Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

adamhg wrote: Fri Sep 10, 2021 10:09 pm

Yes, thanks as usual for the detailed reply. The margin requirement makes sense so that box spreads probably aren't even feasible in iras.

Personally, I'm optimizing for convenience and plan to automate this whole thing. If I get there, I'll pen source the code. The challenge I'm finding with futures is that there aren't a lot of brokers that support apis to trade futures and equity. IBKR is the obvious choice but I refuse to work with their tech.

Your point about the rate may simply be due to the duration of the loan though. A box spread would be close to the 2-3 yr institutional rate whereas the future at the 1-2mo rate. So while your rate is lower, the box locks it in for longer and you should theoretically always be able to sell it back at the same or similar rate as the future if you wanted to roll monthly even.

But I agree, pin pointing the best rate with a 4 legged option bid/ask isn't going to be as easy as just buying a future. In all honestly, my biggest fear with futures though is that I won't be able to resist going 12x STT
Or 50x Eurodollars!

You're right about the box-spread duration. I always remember that when I write one and then promptly forget it. The last one I wrote wasn't 2 full years so .7% was still on the high side. I don't use them a lot so not terribly concerned.

Automating all of this would be quite nice. Take out the risk of emotional error.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by comeinvest »

skierincolorado wrote: Thu Sep 09, 2021 2:35 pm The only scenario where LTT wins are if we assume 1) ITT rates will rise 2) LTT rates don't rise at all and 3) the rate increase of ITT is so large it overcomes the 0.7% return from roll on ITT leaving us with only 0.8% of carry 4) we ignore points #1 and #4 and assume the future will be different than the past and that we can predict when and how it will be different.
This scenario describes a yield curve flattening. I'm wondering if it's really so far-fetched, if we assume that the U.S. follows Europe and Japan i.e. most of the rest of the developed world with a certain lag, and that real returns, and along with them term premia, will permanently converge to zero or negative. I know this scenario is heavily debated, but it's a scenario that many economists predict for a variety of reasons.

Also, in this scenario, your 0.7% roll return may not last long, if it materializes at all. The "roll return", by definition, is based on an assumption of a static yield curve, which is purely hypothetical and does not even take into account current forward rates.

In summary, even after reading this thread and your arguments for STT/ITT vs. LTT, I'm wondering if a "diversification of treasury maturities" that for example NTSX implements is not beneficial, as different maturities thrive in different future interest rate scenarios.

To be clear, I'm not arguing against your rationale to shift to STT and ITT. In fact, you made me re-consider and change my own allocation. I'm just saying, maybe some of the math proposed in this thread is the result of a bit of over-engineering based on history, assumptions, and models that may or may not reflect the future, while the aspect of diversification might have got a bit lost.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

comeinvest wrote: Sun Sep 12, 2021 12:55 am
skierincolorado wrote: Thu Sep 09, 2021 2:35 pm The only scenario where LTT wins are if we assume 1) ITT rates will rise 2) LTT rates don't rise at all and 3) the rate increase of ITT is so large it overcomes the 0.7% return from roll on ITT leaving us with only 0.8% of carry 4) we ignore points #1 and #4 and assume the future will be different than the past and that we can predict when and how it will be different.
This scenario describes a yield curve flattening. I'm wondering if it's really so far-fetched, if we assume that the U.S. follows Europe and Japan i.e. most of the rest of the developed world with a certain lag, and that real returns, and along with them term premia, will permanently converge to zero or negative. I know this scenario is heavily debated, but it's a scenario that many economists predict for a variety of reasons.

Also, in this scenario, your 0.7% roll return may not last long, if it materializes at all. The "roll return", by definition, is based on an assumption of a static yield curve, which is purely hypothetical and does not even take into account current forward rates.

In summary, even after reading this thread and your arguments for STT/ITT vs. LTT, I'm wondering if a "diversification of treasury maturities" that for example NTSX implements is not beneficial, as different maturities thrive in different future interest rate scenarios.

