure later in life. This is called "temporal diversification," and reduces risk when applied to a full portfolio over time. For further details, see: Hedgefundie's Excellent Adventure (HFEA) started as a risk parity strategy and evolved into a leveraged version of a balanced portfolio that backtests extremely well. Per Hedgefundie, it is intended to be a "lottery ticket", not meant to be a full portfolio. The allocation is 55% UPRO (3x S&P 500) and 45% TMF (3x Long-Term Treasuries). Long-Term Treasuries (LTT) were selected due to their low correlation with the S&P 500. For further details, see:
Lifecycle Investing is meant to be a long-term strategy, and the book includes both historical and bootstrap-simulated results for an entire accumulation phase, including additional contributions over time. The theory is sound and the results are impressive. However, after reading about HFEA, I wondered how the results would compare and if the Lifecycle Investing strategy could be improved by levering a balanced portfolio. Since Hedgefundie intended his strategy to be a one-time purchase, the backtesting did not include additional contributions over time. So I built a backtesting spreadsheet to do the apples-to-apples comparison.

I considered 6 strategies for the backtest:
  • A constant allocation of 100% VTSAX
  • The HFEA strategy: a constant allocation of 55% UPRO and 45% TMF
  • The Lifecycle Investing strategy: gliding from 200% equity to 70%
  • An unlevered glidepath: gliding from 100% equity to 70%
  • A constant allocation of 2x leverage 70% S&P 500 30% LTT
  • A modified version of the Lifecycle Investing strategy where bonds are also levered to maintain a constant stock:bond ratio
I did the simulations based roughly on my personal situation: $1M in an investment portfolio and $100k annual contributions to the portfolio, with a 35-year investment horizon. I used data from 1955 to 2020. I started at 1955 because that's when simulated data from leveraged ETFs is available (see siamond's addition of leveraged ETFs to the Simba backtesting sheet). This gives me a sample of 32 periods of 35 years, and I used those 32 periods to determine various percentiles of ending results for each strategy.

Image

From these results, it appears HFEA is just as described: a lottery ticket. It has by far the lowest minimum and the highest maximum outcome. The Lifecycle Investing strategy is a more stable approach but offered little benefit over an unleveraged version. That makes sense for my situation - since I already have a sizable portfolio, significant leverage is only used for the next ~5 years. It seems adding leveraged bonds to this strategy does more harm than good, especially in the lower percentiles.

I think the most attractive option to me is the constant 2x 70-30 portfolio. Its median result is close to HFEA, while its minimum and 10th percentile result are vastly superior.

My current allocation is 100% equities (70% domestic, 30% international). Before I make any changes to my portfolio, I wanted to seek the wisdom of the Bogleheads to correct any mistakes and point out aspects I am not considering. I also have a few more research items I want to complete, though I am hesitant to do much more backtesting for risk of overfitting.

Needs more research:
  • Repeat analysis with bootstrap simulation
  • Consider using a total bond fund rather than LTT, which was used as the bond fund solely because of its inclusion in HFEA
  • Determine optimal way to hold funds across taxable and tax-advantaged accounts. E.g. if I want to use a constant 2x levered 70-30 portfolio, I would have 4 funds: leveraged stocks, stocks, leveraged bonds, bonds. Should I lever proportionally (35/35/15/15) or contain more leverage in one asset class (30/50/20/0)? For the simulation, I assumed proportional leverage and ignored taxes.
  • Determine what to do with international equities. I did not include in the backtesting for two reasons. The first is that VTIAX data only goes back to 1970. The second is that I recognize that international has underperformed U.S. in the past, but I am not confident it will continue to do so in the future. So I'd rather use this analysis to determine a general strategy for my equity allocation and then work in international.
If you'd like to get into detail on the math and help check my work, please see my backtesting Excel workbook (with a macro to run the simulation across periods) or the workbook without macros.
redbarn
Posts: 65
Joined: Mon Feb 10, 2020 5:07 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by redbarn »

My main concern would be that strategies become especially prominent when they would have done well in the past. A backtest of a strategy (or a version of a strategy) that gained prominence through past performance is going to be by construction biased in favor of the strategy. The results about the worst case scenario seem particularly illusory -- a strategy that was identified because it did well wouldn't have done badly during the period it did well in. A Monte Carlo simulation should be a bare minimum requirement for this type of evaluation, that would at least test the strategy against a greater range of potential shocks. A simpler exercise where you parametrically feed in estimates of expected returns for different asset classes (and inflation) based on today's circumstances would also be helpful in testing the strategy against circumstances more different from the ones on the basis of which it gained prominence in the first place.
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

The primary purpose of the lifecycle investing strategy is to solve the lack of time diversification caused by continuing contributions and accumulation. You mention that you are starting with $1M and you applied leverage for only the first 5 years due to the shortened 35-year time horizon. For these two reasons it doesn't surprise me that the lifecycle strategy compares poorly. There is no doubt that the lifecycle strategy would perform best when used as intended: time diversification for younger individuals (20-45) in the accumulation phase.

For your personal situation, I have no problems with a highly leveraged stock and bond position if the risk level is tolerable for you. Backtesting is likely to underestimate the true risk because the future may not be similar to the past. A 2x 70-30 portfolio is similar to my own portfolio (2.5x 50-50), although I intend to de-leverage as I accumulate and age, more similar to a lifecycle strategy.

No matter what your goals are, it's likely some version of lifecycle / glide-path de-leveraging would be appropriate.

Also, while the use of LETF is probably acceptable for an early-mid stage accumulator, they're probably not appropriate for a large portfolio. The fees are significant. Significant improvements could be achieved with the use of margin + box spreads and/or futures contracts.
Last edited by skierincolorado on Mon Aug 09, 2021 12:00 am, edited 5 times in total.
Marseille07
Posts: 5392
Joined: Fri Nov 06, 2020 1:41 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Marseille07 »

I think the big question mark is how the bonds would perform going forward.

Of course, we don't know if we'll ever see a rising rates environment, but it was painful when 10Y jumped from 0.5% to 1.75% not too long ago. And no one knows why 10Y crashed from 1.75% to 1.2% either.
TheCleverest
Posts: 39
Joined: Sat Dec 26, 2015 11:14 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by TheCleverest »

skierincolorado wrote: Sun Aug 08, 2021 11:47 pm The primary purpose of the lifecycle investing strategy is to solve the lack of time diversification caused by continuing contributions and accumulation. You mention that your simulation did not include additional contributions and that the version of the lifecycle strategy you applied involved leverage for only the first 5 years due to the shortened 35-year time horizon. For these two reasons it doesn't surprise me that the lifecycle strategy compares poorly. There is no doubt that the lifecycle strategy would perform best when used as intended: time diversification for individuals in the accumulation phase.
Well said.
No use in being clever - have to be the cleverest
jarjarM
Posts: 1457
Joined: Mon Jul 16, 2018 1:21 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by jarjarM »

Making significant course correction or leverage decision based on backtesting likely lead to making decision by random chances. With leverage, there's significant start time sensitivity (ie return with simulated TQQQ will differ quite a bit between starting at 2000 or 2001). Hence I think you have to ask if the current forward looking investment environment is similar to the past 60 years or will it be different. Will bond serve as a tailwind for majority of the test period? Will stock growth significantly outpace inflation?

Also keep in mind that any leverage will amplify the volatilities and potential drawdown so be prepare for the dark days. Personally, I think a 2x 70-30 maybe a bit aggressive to start off with, may want to slowly ramp up the leverage as you get used to the daily ups/downs. Maybe look into more diversification with small cap? Good luck.
User avatar
Svensk Anga
Posts: 907
Joined: Sun Dec 23, 2012 5:16 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Svensk Anga »

It looks like your back test has no periods ending in the awful 1974 to 1982 period. The price return of the Dow was zero here and rising rates and high inflation destroyed bond real returns. It has been mostly favorable since with buoyant stocks, mild inflation, and a 40 year bond bull market. If you were withdrawing in this period rather than contributing, it was especially bad.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

redbarn wrote: Sun Aug 08, 2021 11:32 pm A backtest of a strategy (or a version of a strategy) that gained prominence through past performance is going to be by construction biased in favor of the strategy...A Monte Carlo simulation should be a bare minimum requirement for this type of evaluation, that would at least test the strategy against a greater range of potential shocks. A simpler exercise where you parametrically feed in estimates of expected returns for different asset classes (and inflation) based on today's circumstances would also be helpful in testing the strategy against circumstances more different from the ones on the basis of which it gained prominence in the first place.
I agree with you and others on the limitations of backtesting. I plan to repeat the analysis using bootstrap-sampled Monte Carlo simulations. That will still assume future possibilities are constrained to historical results, but it will at least be on an annual basis rather than a 35-year period. I suspect this will have more realistic (lower) minimum results. I also am intentionally only testing broad strategies, not specific parameters like the maximum leverage or ending equity percentage, in an attempt to mitigate overfitting. But I acknowledge there is still significant risk of doing so.

I'm personally not very interested in selecting parameters to determine a distribution of returns, as the results would be very sensitive to my selections, and I do not believe I can predict future results.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

skierincolorado wrote: Sun Aug 08, 2021 11:47 pm The primary purpose of the lifecycle investing strategy is to solve the lack of time diversification caused by continuing contributions and accumulation. You mention that you are starting with $1M and you applied leverage for only the first 5 years due to the shortened 35-year time horizon. For these two reasons it doesn't surprise me that the lifecycle strategy compares poorly. There is no doubt that the lifecycle strategy would perform best when used as intended: time diversification for younger individuals (20-45) in the accumulation phase.

For your personal situation, I have no problems with a highly leveraged stock and bond position if the risk level is tolerable for you. Backtesting is likely to underestimate the true risk because the future may not be similar to the past. A 2x 70-30 portfolio is similar to my own portfolio (2.5x 50-50), although I intend to de-leverage as I accumulate and age, more similar to a lifecycle strategy.

No matter what your goals are, it's likely some version of lifecycle / glide-path de-leveraging would be appropriate.

Also, while the use of LETF is probably acceptable for an early-mid stage accumulator, they're probably not appropriate for a large portfolio. The fees are significant. Significant improvements could be achieved with the use of margin + box spreads and/or futures contracts.
This is all very helpful, thank you! Looking at the results, I wouldn't say Lifecycle Investing performs poorly, just that it doesn't knock it out of the park like in the book examples. It's also worth noting that my "Unlevered Glidepath" is still following the basic methodology of Lifecycle Investing (considering the PV of future contributions as part of the portfolio) but without leverage, resulting in a higher equity allocation than you would have with a more traditional target retirement fund. So it makes sense the results for those two end up very close.

It sounds like you are currently holding a leveraged balanced portfolio, with the plan to delever over time. This is what I was testing with the "Lifecycle w Levered Bonds" approach - essentially a constant 70/30 equity/bond split that starts with leverage and gradually reduces to an end point of a standard 70/30 portfolio without leverage. I thought this may get the best of both worlds from HFEA and Lifecycle Investing, but I was surprised to see it tested poorly, consistently beat by the Lifecycle strategy holding only equities while leveraged. Could you share any info on how you landed on this strategy / additional reading I can do to see if I might have missed something?
User avatar
OohLaLa
Posts: 364
Joined: Tue Mar 09, 2021 8:26 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by OohLaLa »

millennialmillions wrote: Mon Aug 09, 2021 9:01 am
skierincolorado wrote: Sun Aug 08, 2021 11:47 pm [...]
For your personal situation, I have no problems with a highly leveraged stock and bond position if the risk level is tolerable for you. Backtesting is likely to underestimate the true risk because the future may not be similar to the past. A 2x 70-30 portfolio is similar to my own portfolio (2.5x 50-50), although I intend to de-leverage as I accumulate and age, more similar to a lifecycle strategy.

No matter what your goals are, it's likely some version of lifecycle / glide-path de-leveraging would be appropriate.
[...]
[...]
It sounds like you are currently holding a leveraged balanced portfolio, with the plan to delever over time. This is what I was testing with the "Lifecycle w Levered Bonds" approach - essentially a constant 70/30 equity/bond split that starts with leverage and gradually reduces to an end point of a standard 70/30 portfolio without leverage. I thought this may get the best of both worlds from HFEA and Lifecycle Investing, but I was surprised to see it tested poorly, consistently beat by the Lifecycle strategy holding only equities while leveraged. Could you share any info on how you landed on this strategy / additional reading I can do to see if I might have missed something?
I have a similar mindset as skier.

I find it very hard to believe that progressive de-leveraging provides the worst outcome out of all the approaches. Barring a horrible sequence of events during 3x, just before moving into lower leverage and making 2x recovery more difficult, I would believe it to remain a solid choice.

One thing that can be extremely misleading in the selected AA is that 55/45 does not give anywhere close to the same ratio of equities to bonds as 70/30. The latter, when leveraged 2-3x, is really not something I envision holding for the long-term.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

OohLaLa wrote: Mon Aug 09, 2021 6:28 pm I find it very hard to believe that progressive de-leveraging provides the worst outcome out of all the approaches. Barring a horrible sequence of events during 3x, just before moving into lower leverage and making 2x recovery more difficult, I would believe it to remain a solid choice.
I thought the same way before doing the analysis, but the results speak for themselves. The Lifecycle w Levered Bonds approach loses to the standard Lifecycle approach (no bonds when leveraged) in 25/32 periods. The problem seems to be that including leveraged bonds adds volatility that results in some serious drawdowns that are tough to recover from. Also, even when LTTs return around 0% for the year, the 3x leveraged fund can have a very negative return due to volatility decay. Here is a good example - this is the worst 35-year period for both the Lifecycle w Levered Bonds approach and the standard Lifecycle approach. The starting year is 1968.

Lifecycle w Levered Bonds
Image

Standard Lifecyle (no bonds when leveraged)
Image

As you said, the ending years are near identical, so the difference occurs when leverage is significant. Holding bonds requires a greater portion of your portfolio to be in leveraged funds, meaning your results are driven by the outcomes of the 3x funds. Over a year, these can significantly underperform the unleveraged asset when volatility is high. For example, see year 2 (1969). While U.S. stocks and LTTs were down 10.3% and 5.8% respectively, the 3x counterparts were down 38.0% and 31.7%. This results in the Lifecycle w Levered Bonds strategy having a much worse return. The same thing happens in year 11 (1978).

Of course, there are multiple ways to build a glidepath for a leveraged fund with stocks and bonds. I chose a simple method of maintaining a constant stock:bond ratio using a target equity allocation consistent with the standard lifecycle strategy. I'm sure there are ways to beat this, but I did not want to backtest to tune parameters, which would certainly result in overfitting.
TheCleverest
Posts: 39
Joined: Sat Dec 26, 2015 11:14 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by TheCleverest »

I'm curious. If a lottery ticket and only 5-15% of your portfolio is what you're willing to use, then have you considered HFEA vs...
a) SCV equities
b) VC funds
c) real estate deals
d) speculative digital assets
e) Frontier and Emerging Markets

If so, what were your thoughts?

I'm not trying to agitate you. Rather, I'd like to see what purpose you're attempting to achieve (upside, volatility, risk parity, etc).
No use in being clever - have to be the cleverest
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

millennialmillions wrote: Mon Aug 09, 2021 10:50 pm
OohLaLa wrote: Mon Aug 09, 2021 6:28 pm I find it very hard to believe that progressive de-leveraging provides the worst outcome out of all the approaches. Barring a horrible sequence of events during 3x, just before moving into lower leverage and making 2x recovery more difficult, I would believe it to remain a solid choice.
I thought the same way before doing the analysis, but the results speak for themselves. The Lifecycle w Levered Bonds approach loses to the standard Lifecycle approach (no bonds when leveraged) in 25/32 periods. The problem seems to be that including leveraged bonds adds volatility that results in some serious drawdowns that are tough to recover from. Also, even when LTTs return around 0% for the year, the 3x leveraged fund can have a very negative return due to volatility decay. Here is a good example - this is the worst 35-year period for both the Lifecycle w Levered Bonds approach and the standard Lifecycle approach. The starting year is 1968.

Lifecycle w Levered Bonds
Image

Standard Lifecyle (no bonds when leveraged)
Image

As you said, the ending years are near identical, so the difference occurs when leverage is significant. Holding bonds requires a greater portion of your portfolio to be in leveraged funds, meaning your results are driven by the outcomes of the 3x funds. Over a year, these can significantly underperform the unleveraged asset when volatility is high. For example, see year 2 (1969). While U.S. stocks and LTTs were down 10.3% and 5.8% respectively, the 3x counterparts were down 38.0% and 31.7%. This results in the Lifecycle w Levered Bonds strategy having a much worse return. The same thing happens in year 11 (1978).

Of course, there are multiple ways to build a glidepath for a leveraged fund with stocks and bonds. I chose a simple method of maintaining a constant stock:bond ratio using a target equity allocation consistent with the standard lifecycle strategy. I'm sure there are ways to beat this, but I did not want to backtest to tune parameters, which would certainly result in overfitting.
Been meaning to get back to you on your previous post. Hopefully will have time to take a more thorough look in coming days but for the time being:

1) Adding bonds on top of a 2x equity lifecycle strategy won't do well because it's simply too much leverage and you get decay or possible total loss. It's also hurts that the leveraged portion of your lifecycle is just 10 years.

2) I don't touch LTT because the risk-adjusted returns are terrible compared to short and intermediate. That relationship has held true for over a century.

3) Standard lifecycle still beats unleveraged glidepath despite having a pretty short period of leverage and starting with 1 million. Since this isn't really the intended application of lifecyle, it's not surprising it doesn't do a lot better. If you started with $0 or from a younger age it would do better. I'm sure you know this already but figured I would mention again. How old are you by the way?

4) Most of your results are due to which strategies contain more overall leverage.

a) UPRO+TMF has the most leverage by far and has the widest variability and best top end results.

b) Standard lifecycle beats 100% VTSAX at the low end due to the time diversification, but fails at the top end because 100% VTSAX is more leveraged for the last 25 years of the simulation (100% vs 70%). The whole purpose of lifecycle is to win at that low-end, which it does. A more fair comparison would be starting 200% ending 100%. If you are considering HFEA or 2x 70-30 for your whole portfolio even late in life, you obviously have an extremely high risk tolerance. If it's not your whole portfolio it's not really a fair comparison is it because you constrain risk in the lifecycle and glidepath scenario late in life such that they would be appropriate for whole portfolio. The risk in HFEA and 2x 70-30 remains unconstrained throughout.. which is fine if that's your risk tolerance but you can't just say well with those strategies they wouldn't be my whole portfolio because with lifecycel the whole idea is that IS your whole portfolio.

c) Unleveraged glidepath has time diversification so the bottom end is a bit better than VTSAX but it has less leverage overall than 100% VTSAX or standard lifecycle, so the top-end is low.

d) 2x 70-30 is the most leveraged other than UPRO+TMF, so has a wide range with very high top end.

e) Lifecycle with leveraged bonds is simply too much leverage for too short a period and does poorly. It's sort of a whacky strategy to be 3x 70-30 for 10 years and then quickly drop off to 1x 70-30. It ends up not being very much leverage overall... for nearly the last 25 years it only has 70% in stock - less than a 100% VTSAX strategy.



The lifecycle book proved that the lifecycle starting at age 20 *always* wins. I guess what you are proving is that the lifecycle starting at age 45 with a 35 year time horizon doesn't always win. Although the traditional lifecycle still edges it out over 100% VTSAX because it wins in the lower percentiles which are the ones you care about more (percentiles 0-50... the 100th percentile is just fun to look at). And you've also proved that adding LTT leverage on top of a lifecycle strategy is too much leverage especially if you're only going to do it for 10 years. The more leverage you have the longer you need for it to pay off.

I'd also point out if you're willing to put any amount of money towards a 2x 70-30 or a HFEA strategy, you should be comfortable with a more gradual glidepath as well in your lifecycle strategies with a longer period of leverage.

One last observation... LTT 3x does terribly in the 70s and 80s ... even compared to VLGSX... LTT has borrowing costs which were huge and there was very little spread between STT and LTT prior to 1990.. there was no premium to capture via leverage. 3x leverage LTT prior to 1990 is just dumb IMO.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

TheCleverest wrote: Mon Aug 09, 2021 11:04 pm I'm curious. If a lottery ticket and only 5-15% of your portfolio is what you're willing to use, then have you considered HFEA vs...
a) SCV equities
b) VC funds
c) real estate deals
d) speculative digital assets
e) Frontier and Emerging Markets

If so, what were your thoughts?

I'm not trying to agitate you. Rather, I'd like to see what purpose you're attempting to achieve (upside, volatility, risk parity, etc).
While using only ~10% of your portfolio was Hedgefundie's recommendation, I am not interested in that. For a strategy to be meaningful to me, it must measure the impact on a total portfolio basis. To the extent holding some allocation of the risky assets you mention has been shown to increase overall returns or reduce downside, I would be interested. But I generally believe in total market funds rather than specific slices, and I suspect the other assets would be difficult to test.

I do have some equity in a private company from prior employer options, but I plan to unload that and transfer to a total market fund as soon as I can.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

skierincolorado wrote: Mon Aug 09, 2021 11:36 pm
The lifecycle book proved that the lifecycle starting at age 20 *always* wins. I guess what you are proving is that the lifecycle starting at age 45 with a 35 year time horizon doesn't always win. Although the traditional lifecycle still edges it out over 100% VTSAX because it wins in the lower percentiles which are the ones you care about more (percentiles 0-50... the 100th percentile is just fun to look at). And you've also proved that adding LTT leverage on top of a lifecycle strategy is too much leverage especially if you're only going to do it for 10 years. The more leverage you have the longer you need for it to pay off.