To be clear, I'm not arguing against your rationale to shift to STT and ITT. In fact, you made me re-consider and change my own allocation. I'm just saying, maybe some of the math proposed in this thread is the result of a bit of over-engineering based on history, assumptions, and models that may or may not reflect the future, while the aspect of diversification might have got a bit lost.
I agree it's not terribly far-fetched, but it shouldn't be our starting point of assumptions which would be market timing. I don't think the math is over-engineered - I'm not sure how far along you've read but eventually we do calculate expected returns factoring in forward rates. The results weren't *quite* as rosy for ITT as I expected, because rate increases for ITT were a little higher than I expected, but the expected return of LTT was very poor as well. We calculated 0.1% expected return for both ITT and LTT. But given the risk on LTT is higher, .1% for ITT is better than .1% for LTT.

This .1% is not the term premium, it's the expected return. Despite the slightly negative term premium, expected returns are probably slightly positive because future short term rates are expected to be higher than current short term rates. The expected return captures both aspects of return - the path of future short term rates (positive), and the term premium (slightly negative).

This fits an expectation of term-premia declining to zero. If the term-premia is zero for all durations, that's a very strong argument for owning shorter durations. Owning shorter durations captures the return of the other component of bond returns - the expected evolution of short-term interest rates. Since the term-premia is zero, why would we bother to try capture it by owning longer durations? That would be taking duration risk without compensation, by definition.

My understanding is even as rates have declined across the rest of the developed world, risk adjusted returns on shorter durations have still generally been higher than for longer durations.

So while term premia declining to zero seems like an argument for shorter durations to me, it also seems like an argument for a lower bond allocation in general - which is something else I've been suggesting in this thread. While the historical efficient frontier since 1978 is like 67% ITT (or like 70% STT + 10% LTT) I think a somewhat lower allocation is prudent going forward given lower expected returns (due to the apparent falling of the term premia). This would be especially true when current financing costs exceed future expected short-term interest rates. Overall, I agree some diversification is prudent and the historical efficient frontier backs this up as the sharpe ratio for 70% STT + 10% LTT is a hair higher than for ITT alone. Presumably this is the diversification benefit showing up in the efficient frontier.

For example, this paper suggests that a low term-premium is an argument for owning shorter duration bonds: https://www.nb.com/documents/public/glo ... undrum.pdf

It does appear the term-premium has fallen, but I haven't read anything yet that explains why this is the case. You've mentioned a number of economists forecast this to be a permanent phenomenon, can you point me to some of those papers?
Last edited by skierincolorado on Sun Sep 12, 2021 2:55 am, edited 1 time in total.
seajay
Posts: 1656
Joined: Sat May 01, 2021 3:26 pm
Contact:

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by seajay »

LTCM wrote: Sat Sep 04, 2021 3:25 am Back test of SPY/8xSTT since 1991:

Image

Back test of SPY/3xITT since 1991:

Image

Back test of SPY/LTT since 1991:

Image

Adjust volatility to taste.

8/9 portfolios do worst in 1994 when markets were flat and the short term borrowing cost was at 4%.
Assumption of CASHX as the cost of borrowing is misplaced IMO. Margin/leverage costs tend to be higher, more like 50/50 VBMFX/SHY

Factor that in and your reference adjusts to this i.e. the apparent benefit is lost.

Image

More strictly, historically LIBOR+x% was the common cost to leverage, and at times LIBOR spiked considerably around periods of high volatility.

The benefits since the financial crisis (2009) are somewhat unique/infrequent due to high levels of yield curve manipulation.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

seajay wrote: Sun Sep 12, 2021 2:52 am
Assumption of CASHX as the cost of borrowing is misplaced IMO. Margin/leverage costs tend to be higher, more like
Factor that in and your reference adjusts to
i.e. the apparent benefit is lost.

Image

More strictly, historically LIBOR+x% was the common cost to leverage, and at times LIBOR spiked considerably around periods of high volatility.

The benefits since the financial crisis (2009) are somewhat unique/infrequent due to high levels of yield curve manipulation.
As discussed in this thread, the historical cost of borrowing with Treasury futures has been LIBOR. And the cost of borrowing implicit in S&P500 futures has been low enough that owning futures + CASH (unleveraged) has been equally as profitable as owning an S&P index fund, after fees. For example holding 100k of SPY has had the same return as holding 100k of CASH and futures with exposure to 100k of S&P500. Again, this rate is very close to LIBOR.