I'd also point out if you're willing to put any amount of money towards a 2x 70-30 or a HFEA strategy, you should be comfortable with a more gradual glidepath as well in your lifecycle strategies with a longer period of leverage.

One last observation... LTT 3x does terribly in the 70s and 80s ... even compared to VLGSX... LTT has borrowing costs which were huge and there was very little spread between STT and LTT prior to 1990.. there was no premium to capture via leverage. 3x leverage LTT prior to 1990 is just dumb IMO.
Thank you for your observations. I agree on every front. I am not trying to invalidate the Lifecycle strategy by any means. The logic is sound, and as you said, when used over an entire accumulation phase, it wins every time vs. a more typical glidepath. I did not even include a typical glidepath strategy in this analysis, because it would certainly lag the Lifecycle strategy. The Unlevered Lifecycle strategy I did include is more aggressive with equity allocation than a typical glidepath you'd find in a target retirement fund.

To clarify, this analysis was done with a total portfolio allocation for all strategies, including HFEA. And I am looking to decide how to allocate my total portfolio, not just a portion. I am 30 years old and plan to be in the accumulation phase until 65, which is why I included 35 years in this analysis. I do have a high risk tolerance and would be comfortable with a more gradual glidepath; I was simply using the leverage calculation from Lifecycle Investing. I did not include Social Security in my analysis, though, which would create a more gradual glidepath.

I did not have any particular reason for using LTT in this analysis other than it being used in HFEA. Also, the correlation between annual returns of VLGSX and VBTLX is 85%, so I thought it was an okay stand-in for a bonds in general. I am interested in an alternative bond fund if you know of one that has historical data for a leveraged fund.
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

millennialmillions wrote: Tue Aug 10, 2021 9:26 am
skierincolorado wrote: Mon Aug 09, 2021 11:36 pm
The lifecycle book proved that the lifecycle starting at age 20 *always* wins. I guess what you are proving is that the lifecycle starting at age 45 with a 35 year time horizon doesn't always win. Although the traditional lifecycle still edges it out over 100% VTSAX because it wins in the lower percentiles which are the ones you care about more (percentiles 0-50... the 100th percentile is just fun to look at). And you've also proved that adding LTT leverage on top of a lifecycle strategy is too much leverage especially if you're only going to do it for 10 years. The more leverage you have the longer you need for it to pay off.

I'd also point out if you're willing to put any amount of money towards a 2x 70-30 or a HFEA strategy, you should be comfortable with a more gradual glidepath as well in your lifecycle strategies with a longer period of leverage.

One last observation... LTT 3x does terribly in the 70s and 80s ... even compared to VLGSX... LTT has borrowing costs which were huge and there was very little spread between STT and LTT prior to 1990.. there was no premium to capture via leverage. 3x leverage LTT prior to 1990 is just dumb IMO.
Thank you for your observations. I agree on every front. I am not trying to invalidate the Lifecycle strategy by any means. The logic is sound, and as you said, when used over an entire accumulation phase, it wins every time vs. a more typical glidepath. I did not even include a typical glidepath strategy in this analysis, because it would certainly lag the Lifecycle strategy. The Unlevered Lifecycle strategy I did include is more aggressive with equity allocation than a typical glidepath you'd find in a target retirement fund.

To clarify, this analysis was done with a total portfolio allocation for all strategies, including HFEA. And I am looking to decide how to allocate my total portfolio, not just a portion. I am 30 years old and plan to be in the accumulation phase until 65, which is why I included 35 years in this analysis. I do have a high risk tolerance and would be comfortable with a more gradual glidepath; I was simply using the leverage calculation from Lifecycle Investing. I did not include Social Security in my analysis, though, which would create a more gradual glidepath.

I did not have any particular reason for using LTT in this analysis other than it being used in HFEA. Also, the correlation between annual returns of VLGSX and VBTLX is 85%, so I thought it was an okay stand-in for a bonds in general. I am interested in an alternative bond fund if you know of one that has historical data for a leveraged fund.
I thought the "unlevered glidepath" in your chart was a more typical glidepath?

One thing to experiment with would be making the average risk/leverage across all scenarios equal or similar. You've proved if you take reasonable amounts of risk for long enough, you win even at the low percentiles, at least historically (in the ~30 cohorts tested). The problem is a 2x 70/30 is probably more risk than even a risk tolerant person would want at age 65, partially because the future may not be like the past.

The problems is it's really not a long enough period to test. Bonds are going to look bad in any lifecycle+bonds approach because bonds did poorly in the first half 1955-1982. And it's going to exaggerate the benefits of having leverage when you're old because most of the cohorts end in a 40 year bull stock market.

I played around with your spreadsheet and a 150% stock + bonds glidepath down to 100% + bonds tested well. I had to edit the spreadsheet so the bonds are a constant proportion of the stock, because right now it only does bonds if there is equity leverage. If you switched out LTT for ITT I think you'd see a solid improvement. Holding 1.5-2x as much ITT significantly outperforms LTT, especially prior to 1990. The yields on LTT were only a hair above ITT prior to 1990, but with ~2x the duration risk -> volatility decay. Risk adjusted returns on LTT since 1990 are better, but still well short of ITT and STT.
TheCleverest
Posts: 39
Joined: Sat Dec 26, 2015 11:14 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by TheCleverest »

millennialmillions wrote: Sun Aug 08, 2021 10:09 pm That makes sense for my situation - since I already have a sizable portfolio, significant leverage is only used for the next ~5 years.
Labeling what you wrote as 'lifecycle investing' is not consistent with what Ayres and Nalebuff wrote.

On page 38 they suggest the following:
Phase 1: A leverage of 2:1 with usually first ten years of one's career.
Phase 2: Usually lasts until mid-50s with a leverage between 2:1 and 1:1.
Phase 3: Lasts until retirement with a portfolio that is fully unleveraged and includes bonds.

So,
1. Five years as you proposed is not the 10 years they wrote about.
2. If you are 30 now that means Phase 2 will last roughly for 15 years (Age 40 to mid 50s). Your calculations go from 200% to 70%. You essentially go from Phase 1 to Phase 3.

This is not lifecycle investing.
No use in being clever - have to be the cleverest
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

TheCleverest wrote: Tue Aug 10, 2021 4:38 pm
millennialmillions wrote: Sun Aug 08, 2021 10:09 pm That makes sense for my situation - since I already have a sizable portfolio, significant leverage is only used for the next ~5 years.
Labeling what you wrote as 'lifecycle investing' is not consistent with what Ayres and Nalebuff wrote.

On page 38 they suggest the following:
Phase 1: A leverage of 2:1 with usually first ten years of one's career.
Phase 2: Usually lasts until mid-50s with a leverage between 2:1 and 1:1.
Phase 3: Lasts until retirement with a portfolio that is fully unleveraged and includes bonds.

So,
1. Five years as you proposed is not the 10 years they wrote about.
2. If you are 30 now that means Phase 2 will last roughly for 15 years (Age 40 to mid 50s). Your calculations go from 200% to 70%. You essentially go from Phase 1 to Phase 3.

This is not lifecycle investing.
This analysis exactly follows the Lifecycle Investing strategy, as outlined by Ayres and Nalebuff. You removed the word "typically" from their summary of how it usually works. Read the section before that summary, which describes the precise methodology. It involves calculating the present value of your future contributions and adding that to your current portfolio, then multiplying that total lifetime savings amount by the target equity percentage. This gives a target equity amount, which is then capped at 200% leverage.

That's exactly what my spreadsheet does. The glidepath depends a bit on early portfolio performance, but it typically gets to 100% equities around year 10 (age 40 for me) and gradually drops down to 70% equities.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

skierincolorado wrote: Tue Aug 10, 2021 11:07 am I thought the "unlevered glidepath" in your chart was a more typical glidepath?

One thing to experiment with would be making the average risk/leverage across all scenarios equal or similar. You've proved if you take reasonable amounts of risk for long enough, you win even at the low percentiles, at least historically (in the ~30 cohorts tested). The problem is a 2x 70/30 is probably more risk than even a risk tolerant person would want at age 65, partially because the future may not be like the past.

The problems is it's really not a long enough period to test. Bonds are going to look bad in any lifecycle+bonds approach because bonds did poorly in the first half 1955-1982. And it's going to exaggerate the benefits of having leverage when you're old because most of the cohorts end in a 40 year bull stock market.

I played around with your spreadsheet and a 150% stock + bonds glidepath down to 100% + bonds tested well. I had to edit the spreadsheet so the bonds are a constant proportion of the stock, because right now it only does bonds if there is equity leverage. If you switched out LTT for ITT I think you'd see a solid improvement. Holding 1.5-2x as much ITT significantly outperforms LTT, especially prior to 1990. The yields on LTT were only a hair above ITT prior to 1990, but with ~2x the duration risk -> volatility decay. Risk adjusted returns on LTT since 1990 are better, but still well short of ITT and STT.
My "unlevered glidepath" approach still follows the Lifecycle Investing strategy of calculating "total lifetime savings", including the present value of future contributions. This is then used to calculate a target equity percentage, and the actual equity is capped at 100%. This method results in holding 100% equities for several years before starting the glide down to 70%. The exact timing depends on portfolio performance. This was called out as an alternative in the Lifecycle Investing book for those who don't have access to leverage / can't tolerate the drawdowns, and it significantly outperforms the typical glidepath you'd see in a target retirement fund.

To your point on the total leverage being too high in the Lifecycle w Levered Bonds approach, I tried capping total leverage at 200% rather than capping equities at 200%. So this strategy no longer uses more leverage than the normal Lifecycle approach without bonds while leveraged.

Image

As you can see, it reduces the downside, as predicted. But it still performs worse than the strategy without leveraged bonds.

Your points on LTT vs. ITT make sense. Ultimately I would like to use a total bond fund. The thing I am having trouble with is the 3x LTT EFT is the only one with much (simulated) historical data. Unfortunately simulating historical returns of an ETF (or an alternative method of leveraging) is beyond my capability.
bumbojumbo
Posts: 56
Joined: Sun Feb 04, 2018 3:25 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by bumbojumbo »

Does anyone know why long-term treasuries (LTT) have had lower risk-adjusted returns than STT or ITT?
Bentonkb
Posts: 48
Joined: Sun Mar 07, 2021 6:47 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Bentonkb »

I've read and re-read both the Hedgefundie and MarketTimer threads a couple times. The idea of leveraging a reduced-volatility portfolio is compelling and I've introduced it into my own investing using futures contracts.