To be clear, you've used a combination of 1-3 year treasury bills and total bond market as the borrowing rate, which is much much higher than the actual financing cost of Treasury and S&P500 futures. There are several papers in this thread proving this to be the case.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by comeinvest »

skierincolorado wrote: Sun Sep 12, 2021 2:25 am
comeinvest wrote: Sun Sep 12, 2021 12:55 am
skierincolorado wrote: Thu Sep 09, 2021 2:35 pm The only scenario where LTT wins are if we assume 1) ITT rates will rise 2) LTT rates don't rise at all and 3) the rate increase of ITT is so large it overcomes the 0.7% return from roll on ITT leaving us with only 0.8% of carry 4) we ignore points #1 and #4 and assume the future will be different than the past and that we can predict when and how it will be different.
This scenario describes a yield curve flattening. I'm wondering if it's really so far-fetched, if we assume that the U.S. follows Europe and Japan i.e. most of the rest of the developed world with a certain lag, and that real returns, and along with them term premia, will permanently converge to zero or negative. I know this scenario is heavily debated, but it's a scenario that many economists predict for a variety of reasons.

Also, in this scenario, your 0.7% roll return may not last long, if it materializes at all. The "roll return", by definition, is based on an assumption of a static yield curve, which is purely hypothetical and does not even take into account current forward rates.

In summary, even after reading this thread and your arguments for STT/ITT vs. LTT, I'm wondering if a "diversification of treasury maturities" that for example NTSX implements is not beneficial, as different maturities thrive in different future interest rate scenarios.

To be clear, I'm not arguing against your rationale to shift to STT and ITT. In fact, you made me re-consider and change my own allocation. I'm just saying, maybe some of the math proposed in this thread is the result of a bit of over-engineering based on history, assumptions, and models that may or may not reflect the future, while the aspect of diversification might have got a bit lost.
I agree it's not terribly far-fetched, but it shouldn't be our starting point of assumptions which would be market timing. I don't think the math is over-engineered - I'm not sure how far along you've read but eventually we do calculate expected returns factoring in forward rates. The results weren't *quite* as rosy for ITT as I expected, because rate increases for ITT were a little higher than I expected, but the expected return of LTT was very poor as well. We calculated 0.1% expected return for both ITT and LTT. But given the risk on LTT is higher, .1% for ITT is better than .1% for LTT.

This .1% is not the term premium, it's the expected return. Despite the slightly negative term premium, expected returns are probably slightly positive because future short term rates are expected to be higher than current short term rates. The expected return captures both aspects of return - the path of future short term rates (positive), and the term premium (slightly negative).

This fits an expectation of term-premia declining to zero. If the term-premia is zero for all durations, that's a very strong argument for owning shorter durations. Owning shorter durations captures the return of the other component of bond returns - the expected evolution of short-term interest rates. Since the term-premia is zero, why would we bother to try capture it by owning longer durations? That would be taking duration risk without compensation, by definition.

My understanding is even as rates have declined across the rest of the developed world, risk adjusted returns on shorter durations have still generally been higher than for longer durations.

So while term premia declining to zero seems like an argument for shorter durations to me, it also seems like an argument for a lower bond allocation in general - which is something else I've been suggesting in this thread. While the historical efficient frontier since 1978 is like 67% ITT (or like 70% STT + 10% LTT) I think a somewhat lower allocation is prudent going forward given lower expected returns (due to the apparent falling of the term premia). This would be especially true when current financing costs exceed future expected short-term interest rates. Overall, I agree some diversification is prudent and the historical efficient frontier backs this up as the sharpe ratio for 70% STT + 10% LTT is a hair higher than for ITT alone. Presumably this is the diversification benefit showing up in the efficient frontier.

For example, this paper suggests that a low term-premium is an argument for owning shorter duration bonds: https://www.nb.com/documents/public/glo ... undrum.pdf

It does appear the term-premium has fallen, but I haven't read anything yet that explains why this is the case. You've mentioned a number of economists forecast this to be a permanent phenomenon, can you point me to some of those papers?
"I'm not sure how far along you've read" - I'm admittedly in the process of catching up with this and related threads. Therefore my apologies if I'm asking something that would be evident in the math done already.