The whole idea of analyzing the Sharpe ratio of individual components of the portfolio bugs me. My gut says that this is the wrong path. It makes me even more nervous to see people analyzing the leverage ratio for individual portfolio components and choosing, for example, ITT over LTT based on their individual Sharpe ratios.

I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
Rob Bertram
Posts: 857
Joined: Mon May 05, 2014 12:15 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Rob Bertram »

millennialmillions wrote: Sun Aug 08, 2021 10:09 pmMy current allocation is 100% equities (70% domestic, 30% international). Before I make any changes to my portfolio, I wanted to seek the wisdom of the Bogleheads to correct any mistakes and point out aspects I am not considering. I also have a few more research items I want to complete, though I am hesitant to do much more backtesting for risk of overfitting.
Here are some questions to ask yourself:
  • If you are going to be leveraged, what real role do bonds play in your portfolio? Bonds create drag during the accumulation phase and reduce your return. Lowering your leverage ratio does a better job of reducing your risk than adding bonds, so they don't have a meaningful role in your glide path to lower risk.
  • If you are comfortable with the concept of going 200% stocks -100% cash early in the lifecycle, how do you feel about staying at 150% stock -50% cash indefinitely? If you believe that markets will eventually revert to the mean, this portfolio would be unlikely to suffer a margin call (assuming a 10% maintenance margin) if it were to experience a 55% drop. If that has you too nervous, how about 140% stock?1.12
  • Let's say that this leveraged approach returns 12%/year for 35 years. That is a 52.8x growth (1.12^52 = 52.7996). When you are ready to take income from your portfolio, would you consider withdrawing cash from the portfolio as a margin loan instead of selling assets? With a sufficiently large portfolio, this would be an insignificant change in your leverage ratio.
Hydromod
Posts: 663
Joined: Tue Mar 26, 2019 10:21 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Hydromod »

Bentonkb wrote: Wed Aug 11, 2021 7:31 am I've read and re-read both the Hedgefundie and MarketTimer threads a couple times. The idea of leveraging a reduced-volatility portfolio is compelling and I've introduced it into my own investing using futures contracts.

The whole idea of analyzing the Sharpe ratio of individual components of the portfolio bugs me. My gut says that this is the wrong path. It makes me even more nervous to see people analyzing the leverage ratio for individual portfolio components and choosing, for example, ITT over LTT based on their individual Sharpe ratios.

I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
I agree that the Sharpe ratio and leverage ratio are mostly useful when they are applied to the whole portfolio. Perhaps there is some information about how often to rebalance when applied to individual components, but I haven't seen anything along those lines.
econalex
Posts: 36
Joined: Fri Aug 31, 2018 9:44 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by econalex »

For a brief moment I thought OP was the alter ego of dear me, since I'm the same age (welp 31 :oops: ), have the exact same asset allocation (70 vti 30 vxus and just adding edv) and wondering the same thing (HFEA and adding leverage to stock/bond portfolio).

I just want to thank OP and replies to have done the numbers and offered their sage advices! Indeed I just moved all of my Roth to HFEA and was planning to do something for my 401k (currently in Vanguard target date 2060, which is 100% boglehead-approved I believe). I was thinking about PSLDX but not sure if the company plan offers that. Anyway I've been actively interviewing so there's that :sharebeer

I'm also worried about rising rate but have to confess that I know nothing about macroeconomics----indeed I failed macro for my PhD qualifying exam some years before :twisted:
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

Bentonkb wrote: Wed Aug 11, 2021 7:31 am I've read and re-read both the Hedgefundie and MarketTimer threads a couple times. The idea of leveraging a reduced-volatility portfolio is compelling and I've introduced it into my own investing using futures contracts.

The whole idea of analyzing the Sharpe ratio of individual components of the portfolio bugs me. My gut says that this is the wrong path. It makes me even more nervous to see people analyzing the leverage ratio for individual portfolio components and choosing, for example, ITT over LTT based on their individual Sharpe ratios.

I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
How far back are you going with PV? And what tickers are you using? I'd like to run it myself. I never figured out how to leverage the PV asset class side, and I don't know what tickers to use to go prior to year 2000.

Unless you are getting back prior to 1980, the sharpe ratio of LTT is basically inflated due to falling rates. And when you take this inflated performance and inflated sharpe and combine it with stocks, it does a lot better than if you backtested prior to 1980. There's simply no way in my mind that the small amount of extra yield you get by going from 8 years to 20+ years of duration risk is worth it unless you are a pension fund trying to duration match your liabilities and holding the bonds to expiration.

We're seeing in the OPs testing how backtesting prior to 1980 with LTT is not only a worse sharpe ratio, but also has worse absolute performance than not including treasuries at all. Not only do they fail to realize any diversification benefit, they have a significant negative effect. LTT prior to 1980 are a portfolio wrecking ball.

I'm not suggesting picking ITT over LTT based on individual sharpe ratios. I'm suggesting picking it over LTT because the portfolio as a whole will do better and have a better sharpe ratio. We're seeing how poorly LTT are doing in the OP's own testing.
Last edited by skierincolorado on Wed Aug 11, 2021 10:53 am, edited 1 time in total.
klaus14
Posts: 765
Joined: Sun Nov 25, 2018 7:43 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by klaus14 »

Rob Bertram wrote: Wed Aug 11, 2021 8:26 am
millennialmillions wrote: Sun Aug 08, 2021 10:09 pmMy current allocation is 100% equities (70% domestic, 30% international). Before I make any changes to my portfolio, I wanted to seek the wisdom of the Bogleheads to correct any mistakes and point out aspects I am not considering. I also have a few more research items I want to complete, though I am hesitant to do much more backtesting for risk of overfitting.
Here are some questions to ask yourself:
  • If you are going to be leveraged, what real role do bonds play in your portfolio? Bonds create drag during the accumulation phase and reduce your return. Lowering your leverage ratio does a better job of reducing your risk than adding bonds, so they don't have a meaningful role in your glide path to lower risk.
  • If you are comfortable with the concept of going 200% stocks -100% cash early in the lifecycle, how do you feel about staying at 150% stock -50% cash indefinitely? If you believe that markets will eventually revert to the mean, this portfolio would be unlikely to suffer a margin call (assuming a 10% maintenance margin) if it were to experience a 55% drop. If that has you too nervous, how about 140% stock?1.12
  • Let's say that this leveraged approach returns 12%/year for 35 years. That is a 52.8x growth (1.12^52 = 52.7996). When you are ready to take income from your portfolio, would you consider withdrawing cash from the portfolio as a margin loan instead of selling assets? With a sufficiently large portfolio, this would be an insignificant change in your leverage ratio.
I think "expected returns" needs to be considered.
* With treasury futures, you are borrowing with close to zero rate and earning like 2% for the long term bonds.
* US Stocks have very low expected returns as of now (based on forward PE), this is not a good time to lever. I think leverage should be spared for a time when expected returns are normal/high.
My investment algorithm: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=351899&p=6112869#p6112869
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

klaus14 wrote: Wed Aug 11, 2021 10:08 am
Rob Bertram wrote: Wed Aug 11, 2021 8:26 am
millennialmillions wrote: Sun Aug 08, 2021 10:09 pmMy current allocation is 100% equities (70% domestic, 30% international). Before I make any changes to my portfolio, I wanted to seek the wisdom of the Bogleheads to correct any mistakes and point out aspects I am not considering. I also have a few more research items I want to complete, though I am hesitant to do much more backtesting for risk of overfitting.
Here are some questions to ask yourself:
  • If you are going to be leveraged, what real role do bonds play in your portfolio? Bonds create drag during the accumulation phase and reduce your return. Lowering your leverage ratio does a better job of reducing your risk than adding bonds, so they don't have a meaningful role in your glide path to lower risk.
  • If you are comfortable with the concept of going 200% stocks -100% cash early in the lifecycle, how do you feel about staying at 150% stock -50% cash indefinitely? If you believe that markets will eventually revert to the mean, this portfolio would be unlikely to suffer a margin call (assuming a 10% maintenance margin) if it were to experience a 55% drop. If that has you too nervous, how about 140% stock?1.12
  • Let's say that this leveraged approach returns 12%/year for 35 years. That is a 52.8x growth (1.12^52 = 52.7996). When you are ready to take income from your portfolio, would you consider withdrawing cash from the portfolio as a margin loan instead of selling assets? With a sufficiently large portfolio, this would be an insignificant change in your leverage ratio.
I think "expected returns" needs to be considered.
* With treasury futures, you are borrowing with close to zero rate and earning like 2% for the long term bonds.
* US Stocks have very low expected returns as of now (based on forward PE), this is not a good time to lever. I think leverage should be spared for a time when expected returns are normal/high.
The yield curve is actually moderately steep right now. That is the difference between borrowing at 0% and lending at 1% (ITT) is near the historical norm. Of course the curve is only half the story since the whole curve could shift up and you'd lose money. But if you believe in EMH you shouldn't really be trying to time any of this. I'll admit I haven't gone all in on lifecycle/leverage for the reasons you suggest.

It's far riskier to try to time the market than to implement a lifecycle strategy.
Last edited by skierincolorado on Wed Aug 11, 2021 11:42 am, edited 1 time in total.
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

Bentonkb wrote: Wed Aug 11, 2021 7:31 am I've read and re-read both the Hedgefundie and MarketTimer threads a couple times. The idea of leveraging a reduced-volatility portfolio is compelling and I've introduced it into my own investing using futures contracts.

The whole idea of analyzing the Sharpe ratio of individual components of the portfolio bugs me. My gut says that this is the wrong path. It makes me even more nervous to see people analyzing the leverage ratio for individual portfolio components and choosing, for example, ITT over LTT based on their individual Sharpe ratios.

I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
The two most recent periods of flat or nearly flat rates are 2012-2019 (flat rates) and 2003-2018 (slightly falling rates). Over both periods a portfolio of 100% VTI + 200% ITT (VFITX) had a higher sharpe ratio than a portfolio of 100% VTI + 100% LTT (VUSTX). The bet on LTT relies on continuously falling rates. Otherwise the duration risk has not been adequately rewarded. This is most likely due to leverage-constrained entities chasing yield and/or pensions duration matching their liabilities.

I've done crude backtesting, similar to the OPs, prior to 1980 and the results are even more heavily skewed toward ITT.

The yield premium on LTT prior to 1990 was so small you'd be crazy to lever into them. After 1990 it's been noticeably better. The sharpe ratio of a portfolio paired with VTI is still better with ITT once you account for falling rates, but it's not as lopsided a win for ITT as it was before 1990.

Also just testing the longest period possible with the tickers I chose, 2001-present, the sharpe ratio of the portfolio with ITT is significantly better, despite falling rates.


If you're using leveraged ETFs with high borrowing costs and high fees, LTT might win because of the lower amount of money required to be invested to gain the same risk/return. But if you're using futures contracts, ITT is the clear winner in my backtesting. Especially prior to 1990.
Bentonkb
Posts: 48
Joined: Sun Mar 07, 2021 6:47 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Bentonkb »

skierincolorado wrote: Wed Aug 11, 2021 11:08 am
If you're using leveraged ETFs with high borrowing costs and high fees, LTT might win because of the lower amount of money required to be invested to gain the same risk/return. But if you're using futures contracts, ITT is the clear winner in my backtesting. Especially prior to 1990.
You might be right. Maybe it is just my squeemishness about high leverage ratios that is preventing me from shifting from /ZB (12 year duration) to /ZN (6 year duration). I guess I've already moved a bit in that direction because I shifted 1/5 of my bond exposure to GOVT, which had a similar duration to /ZN last time I checked.
Hydromod
Posts: 663
Joined: Tue Mar 26, 2019 10:21 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Hydromod »

skierincolorado wrote: Wed Aug 11, 2021 11:08 am
Bentonkb wrote: Wed Aug 11, 2021 7:31 am I've read and re-read both the Hedgefundie and MarketTimer threads a couple times. The idea of leveraging a reduced-volatility portfolio is compelling and I've introduced it into my own investing using futures contracts.

The whole idea of analyzing the Sharpe ratio of individual components of the portfolio bugs me. My gut says that this is the wrong path. It makes me even more nervous to see people analyzing the leverage ratio for individual portfolio components and choosing, for example, ITT over LTT based on their individual Sharpe ratios.

I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
The two most recent periods of flat or nearly flat rates are 2012-2019 (flat rates) and 2003-2018 (slightly falling rates). Over both periods a portfolio of 100% VTI + 200% ITT (VFITX) had a higher sharpe ratio than a portfolio of 100% VTI + 100% LTT (VUSTX). The bet on LTT relies on continuously falling rates. Otherwise the duration risk has not been adequately rewarded. This is most likely due to leverage-constrained entities chasing yield and/or pensions duration matching their liabilities.

I've done crude backtesting, similar to the OPs, prior to 1980 and the results are even more heavily skewed toward ITT.

The yield premium on LTT prior to 1990 was so small you'd be crazy to lever into them. After 1990 it's been noticeably better. The sharpe ratio of a portfolio paired with VTI is still better with ITT once you account for falling rates, but it's not as lopsided a win for ITT as it was before 1990.

Also just testing the longest period possible with the tickers I chose, 2001-present, the sharpe ratio of the portfolio with ITT is significantly better, despite falling rates.

If you're using leveraged ETFs with high borrowing costs and high fees, LTT might win because of the lower amount of money required to be invested to gain the same risk/return. But if you're using futures contracts, ITT is the clear winner in my backtesting. Especially prior to 1990.
Since 1986, 55/45 UPRO/TMF has the same ratio of equity to treasury volatility as 40/60 UPRO/TYD, albeit with higher volatility and smaller Sharpe ratio.

The bet on LTT gets you a larger UPRO allocation. That higher UPRO allocation corresponds to roughly 1/3 larger CAGR (e.g., 15% goes to 20% CAGR), assuming zero returns from treasuries. That's a significant increase.

Going from 33/67 to 50/50 gets you roughly 50% larger CAGR, again assuming zero treasury returns.

If and when rates prior to 1980 return, time to revisit the strategy.
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

Hydromod wrote: Wed Aug 11, 2021 12:03 pm
skierincolorado wrote: Wed Aug 11, 2021 11:08 am
Bentonkb wrote: Wed Aug 11, 2021 7:31 am I've read and re-read both the Hedgefundie and MarketTimer threads a couple times. The idea of leveraging a reduced-volatility portfolio is compelling and I've introduced it into my own investing using futures contracts.

The whole idea of analyzing the Sharpe ratio of individual components of the portfolio bugs me. My gut says that this is the wrong path. It makes me even more nervous to see people analyzing the leverage ratio for individual portfolio components and choosing, for example, ITT over LTT based on their individual Sharpe ratios.

I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
The two most recent periods of flat or nearly flat rates are 2012-2019 (flat rates) and 2003-2018 (slightly falling rates). Over both periods a portfolio of 100% VTI + 200% ITT (VFITX) had a higher sharpe ratio than a portfolio of 100% VTI + 100% LTT (VUSTX). The bet on LTT relies on continuously falling rates. Otherwise the duration risk has not been adequately rewarded. This is most likely due to leverage-constrained entities chasing yield and/or pensions duration matching their liabilities.

I've done crude backtesting, similar to the OPs, prior to 1980 and the results are even more heavily skewed toward ITT.

The yield premium on LTT prior to 1990 was so small you'd be crazy to lever into them. After 1990 it's been noticeably better. The sharpe ratio of a portfolio paired with VTI is still better with ITT once you account for falling rates, but it's not as lopsided a win for ITT as it was before 1990.

Also just testing the longest period possible with the tickers I chose, 2001-present, the sharpe ratio of the portfolio with ITT is significantly better, despite falling rates.

If you're using leveraged ETFs with high borrowing costs and high fees, LTT might win because of the lower amount of money required to be invested to gain the same risk/return. But if you're using futures contracts, ITT is the clear winner in my backtesting. Especially prior to 1990.
Since 1986, 55/45 UPRO/TMF has the same ratio of equity to treasury volatility as 40/60 UPRO/TYD, albeit with higher volatility and smaller Sharpe ratio.

The bet on LTT gets you a larger UPRO allocation. That higher UPRO allocation corresponds to roughly 1/3 larger CAGR (e.g., 15% goes to 20% CAGR), assuming zero returns from treasuries. That's a significant increase.

Going from 33/67 to 50/50 gets you roughly 50% larger CAGR, again assuming zero treasury returns.

If and when rates prior to 1980 return, time to revisit the strategy.
Only applicable in the world of LETF. With futures there are no such constraints.

If/when rates return to rates prior to 1980 it's too late... the TMF has created a -60%+ drag on your portfolio at that point as you rebalance into a losing asset. ITT would do poorly of course too, but not as poorly.
Bentonkb
Posts: 48
Joined: Sun Mar 07, 2021 6:47 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Bentonkb »

skierincolorado wrote: Wed Aug 11, 2021 9:59 am
Bentonkb wrote: Wed Aug 11, 2021 7:31 am
I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
How far back are you going with PV? And what tickers are you using? I'd like to run it myself. I never figured out how to leverage the PV asset class side, and I don't know what tickers to use to go prior to year 2000.

Unless you are getting back prior to 1980, the sharpe ratio of LTT is basically inflated due to falling rates. And when you take this inflated performance and inflated sharpe and combine it with stocks, it does a lot better than if you backtested prior to 1980. There's simply no way in my mind that the small amount of extra yield you get by going from 8 years to 20+ years of duration risk is worth it unless you are a pension fund trying to duration match your liabilities and holding the bonds to expiration.

We're seeing in the OPs testing how backtesting prior to 1980 with LTT is not only a worse sharpe ratio, but also has worse absolute performance than not including treasuries at all. Not only do they fail to realize any diversification benefit, they have a significant negative effect. LTT prior to 1980 are a portfolio wrecking ball.
I did the asset class backtests for an unlevered portfolio starting in (I think) 1978. I ran it for 10 year periods starting in every even-numbered year. The levered portfolio backtests were only done back to 2005. I was mostly looking at maximum drawdowns rather than trying to find maximum returns. The Sharpe ratio isn't the best tool in my case. The maximum drawdowns tend to be fat-tail events and the Sharpe ratio tells you more about the Gaussian middle part of the distribution.

I don't know about earlier, but the 1978 to 1985 backtest for stocks and LTT is significantly improved by adding gold to the portfolio. Lots of people hate the idea of gold, but I prefer the idea of owning gold over cash or short term treasuries. ITT might be a good middle ground, I don't know.

In one of the threads (I can't recall which one) there was a discussion of going down to 2 year notes and jacking up the leverage to 20x. I just can't bring myself to consider it. I might be able to talk myself into using 10 year notes and 5x leverage.
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

Bentonkb wrote: Wed Aug 11, 2021 12:12 pm
skierincolorado wrote: Wed Aug 11, 2021 9:59 am
Bentonkb wrote: Wed Aug 11, 2021 7:31 am
I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
How far back are you going with PV? And what tickers are you using? I'd like to run it myself. I never figured out how to leverage the PV asset class side, and I don't know what tickers to use to go prior to year 2000.

Unless you are getting back prior to 1980, the sharpe ratio of LTT is basically inflated due to falling rates. And when you take this inflated performance and inflated sharpe and combine it with stocks, it does a lot better than if you backtested prior to 1980. There's simply no way in my mind that the small amount of extra yield you get by going from 8 years to 20+ years of duration risk is worth it unless you are a pension fund trying to duration match your liabilities and holding the bonds to expiration.

We're seeing in the OPs testing how backtesting prior to 1980 with LTT is not only a worse sharpe ratio, but also has worse absolute performance than not including treasuries at all. Not only do they fail to realize any diversification benefit, they have a significant negative effect. LTT prior to 1980 are a portfolio wrecking ball.
I did the asset class backtests for an unlevered portfolio starting in (I think) 1978. I ran it for 10 year periods starting in every even-numbered year. The levered portfolio backtests were only done back to 2005. I was mostly looking at maximum drawdowns rather than trying to find maximum returns. The Sharpe ratio isn't the best tool in my case. The maximum drawdowns tend to be fat-tail events and the Sharpe ratio tells you more about the Gaussian middle part of the distribution.

I don't know about earlier, but the 1978 to 1985 backtest for stocks and LTT is significantly improved by adding gold to the portfolio. Lots of people hate the idea of gold, but I prefer the idea of owning gold over cash or short term treasuries. ITT might be a good middle ground, I don't know.

In one of the threads (I can't recall which one) there was a discussion of going down to 2 year notes and jacking up the leverage to 20x. I just can't bring myself to consider it. I might be able to talk myself into using 10 year notes and 5x leverage.
I mean once you're past 2x leverage it kind of stops mattering right. I suppose you might come out alive in a partial U.S. default (financially and literally).

EfficientInvestor is the one who posted about 20x leverage on STT in the Lifecycle thread. I think it's a fairly common strategy for hedge funds. There hasn't been much premium on STT lately so I haven't explored it yet.

In your 10 year period backtests, I assume ITT would have the lower drawdown if not using leverage. With leverage, I would still expect it to have the lower drawdown since the sharpe is better, although sharpe and max drawdown aren't the same. In the periods I tested levered VTI+ITT had lower max drawdown for the same return.

I've been meaning to explore some gold as balast but not sure it is worth the added complexity. I'll try to take a look though.