"We calculated 0.1% expected return for both ITT and LTT" - In your calculation of the expected future ITT and LTT interest rates for purpose of calculating the expected returns of ITT and LTT, did you take into account the fact that realized rates are usually lower than forward rates (forward rates for example based on STIR futures), because forward rates basically reflect the term (risk) premium, while realized rates don't? In other words, the forward rates (and by implication future ITT and LTT rates implied by forward rates) are NOT the expected rates that people actually expect to prevail on a given future date. That's my understanding. Correct me if I'm wrong. If we accept that forward rates are not expected rates (i.e. the expectations hypothesis is not true, and I think it is not considered true), then by implication, I'm questioning the approach of calculating expected returns from forward rates.

"My understanding is even as rates have declined across the rest of the developed world, risk adjusted returns on shorter durations have still generally been higher than for longer durations" - I think not true for Europe for a while. I've been shorting the 2-year Bund futures for some time as an insurance against rising interest rates worldwide, and at the same time I have made some money from this "insurance" short position. Whatever the risk of the 2-year Bund is (probably more an opinion as to how far negative real rates can go, than a quantifiable risk), the carry is negative, while the longer maturities have positive carry.

Another question: If indeed the expected return of ITT and LTT is 0.1%, then it is currently about equal to the funding rate, and will be lower than the funding rate soon, i.e. the expected return of the futures will be negative soon. I think someone showed a long time ago in the HFEA thread that rebalancing with an asset with zero or negative expected returns does not make sense, even when correlations are negative, i.e. you are better off just investing in the asset with positive expected returns and "rebalancing with cash". Can you elaborate if that argument might be applicable to the situation of zero or negative expected returns from the treasury futures, i.e. shouldn't the treasury futures allocation be zero under your assumption of zero expected return for treasury futures?

Whether the term premium might possibly be permanently gone: I don't have a citation handy right now, but I think the argument is similar to that of lower expected equity returns, higher valuations across almost all asset classes, etc.: A glut of savings worldwide on one hand, and reduced need from businesses for capital on the other hand, compared to the era of industrialization. I've also seen the argument that the world is perceived "safer" now than in the past, which would lower the discount rates that investors demand to "store" capital.
Not from an explanatory, but purely from an observational point of view, let's also not forget that the rest of the developed world has been in that mode for like 10+ years, with no end in sight. I have a hard time ignoring what might seem like, and could be, a race to zero or negative real and nominal returns. The last 10+ years have seen economically good and bad times, but nothing reversed that trend. I'm wondering what could in fact reverse this trend. I know that "other experts" argue for "mean reversion", and in business schools they teach theories that I don't know well enough about what an "equilibrium" interest rate should be (theories which didn't play out in the developed world for 10+ years), but whatever the future brings, I think we can agree that there is no guarantee or "entitlement", or law of physics, that says investors can increase their NAV on average over time just by "sitting it out", rather than consuming their assets now.
User avatar
Bentonkb
Posts: 83
Joined: Sun Mar 07, 2021 5:47 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by Bentonkb »

I would like to point out that Skier's suggestion that lower allocations to bonds vs stocks and comeinvest's idea of rebalancing into cash rather than bonds are effectively the same thing. For a futures trader they mean smaller position sizes and lower leverage.

Probably good advice, especially for me. I seem to be more levered up than the rest of you.
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

seajay wrote: Sun Sep 12, 2021 2:52 am Factor that in and your reference adjusts to this i.e. the apparent benefit is lost.
This has borrowing at 5.5% in 2020?
55% VUG - 20% VEA - 20% EDV - 5% BNDX
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

comeinvest wrote: Sun Sep 12, 2021 4:30 am In your calculation of the expected future ITT and LTT interest rates for purpose of calculating the expected returns of ITT and LTT, did you take into account the fact that realized rates are usually lower than forward rates (forward rates for example based on STIR futures), because forward rates basically reflect the term (risk) premium, while realized rates don't? In other words, the forward rates (and by implication future ITT and LTT rates implied by forward rates) are NOT the expected rates that people actually expect to prevail on a given future date. That's my understanding. Correct me if I'm wrong. If we accept that forward rates are not expected rates (i.e. the expectations hypothesis is not true, and I think it is not considered true), then by implication, I'm questioning the approach of calculating expected returns from forward rates.
We did find a discrepancy between the forward rates and the current rates. Basically when we used forward rates to predict the current rates it was overestimating by a small factor. Skier then calculated the change between the forward calculated current rate and the forward calculated forward rate and applied that change to the real current rate.