Personally I own ZF, but not as much leverage as you it sounds like. 5x on ZN sounds like a lot to me. I'll probably pick up one ZN at some point.
Last edited by skierincolorado on Wed Aug 11, 2021 12:35 pm, edited 1 time in total.
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

Bentonkb wrote: Wed Aug 11, 2021 12:12 pm
skierincolorado wrote: Wed Aug 11, 2021 9:59 am
Bentonkb wrote: Wed Aug 11, 2021 7:31 am
I think that the Sharpe ratio and leverage ratio are only useful when they are applied to the whole portfolio. When I run Portfolio Visualizer models of leveraged portfolios there doesn't seem to be a significant difference between stocks/LTT at a lower leverage and stocks/ITT with a higher leverage ratio.
How far back are you going with PV? And what tickers are you using? I'd like to run it myself. I never figured out how to leverage the PV asset class side, and I don't know what tickers to use to go prior to year 2000.

Unless you are getting back prior to 1980, the sharpe ratio of LTT is basically inflated due to falling rates. And when you take this inflated performance and inflated sharpe and combine it with stocks, it does a lot better than if you backtested prior to 1980. There's simply no way in my mind that the small amount of extra yield you get by going from 8 years to 20+ years of duration risk is worth it unless you are a pension fund trying to duration match your liabilities and holding the bonds to expiration.

We're seeing in the OPs testing how backtesting prior to 1980 with LTT is not only a worse sharpe ratio, but also has worse absolute performance than not including treasuries at all. Not only do they fail to realize any diversification benefit, they have a significant negative effect. LTT prior to 1980 are a portfolio wrecking ball.
I did the asset class backtests for an unlevered portfolio starting in (I think) 1978. I ran it for 10 year periods starting in every even-numbered year. The levered portfolio backtests were only done back to 2005. I was mostly looking at maximum drawdowns rather than trying to find maximum returns. The Sharpe ratio isn't the best tool in my case. The maximum drawdowns tend to be fat-tail events and the Sharpe ratio tells you more about the Gaussian middle part of the distribution.

I don't know about earlier, but the 1978 to 1985 backtest for stocks and LTT is significantly improved by adding gold to the portfolio. Lots of people hate the idea of gold, but I prefer the idea of owning gold over cash or short term treasuries. ITT might be a good middle ground, I don't know.

In one of the threads (I can't recall which one) there was a discussion of going down to 2 year notes and jacking up the leverage to 20x. I just can't bring myself to consider it. I might be able to talk myself into using 10 year notes and 5x leverage.
You might be interested in these two papers:

https://coexpartnersaig.files.wordpress ... -carry.pdf
https://www.pimco.com/handlers/displayd ... d%2BfxA%3D
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

Rob Bertram wrote: Wed Aug 11, 2021 8:26 am Here are some questions to ask yourself:
  • If you are going to be leveraged, what real role do bonds play in your portfolio? Bonds create drag during the accumulation phase and reduce your return. Lowering your leverage ratio does a better job of reducing your risk than adding bonds, so they don't have a meaningful role in your glide path to lower risk.
  • If you are comfortable with the concept of going 200% stocks -100% cash early in the lifecycle, how do you feel about staying at 150% stock -50% cash indefinitely? If you believe that markets will eventually revert to the mean, this portfolio would be unlikely to suffer a margin call (assuming a 10% maintenance margin) if it were to experience a 55% drop. If that has you too nervous, how about 140% stock?1.12
  • Let's say that this leveraged approach returns 12%/year for 35 years. That is a 52.8x growth (1.12^52 = 52.7996). When you are ready to take income from your portfolio, would you consider withdrawing cash from the portfolio as a margin loan instead of selling assets? With a sufficiently large portfolio, this would be an insignificant change in your leverage ratio.
My original thought, based on reading the HFEA and Lifecycle Investing threads, is that levering a portfolio closely to the efficient frontier can either get the same returns with less volatility or greater returns with the same volatility when compared to a 100% equity portfolio. And that there may be some rebalancing bonus from putting more into the asset that has recently underperformed.

You bring up a good point on maintaining leverage late in the lifecycle. The constant 2x 70/30 portfolio has 140% equities at the end, so it's not surprising it beats the Lifecycle 200% to 70% glidepath on average and in the upper percentiles. I just ran the simulation for a constant 200% equity allocation, and the results are very similar to the constant 2x 70/30 (the worst result is actually better for the constant 200% equity). So you're right that bonds aren't doing much for me when leveraged.
Bentonkb
Posts: 48
Joined: Sun Mar 07, 2021 6:47 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Bentonkb »

Exhibit 3 from the PIMCO paper makes 30Y treasuries look best too me. I'm looking for small, positive returns that are as stable as possible. That and volatility that is inversely correlated with stocks.
Rob Bertram
Posts: 857
Joined: Mon May 05, 2014 12:15 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Rob Bertram »

millennialmillions wrote: Wed Aug 11, 2021 8:47 pm My original thought, based on reading the HFEA and Lifecycle Investing threads, is that levering a portfolio closely to the efficient frontier can either get the same returns with less volatility or greater returns with the same volatility when compared to a 100% equity portfolio.

The concept that you are describing is the Capital Asset Pricing Model (CAPM).
Image

It is a very amazing piece of mathematics, but it does not work the way that we want or think. The efficient frontier in this case is based on expected future volatility, return, and correlation of assets. My crystal ball is hazy (at best) with return and volatility but is silent on correlation. Perhaps yours is better? Additionally, CAPM only applies to a specific period of time in the near future -- usually the next week or month. And then you need to redo your expected volatility, return and correlation to find a new optimal portfolio.
millennialmillions wrote: Wed Aug 11, 2021 8:47 pmAnd that there may be some rebalancing bonus from putting more into the asset that has recently underperformed.
From a long-term "buy and hold" perspective, the rebalancing bonus is a myth that we debunked on the forum. For periods where stocks and bonds had a negative correlation, there was a bonus. When they were positively correlated, there was a loss. For periods where there was no correlation, the total return was explained completely by the asset allocation.

In practice, stocks and bonds have completely different risk profiles. Rebalancing has you sell your riskier asset (stocks) and buying your safer asset (bonds). It is a fair way to reduce risk for those who are uncomfortable with the volatility of 100% stocks. For those who have the risk tolerance to go beyond 100% stocks, you have leverage which also allows you to control your risk.

Having two dials to control risk might be beneficial, but it means you would need to predict the future volatility, return, and correlation of bonds in addition to stocks. This drastically increases the complexity of your leveraged model as well as increases your dependence on your crystal ball. Personally, I like to keep things simple. Sticking with one risky asset class reduces my dependence on educated guesses and increases my confidence in my model.

Your comfort level may be different from mine which is why I suggested that you think heavily on the role of bonds in your leveraged portfolio.
millennialmillions wrote: Wed Aug 11, 2021 8:47 pm You bring up a good point on maintaining leverage late in the lifecycle. The constant 2x 70/30 portfolio has 140% equities at the end, so it's not surprising it beats the Lifecycle 200% to 70% glidepath on average and in the upper percentiles. I just ran the simulation for a constant 200% equity allocation, and the results are very similar to the constant 2x 70/30 (the worst result is actually better for the constant 200% equity). So you're right that bonds aren't doing much for me when leveraged.
From a modeling perspective, how do you handle market movements? I ask because this makes back testing and simulations complicated. For example, you have a 200% stock portfolio. The leveraged value is $200k ($100k yours + $100k margin).
Market moves 10% up:
  • The leveraged portfolio value is now $220k ($120k yours + $100k margin) for a ratio of 220k/120k = 183%
  • What would you do to get back to 200% leverage? I would buy an additional $20k assets and have a resulting portfolio of $240k ($120k yours + $120k margin)
  • How do you model this in your testing? You might be missing half of the benefit of leverage and time diversification.
Market moves 10% down:
  • The leveraged portfolio value is now $180k ($80k yours + $100k margin) for a ratio of 180k/80k = 225%
  • Does your model force you to get back to 200% leverage by selling assets? Ayers and Nalebuff suggest staying the course and wait for the market to revert to the mean. Or wait for regular contributions (or dividends) to restore balance while in the accumulation phase. My strategy follows this approach as well.
  • If you sell, how do you model this? If you sell, know that volatility will have a huge impact on returns. (This is the main reason that leveraged ETFs are suboptimal as they adjust assets daily.)
One last thing to consider: What is the goal of this portfolio? Are you aiming to have a huge pile of cash, or are you working to generate a stream of income? For most of us, income is our primary goal. If this is the same for you, I am going make an argument that your ending portfolio value is not a meaningful measurement. Instead, it should be the income stream that it generates.

Sequence of returns has been a huge factor in portfolio growth and its ability to provide a steady stream of income in our retirement years. So for example, high returns in the beginning of your portfolio lifecycle when its value is low combined with low or negative returns when you are drawing income from the portfolio will limit its income capability. Lifecycle investing attempts to mitigate this somewhat with time diversification by using higher amounts of leverage early in the lifecycle. This addresses the first half of sequence of returns.

Here is how you can mitigate sequence of returns at the end of your accumulation phase and into the income phase: In a traditional portfolio without leverage, we have bonds or fixed-income assets that we can sell when the stock market drops which allows us to hold onto stocks so that they can revert to the mean. With leverage, you have another option. You can withdraw that income as a margin loan against your portfolio without needing to sell any of your assets. With modest levels of leverage (e.g., 130 - 140% stock), your portfolio can weather a 55% market drop (like we saw in recent bear markets) while still supplying up to 3 years of income (3-4% of peak net value) without triggering a margin call or needing to sell assets.

When you are looking at back testing or simulation results, look at the income that they would generate. For the strategies that include leverage into the income phase, instead of using the final portfolio value, take the maximum net value over its history. I expect that most if not all of your worst-case simulations for the leveraged income portfolio would disappear.
Hydromod
Posts: 663
Joined: Tue Mar 26, 2019 10:21 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Hydromod »

Rob Bertram wrote: Thu Aug 12, 2021 11:59 am The concept that you are describing is the Capital Asset Pricing Model (CAPM).
Image

It is a very amazing piece of mathematics, but it does not work the way that we want or think. The efficient frontier in this case is based on expected future volatility, return, and correlation of assets. My crystal ball is hazy (at best) with return and volatility but is silent on correlation. Perhaps yours is better? Additionally, CAPM only applies to a specific period of time in the near future -- usually the next week or month. And then you need to redo your expected volatility, return and correlation to find a new optimal portfolio.

In practice, stocks and bonds have completely different risk profiles. Rebalancing has you sell your riskier asset (stocks) and buying your safer asset (bonds). It is a fair way to reduce risk for those who are uncomfortable with the volatility of 100% stocks. For those who have the risk tolerance to go beyond 100% stocks, you have leverage which also allows you to control your risk.

Having two dials to control risk might be beneficial, but it means you would need to predict the future volatility, return, and correlation of bonds in addition to stocks. This drastically increases the complexity of your leveraged model as well as increases your dependence on your crystal ball. Personally, I like to keep things simple. Sticking with one risky asset class reduces my dependence on educated guesses and increases my confidence in my model.
If you recognize that 3x LETFs in the HFEA benefit from fairly frequent rebalancing to correct for drift anyway, hewing to a CAPM-like approach is not such an ordeal (especially in tax-advantaged accounts).

I find that imposing minimum variance with a risk budget approach for all assets (no estimates of return but estimates of variance-covariance from recent months) actually works very well in backtesting. Frequent updates keep the estimates reasonable, and the matrix equation is no more complex for 2 or 10 funds once the software is set up.

But complexity is in the eye of the beholder, certainly.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

Rob Bertram wrote: Thu Aug 12, 2021 11:59 am Having two dials to control risk might be beneficial, but it means you would need to predict the future volatility, return, and correlation of bonds in addition to stocks. This drastically increases the complexity of your leveraged model as well as increases your dependence on your crystal ball. Personally, I like to keep things simple. Sticking with one risky asset class reduces my dependence on educated guesses and increases my confidence in my model.
Thanks Rob for the response. I'm not sure I agree that you need to be able to accurately predict volatility, return, and correlation of bonds to utilize a strategy that has leverage with stocks and bonds. With the backtesting approach, I am essentially saying that I believe future performance of stocks and bonds will be somewhat similar to historical periods. I think most investing strategies rely on that assumption. I am curious about your model, though. Can you share more about what "keeping things simple" means to you?
Rob Bertram wrote: Thu Aug 12, 2021 11:59 am From a modeling perspective, how do you handle market movements? I ask because this makes back testing and simulations complicated. For example, you have a 200% stock portfolio. The leveraged value is $200k ($100k yours + $100k margin).
My analysis is using a combination of 3x leveraged ETFs and non-leveraged funds to achieve the target leverage. The leveraged ETFs track daily 3x returns. I used annual data in the analysis, so rebalancing between funds according to the target leverage occurs annually.

Rob Bertram wrote: Thu Aug 12, 2021 11:59 am One last thing to consider: What is the goal of this portfolio? Are you aiming to have a huge pile of cash, or are you working to generate a stream of income? For most of us, income is our primary goal. If this is the same for you, I am going make an argument that your ending portfolio value is not a meaningful measurement. Instead, it should be the income stream that it generates.
My analysis only includes the accumulation phase. I wanted to limit the scope to decisions that are relevant to me now, so I'm assuming that the same strategy would be used in retirement across strategies. So money at the end of the accumulation phase is my target.
Rob Bertram
Posts: 857
Joined: Mon May 05, 2014 12:15 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Rob Bertram »

millennialmillions wrote: Thu Aug 12, 2021 7:24 pm Thanks Rob for the response. I'm not sure I agree that you need to be able to accurately predict volatility, return, and correlation of bonds to utilize a strategy that has leverage with stocks and bonds. With the backtesting approach, I am essentially saying that I believe future performance of stocks and bonds will be somewhat similar to historical periods. I think most investing strategies rely on that assumption. I am curious about your model, though. Can you share more about what "keeping things simple" means to you?
I had originally made the same assumption -- that I could simply use historical data to predict future returns, volatility, and correlations. The forum members politely corrected me of that mistake and provided data. There are some past events that will never happen again -- like decoupling of the gold standard in the 70s resulting in stagflation in the 70s and hyper inflation in the 80s. In other cases, our financial institutions learned from our mistakes of the 1990s and 2000s and have enacted policies that govern how the Fed influences interest rates.

For example, inflation and bond yields are low, and the general consensus is that they will stay low for the foreseeable future. If bond data from the 1980s are influencing your model, then you are in for a terrible surprise -- especially if you do not adjust for inflation. Inflation was as high as 18% and inflation-adjusted bond yields were low if not zero.

Historical data has a place in back testing and Monte Carlo simulations, but they should not be used for asset allocation or return estimates. Many have argued that past volatility can be used to predict future volatility. I have no other way for estimating volatility, so that is the only value of the three that I take from historical data.

For me, keeping things simple means making the fewest predictions about the future. And more importantly -- it also means being honest with myself and assessing my confidence about those predictions (estimates). So for example, future stock and bond returns would be a first-order estimate. Their correlation would be a second-order estimate as it combines the two. Determining your asset allocation would be third-order. What is your final confidence interval?
millennialmillions wrote: Thu Aug 12, 2021 7:24 pm
Rob Bertram wrote: Thu Aug 12, 2021 11:59 am From a modeling perspective, how do you handle market movements? I ask because this makes back testing and simulations complicated. For example, you have a 200% stock portfolio. The leveraged value is $200k ($100k yours + $100k margin).
My analysis is using a combination of 3x leveraged ETFs and non-leveraged funds to achieve the target leverage. The leveraged ETFs track daily 3x returns. I used annual data in the analysis, so rebalancing between funds according to the target leverage occurs annually.
The biggest problem of leveraged ETFs is that they refactor leverage daily. They literally buy when the market goes up and sells when the market drops. Buying high and selling low is a losing strategy. As I mentioned in the last post, volatility destroys these ETFs. The only thing that keeps these on people's radar is that we have had more up days than down. For clarity, Ayers and Nalebuff recommend staying the course when markets drop, so a leveraged ETF do not have a place in their model.

For the record, leveraged ETFs have all of the downside of 3x leverage but only a fraction of the upside. You don't have to go far back to see this in action. Look at 2020. Compare UPRO (3x S&P500) to VTI (Vanguard total-stock market) from the beginning of the year through December (Portfolio Visualizer). UPRO has a negative return while VTI is up over 15%. Yet UPRO had 3x the downside.

Lifecycle Investing as defined by Ayers and Nalebuff doesn't have this flaw. Instead of having a "fixed" leverage ratio like the ETFs, Lifecycle Investing introduces a leverage "floor." You buy more when the market goes up and compound those gains while staying the course when the market drops. Because of this, your annual return is at least your leverage factor (e.g., 140%) times the annual return of the underlying assets. If you are using leveraged ETFs to model their strategy, you are not doing yourself or the strategy justice.
millennialmillions wrote: Thu Aug 12, 2021 7:24 pm
Rob Bertram wrote: Thu Aug 12, 2021 11:59 am One last thing to consider: What is the goal of this portfolio? Are you aiming to have a huge pile of cash, or are you working to generate a stream of income? For most of us, income is our primary goal. If this is the same for you, I am going make an argument that your ending portfolio value is not a meaningful measurement. Instead, it should be the income stream that it generates.
My analysis only includes the accumulation phase. I wanted to limit the scope to decisions that are relevant to me now, so I'm assuming that the same strategy would be used in retirement across strategies. So money at the end of the accumulation phase is my target.
If you are modeling a Lifecycle Investing strategy and want to do it fairly, you need to track the portfolio's market position instead of its value. They recommend staying the course. Simulations that end on a down market would have a higher market position and higher leverage ratio. Because you expect to continue the strategy after the simulation, you would continue the "buy and hold" / stay the course approach.
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

Bentonkb wrote: Thu Aug 12, 2021 9:49 am
Exhibit 3 from the PIMCO paper makes 30Y treasuries look best too me. I'm looking for small, positive returns that are as stable as possible. That and volatility that is inversely correlated with stocks.
Why would you want less carry and roll-down? Perhaps you're reading the chart wrong? More carry and roll-down is better. That's your return. 5-yr and 2-yr bonds almost always have had more carry and roll-down per year of duration than long bonds.
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

Bentonkb wrote: Thu Aug 12, 2021 9:49 am
Exhibit 3 from the PIMCO paper makes 30Y treasuries look best too me. I'm looking for small, positive returns that are as stable as possible. That and volatility that is inversely correlated with stocks.
Why would you want less carry and roll-down? Perhaps you're reading the chart wrong? More carry and roll-down is better. That's your return. 5-yr and 2-yr bonds almost always have had more carry and roll-down per year of duration than long bonds.

Exhibit 3 is just a fancy way of saying that the yield curve is almost completely flat after 15 years and you're getting almost no return for all the extra risk.
It's not a graph of price or volatility or anything like that. It just shows that the shorter durations sometimes have a lot more carry than the long durations, and sometimes only a little more carry. That's like saying they're sometimes a lot better sometimes a little better, but pretty much always better.

And as the "Bet against Beta" paper basically proved this is due to leverage constrained investors chasing risk and return and leading to overpricing of riskier assets (long bonds and high beta stocks).
Bentonkb
Posts: 48
Joined: Sun Mar 07, 2021 6:47 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Bentonkb »

skierincolorado wrote: Fri Aug 13, 2021 3:32 pm

Exhibit 3 from the PIMCO paper makes 30Y treasuries look best too me. I'm looking for small, positive returns that are as stable as possible. That and volatility that is inversely correlated with stocks.
Why would you want less carry and roll-down? Perhaps you're reading the chart wrong? More carry and roll-down is better. That's your return. 5-yr and 2-yr bonds almost always have had more carry and roll-down per year of duration than long bonds.

[/quote]

Maybe I was reading it wrong. It looked like the 2 and 5 year bonds frequently had a negative carry and roll-down. Sometimes the carry on the short term bonds was significantly better and sometimes it was a little worse than the long term. On average, the short term contracts had a better carry yield. That is how I read it. Please correct me if I didn't understand.

I said that I prefer the long term contracts, not because they have a better carry, but because they are less likely to have a negative carry.

The bond portion of my portfolio is not expected to be the driver of returns. They are just supposed to provide some volatility that is negatively correlated with stocks and a small return above anticipated inflation. Gold is supposed to rise when there is unanticipated inflation.

Stocks are supposed to be the real driver of returns. Bonds are an insurance policy that pulls its own weight but not much more. Gold is an insurance policy that keeps up with inflation and only costs me 50 bps.

I'm open to other ideas and willing to consider shifting to short term bonds and more leverage, but I need to understand what the effects will be. My current position has an effective duration of 10 to 11 years, which might not be as far away from your ideal as you think.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

I ran the same strategies as above through bootstrap-sampled Monte Carlo simulations. The fund performance data includes 1955-2020. I assume each year is independent but that the returns across various funds within a year are dependent. E.g. a year in the sample may correspond to historical year 1980, in which case returns for all funds reflect returns from 1980. I ran 10,000 simulations of 35-year periods for each strategy.

Image

Unsurprisingly, there is more spread in the percentiles than the straight historical analysis. The only meaningful difference I see is that the 10th percentile is much closer across strategies.

Looking at both the historical period analysis and the bootstrap analysis, the constant 2x 70-30 portfolio seems like the winner to me. The 10th percentile is close to the best result, and the median result is significantly higher than anything except a constant HFEA.