Personally I’m still a bit unsure of the extent to which predicted rate rises are already priced in. Having them not priced in already seems like an arb opportunity. I’m just not sure how this affects the yield/roll/carry and expected return we’re trying to estimate.
55% VUG - 20% VEA - 20% EDV - 5% BNDX
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

comeinvest wrote: Sun Sep 12, 2021 4:30 am
"I'm not sure how far along you've read" - I'm admittedly in the process of catching up with this and related threads. Therefore my apologies if I'm asking something that would be evident in the math done already.

"We calculated 0.1% expected return for both ITT and LTT" - In your calculation of the expected future ITT and LTT interest rates for purpose of calculating the expected returns of ITT and LTT, did you take into account the fact that realized rates are usually lower than forward rates (forward rates for example based on STIR futures), because forward rates basically reflect the term (risk) premium, while realized rates don't? In other words, the forward rates (and by implication future ITT and LTT rates implied by forward rates) are NOT the expected rates that people actually expect to prevail on a given future date. That's my understanding. Correct me if I'm wrong. If we accept that forward rates are not expected rates (i.e. the expectations hypothesis is not true, and I think it is not considered true), then by implication, I'm questioning the approach of calculating expected returns from forward rates.

"My understanding is even as rates have declined across the rest of the developed world, risk adjusted returns on shorter durations have still generally been higher than for longer durations" - I think not true for Europe for a while. I've been shorting the 2-year Bund futures for some time as an insurance against rising interest rates worldwide, and at the same time I have made some money from this "insurance" short position. Whatever the risk of the 2-year Bund is (probably more an opinion as to how far negative real rates can go, than a quantifiable risk), the carry is negative, while the longer maturities have positive carry.

Another question: If indeed the expected return of ITT and LTT is 0.1%, then it is currently about equal to the funding rate, and will be lower than the funding rate soon, i.e. the expected return of the futures will be negative soon. I think someone showed a long time ago in the HFEA thread that rebalancing with an asset with zero or negative expected returns does not make sense, even when correlations are negative, i.e. you are better off just investing in the asset with positive expected returns and "rebalancing with cash". Can you elaborate if that argument might be applicable to the situation of zero or negative expected returns from the treasury futures, i.e. shouldn't the treasury futures allocation be zero under your assumption of zero expected return for treasury futures?

Whether the term premium might possibly be permanently gone: I don't have a citation handy right now, but I think the argument is similar to that of lower expected equity returns, higher valuations across almost all asset classes, etc.: A glut of savings worldwide on one hand, and reduced need from businesses for capital on the other hand, compared to the era of industrialization. I've also seen the argument that the world is perceived "safer" now than in the past, which would lower the discount rates that investors demand to "store" capital.
Not from an explanatory, but purely from an observational point of view, let's also not forget that the rest of the developed world has been in that mode for like 10+ years, with no end in sight. I have a hard time ignoring what might seem like, and could be, a race to zero or negative real and nominal returns. The last 10+ years have seen economically good and bad times, but nothing reversed that trend. I'm wondering what could in fact reverse this trend. I know that "other experts" argue for "mean reversion", and in business schools they teach theories that I don't know well enough about what an "equilibrium" interest rate should be (theories which didn't play out in the developed world for 10+ years), but whatever the future brings, I think we can agree that there is no guarantee or "entitlement", or law of physics, that says investors can increase their NAV on average over time just by "sitting it out", rather than consuming their assets now.
Regarding the .1% calculation, I believe what you are asking is mostly accounted for. Realized instantaneous forward rates should be considerably lower because they won't include a term premium. But realized 5-yr rates 1-yr from now, should be very close because it still contains most of the term premia. So while the IF 6 years from now would be substantially less (assuming a positive term premia) because it's lost 6 years of term premia, the 5-yr rate 1-yr from now only loses 1-yr of term premia. We were making predictions of rates, based off forward rates, for interest one year from now, so there should be very little loss of term premia. Also the term premia is near zero or negative.