The biggest thing I'm grappling with is Rob's point about bond performance being inherently different in the future. However, even looking at periods that begin in the mid-80's, the performance of the 2x 70-30 portfolio is significantly better than strategies that do not hold bolds while leveraged. I understand it's not reasonable to assume that high of inflation and bond yields going forward, but isn't it reasonable to assume the risk premium of equities over bonds will behave similarly to the past and the strategy that performed best across history is most likely to perform best going forward?
skierincolorado
Posts: 472
Joined: Sat Mar 21, 2020 10:56 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by skierincolorado »

Bentonkb wrote: Sat Aug 14, 2021 9:58 am
Maybe I was reading it wrong. It looked like the 2 and 5 year bonds frequently had a negative carry and roll-down. Sometimes the carry on the short term bonds was significantly better and sometimes it was a little worse than the long term. On average, the short term contracts had a better carry yield. That is how I read it. Please correct me if I didn't understand.

I said that I prefer the long term contracts, not because they have a better carry, but because they are less likely to have a negative carry.

The bond portion of my portfolio is not expected to be the driver of returns. They are just supposed to provide some volatility that is negatively correlated with stocks and a small return above anticipated inflation. Gold is supposed to rise when there is unanticipated inflation.

Stocks are supposed to be the real driver of returns. Bonds are an insurance policy that pulls its own weight but not much more. Gold is an insurance policy that keeps up with inflation and only costs me 50 bps.

I'm open to other ideas and willing to consider shifting to short term bonds and more leverage, but I need to understand what the effects will be. My current position has an effective duration of 10 to 11 years, which might not be as far away from your ideal as you think.
The 5 year had more carry than the 30 for all 23 years except a few months in 2007. It had more carry than the 10 for most of the 23 years as well. The 2 year's carry is volatile, but averages the highest and it's really the average that matters. The carry for a particular duration can get low when it is expected yields will fall at that particular duration. For example, in early 2008 the carry on the 2-yr was negative. But it was actually a phenomenal time to own the 2-yr. Bond investors were (correctly) predicting rates would fall (of course they probably fell more than expected). Owning 500k of 2-yr vs 100k of 10-yr, you would have made more money in 2008-2009, because the 2-yr rate fell by over 4.5% compared to 3.5% for the 10-yr. If you owned the 500k of 2-yr vs 100k of 10-y for the 23 years, you would have made a lot more money because the carry averaged higher. The carry is more volatile because bond investors have more clarity as to how short term rates are likely to change over the next 2-5 years. So the carry can get low when short-term rates are expected to fall a lot.

The exhibit you really care about is exhibit 13. They develope a market timing model to maximize returns which is way too complex for a normal person to do. But the key thing is that the model basically alternates between the 2-yr and 5-yr and occasionally the 10-yr. Ratheer than try to time the market, I'd rather own all three with equal risk assigned to each one. So 100k 10-yr 200k 5-yr and 500k 2-yr. That's what I intend to own long-term, but right now I just did the 5-yr to keep things simple. I might juts stick with the 5-yr because it's nearly as good as the 2-yr and the 2-yr leverage gets a little crazy even for me.

It's past the 10-yr where things really start to get ugly and the bet against beta becomes critical, so yeah I agree if your risk-weighted duration is 10-11 years that's pretty reasonable. Of course if that's just a straight average, then you're really taking the large majority of your risk farther out on the curve. For example 100k of 5 year and 100k of 20 year would be dominated by the 20-yr.
User avatar
PicassoSparks
Posts: 209
Joined: Tue Apr 28, 2020 5:41 am

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by PicassoSparks »

Rob Bertram wrote: Fri Aug 13, 2021 11:26 am For the record, leveraged ETFs have all of the downside of 3x leverage but only a fraction of the upside. You don't have to go far back to see this in action. Look at 2020. Compare UPRO (3x S&P500) to VTI (Vanguard total-stock market) from the beginning of the year through December (Portfolio Visualizer). UPRO has a negative return while VTI is up over 15%. Yet UPRO had 3x the downside.
This is a little misleading. In some circumstances — such as the one you constructed — UPRO has 3x the downside and less of the upside than a 3x leverage. In other circumstances, it has less of the downside and more upside. Compounding is weird.

The UPRO prospectus has a chart that is very helpful for understanding this. The dark grey sections are where UPRO underperforms 3x the index. But look at what's happening in the light grey areas.
Image
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

PicassoSparks wrote: Sat Aug 14, 2021 2:49 pm
This is a little misleading. In some circumstances — such as the one you constructed — UPRO has 3x the downside and less of the upside than a 3x leverage. In other circumstances, it has less of the downside and more upside. Compounding is weird.

The UPRO prospectus has a chart that is very helpful for understanding this. The dark grey sections are where UPRO underperforms 3x the index. But look at what's happening in the light grey areas.
Yep, I was going to post something similar - not really fair to call it 3x the downside without the upside. In practice, with the simulated data back to 1955, the 3x leveraged S&P 500 fund has outperformed 3 * the S&P 500 returns only 12 out of 66 years, underperforming the rest. This is not surprising given the added expenses of borrowing.
Rob Bertram
Posts: 857
Joined: Mon May 05, 2014 12:15 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by Rob Bertram »

millennialmillions wrote: Sat Aug 14, 2021 11:51 am I ran the same strategies as above through bootstrap-sampled Monte Carlo simulations. The fund performance data includes 1955-2020. I assume each year is independent but that the returns across various funds within a year are dependent. E.g. a year in the sample may correspond to historical year 1980, in which case returns for all funds reflect returns from 1980. I ran 10,000 simulations of 35-year periods for each strategy.
How can you do 35-year periods for a fund like UPRO that has only been around since June of 2009?

Instead of using a leveraged ETF, you could attempt to determine a better approximate return rate for a Lifecycle strategy. If you have historical stock performance, you would be better served by multiplying the return by your leverage ratio and giving it a 10% bump for years with positive return and no bump on down years. Subtract the risk-free rate, cash, or short-term treasury rates to represent borrowing costs. So for example, say you were simulating a 200% stock allocation -100% cash. If stocks returned 15% for that period and borrowing was 0.5%, the 200% leveraged fund would have returned 0.15 * 2.0 * 1.1 - 0.005 = 32.5%.

As I mentioned before, the leveraged aspect of Lifecycle Investing has all the upside of compounding incremental growth similar to a leveraged ETF but none of the volatility problems as you are to stay the course and weather a market downturn. This makes it slightly challenging to model correctly, but learning how to do so makes you more aware of the benefit of taking a leveraged position.
millennialmillions wrote: Sat Aug 14, 2021 11:51 am Looking at both the historical period analysis and the bootstrap analysis, the constant 2x 70-30 portfolio seems like the winner to me. The 10th percentile is close to the best result, and the median result is significantly higher than anything except a constant HFEA.

The biggest thing I'm grappling with is Rob's point about bond performance being inherently different in the future. However, even looking at periods that begin in the mid-80's, the performance of the 2x 70-30 portfolio is significantly better than strategies that do not hold bolds while leveraged. I understand it's not reasonable to assume that high of inflation and bond yields going forward, but isn't it reasonable to assume the risk premium of equities over bonds will behave similarly to the past and the strategy that performed best across history is most likely to perform best going forward?
You will be in for a nasty surprise if you validate your model using bond data from the 80s, 90s, or 2000s. Bond yields were falling from their highs (20% or more) to near zero today. When yields fall, the value of bonds in your portfolio goes up. This is extremely unlikely to happen again. Yields might go negative, but that is also unlikely. Yields would need to rise to 20% before market data from the 80s would be relevant again. My crystal ball is fuzzy on most things, but when I ask if bond yields will return to 20% in the next 35 years it consistently tells me "My sources say no."

Over the next 10 years, we expect bond yields to rise. This means that their value will drop. Their total return will be close to zero if not negative. This is why I said that you need to spend time to estimate their future return, variance, and correlation.
User avatar
Topic Author
millennialmillions
Posts: 107
Joined: Sun Aug 18, 2019 6:03 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by millennialmillions »

Rob Bertram wrote: Sat Aug 14, 2021 5:46 pm How can you do 35-year periods for a fund like UPRO that has only been around since June of 2009?
Siamond created simulated returns for a 3x ETF that matches up very well against UPRO. The simulated returns go back to 1955. See this thread for more detail.
Rob Bertram wrote: Sat Aug 14, 2021 5:46 pm As I mentioned before, the leveraged aspect of Lifecycle Investing has all the upside of compounding incremental growth similar to a leveraged ETF but none of the volatility problems as you are to stay the course and weather a market downturn. This makes it slightly challenging to model correctly, but learning how to do so makes you more aware of the benefit of taking a leveraged position.
I am interested in considering methods of leverage other than ETFs, such as futures or portfolio margin. I need to do more research on these to understand how to model and how I would implement in practice.
Rob Bertram wrote: Sat Aug 14, 2021 5:46 pm Over the next 10 years, we expect bond yields to rise. This means that their value will drop. Their total return will be close to zero if not negative. This is why I said that you need to spend time to estimate their future return, variance, and correlation.
I don't share the strong conviction that bond yields will rise over the next 10 years or that the current low rate environment negates a leveraged strategy with stocks and bonds. This topic came up several times across the HFEA threads, and I think the same principles apply here. See this post for returns from 1955-1964, when 10-year yields were still no more than 4.2% and hadn't begun their steep climb of the late 1960s and 1970s.
User avatar
firebirdparts
Posts: 2790
Joined: Thu Jun 13, 2019 4:21 pm

Re: Lifecycle Investing vs. Hedgefundie's Excellent Adventure

Post by firebirdparts »

bumbojumbo wrote: Wed Aug 11, 2021 1:10 am Does anyone know why long-term treasuries (LTT) have had lower risk-adjusted returns than STT or ITT?
Returns are similar, but variation in price due to interest rate expectations are not similar.

It seems to me that is misleading and unhelpful. The reason LTT's figure in this sort of strategizing is that they have higher absolute returns, high volatility, and lack of correlation to stocks. The ideal portfolio would have something in with returns as high as stocks but short-term negative correlation. No such investment exists, but if it did, volatility would actually be useful. In these portfolios of the past, the volatility of long term treasuries has not been a problem, it's been a feature.

I think mistaking the volatility for actual risk is a very bad mistake, commonly made in these forums. If you are really desperate to do some mathematical comparison, it's generally accepted that volatility of stocks could be used a proxy for risk, which is not easily measureable. But obviously, a model of the risk of stocks is not really applicable to treasuries. Treasury pricing is based entirely on some simple fundamental concepts.
The risk of a US treasury bankruptcy is not involved in setting the prices. Using volatility as a proxy for risk is not some fundamental truth; it's a weak attempt at building a meaningful model.

To me it just seemed like total nonsense. This is just my opinion which these days would need to be combined with over $2 to buy you a cup of coffee.
A fool and your money are soon partners
Post Reply