Regarding the inclusion of an asset with a .1% or 0% ER for it's negative correlation purposes, I don't find the same as what you've seen in HFEA thread. The efficient fontier still includes some bonds even with a 0 ER, albeit greatly reduced. I'm not totally convinced that the ER is really as low as .1% for ITT and LTT, the market made this same mistake in 2009 in expecting rates to rise too quickly. So while the market seems to be telling us an ER near .1%, I think history tells us owning bonds has made sense. And I don't think economic theory really supports 0% returns for bond either. It's a very large asset class and it makes sense to at least own some. So we could definitely trim back bonds compared to the historical efficient frontier, but I'm no where close to wanting to eliminate them yet.
Northern Flicker
Posts: 15367
Joined: Fri Apr 10, 2015 12:29 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by Northern Flicker »

What happens if interest rise sharply causing a sharp fall in both stocks and bonds, not to mention an increase in cost of rolling leverage?
adamhg
Posts: 218
Joined: Sat Apr 10, 2021 8:40 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by adamhg »

Northern Flicker wrote: Sun Sep 12, 2021 1:19 pm What happens if interest rise sharply causing a sharp fall in both stocks and bonds, not to mention an increase in cost of rolling leverage?
When was the last time that happened?

https://en.m.wikipedia.org/wiki/History ... ee_actions

All rate increases since 2000 have been accompanied by bull markets, but I haven't checked previous ones before that list
Northern Flicker
Posts: 15367
Joined: Fri Apr 10, 2015 12:29 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by Northern Flicker »

adamhg wrote: Sun Sep 12, 2021 2:54 pm
Northern Flicker wrote: Sun Sep 12, 2021 1:19 pm What happens if interest rise sharply causing a sharp fall in both stocks and bonds, not to mention an increase in cost of rolling leverage?
When was the last time that happened?

https://en.m.wikipedia.org/wiki/History ... ee_actions

All rate increases since 2000 have been accompanied by bull markets, but I haven't checked previous ones before that list
2018, 1969, 1941, and 1931 were years in which both US large cap stocks and the 10-yr note had negative returns for the calendar year.

But when was the last time the Fed rate was effectively zero, the Fed balance sheet was filled with treasury notes and securitized mortgages?

In modern portfolio theory, you would find the optimal portfolio, and then leverage the whole portfolio to increase return, or add cash to decrease risk.
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

Northern Flicker wrote: Sun Sep 12, 2021 1:19 pm What happens if interest rise sharply causing a sharp fall in both stocks and bonds, not to mention an increase in cost of rolling leverage?
2018, 1969, 1941, and 1931 were years in which both US large cap stocks and the 10-yr note had negative returns for the calendar year.
2018 is -2% to -12% depending on leverage and duration choice on the bonds.

1994 is worse because of higher borrowing costs. Up to -47% on a flat S&P.

You can't win 'em all!
55% VUG - 20% VEA - 20% EDV - 5% BNDX
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

Northern Flicker wrote: Sun Sep 12, 2021 1:19 pm What happens if interest rise sharply causing a sharp fall in both stocks and bonds, not to mention an increase in cost of rolling leverage?
You'd lose money. If you maintained a 100% stock allocation, and added bonds on top, you'd do worse than a 100% stock allocation anytime bonds have a negative return. Of course, most of the time bonds have a positive return greater than the cost of leverage and on average, if the future is like the past, you will do much better than 100% stock. Bonds had very negative returns in 1994, and yet the strategy more than pays off over the long-run.
adamhg
Posts: 218
Joined: Sat Apr 10, 2021 8:40 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by adamhg »

Northern Flicker wrote: Sun Sep 12, 2021 3:18 pm
adamhg wrote: Sun Sep 12, 2021 2:54 pm
Northern Flicker wrote: Sun Sep 12, 2021 1:19 pm What happens if interest rise sharply causing a sharp fall in both stocks and bonds, not to mention an increase in cost of rolling leverage?
When was the last time that happened?

https://en.m.wikipedia.org/wiki/History ... ee_actions

All rate increases since 2000 have been accompanied by bull markets, but I haven't checked previous ones before that list
2018, 1969, 1941, and 1931 were years in which both US large cap stocks and the 10-yr note had negative returns for the calendar year.

But when was the last time the Fed rate was effectively zero, the Fed balance sheet was filled with treasury notes and securitized mortgages?

In modern portfolio theory, you would find the optimal portfolio, and then leverage the whole portfolio to increase return, or add cash to decrease risk.
Not disagreeing that it could happen, but the question was specifically if rising rates would trigger such an event. I'm not sure pre-1970s conditions are possible anymore, and 2018 was 3 years into rate hikes where hfea still outperformed over the full time period.

https://www.portfoliovisualizer.com/bac ... tion2_1=45

I'd argue the downturn that year was initially triggered by a volatility liquidity event rather than the rake hike
Northern Flicker
Posts: 15367
Joined: Fri Apr 10, 2015 12:29 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by Northern Flicker »

I don't think there is a precedent positive or negative for the current state of affairs as the initial condition.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

Northern Flicker wrote: Sun Sep 12, 2021 10:08 pm I don't think there is a precedent positive or negative for the current state of affairs as the initial condition.
That's literally always true. There many good arguments for continuing deflationary forces. The idea that inflation is likely is mistaken. It's possible. Even if there is inflation, hfea would likelyoutperform in the long-run. And the modified-hfea discussed here would likely do even better and be less inflation susceptible than hfea.
Northern Flicker
Posts: 15367
Joined: Fri Apr 10, 2015 12:29 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by Northern Flicker »

skierincolorado wrote: Sun Sep 12, 2021 11:33 pm
Northern Flicker wrote: Sun Sep 12, 2021 10:08 pm I don't think there is a precedent positive or negative for the current state of affairs as the initial condition.
That's literally always true. There many good arguments for continuing deflationary forces. The idea that inflation is likely is mistaken. It's possible. Even if there is inflation, hfea would likelyoutperform in the long-run. And the modified-hfea discussed here would likely do even better and be less inflation susceptible than hfea.
Good luck with it.
seajay
Posts: 1656
Joined: Sat May 01, 2021 3:26 pm
Contact:

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by seajay »

skierincolorado wrote: Sun Sep 12, 2021 2:58 am
seajay wrote: Sun Sep 12, 2021 2:52 am
Assumption of CASHX as the cost of borrowing is misplaced IMO. Margin/leverage costs tend to be higher, more like
Factor that in and your reference adjusts to
i.e. the apparent benefit is lost.

Image

More strictly, historically LIBOR+x% was the common cost to leverage, and at times LIBOR spiked considerably around periods of high volatility.

The benefits since the financial crisis (2009) are somewhat unique/infrequent due to high levels of yield curve manipulation.
As discussed in this thread, the historical cost of borrowing with Treasury futures has been LIBOR. And the cost of borrowing implicit in S&P500 futures has been low enough that owning futures + CASH (unleveraged) has been equally as profitable as owning an S&P index fund, after fees. For example holding 100k of SPY has had the same return as holding 100k of CASH and futures with exposure to 100k of S&P500. Again, this rate is very close to LIBOR.

To be clear, you've used a combination of 1-3 year treasury bills and total bond market as the borrowing rate, which is much much higher than the actual financing cost of Treasury and S&P500 futures. There are several papers in this thread proving this to be the case.
Thanks. In the UK spreadbets levy something like LIBOR+2% for margin, Futures a lot less expensive. LIBOR can however spike at times, typically during periods of stress. TED spread (3 month LIBOR / 3 month T-Bill) for instance spiked to 4.5% during 2008 https://www.macrotrends.net/1447/ted-sp ... ical-chart
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

seajay wrote: Tue Sep 14, 2021 4:04 am
skierincolorado wrote: Sun Sep 12, 2021 2:58 am
seajay wrote: Sun Sep 12, 2021 2:52 am
Assumption of CASHX as the cost of borrowing is misplaced IMO. Margin/leverage costs tend to be higher, more like
Factor that in and your reference adjusts to
i.e. the apparent benefit is lost.

Image

More strictly, historically LIBOR+x% was the common cost to leverage, and at times LIBOR spiked considerably around periods of high volatility.

The benefits since the financial crisis (2009) are somewhat unique/infrequent due to high levels of yield curve manipulation.
As discussed in this thread, the historical cost of borrowing with Treasury futures has been LIBOR. And the cost of borrowing implicit in S&P500 futures has been low enough that owning futures + CASH (unleveraged) has been equally as profitable as owning an S&P index fund, after fees. For example holding 100k of SPY has had the same return as holding 100k of CASH and futures with exposure to 100k of S&P500. Again, this rate is very close to LIBOR.

To be clear, you've used a combination of 1-3 year treasury bills and total bond market as the borrowing rate, which is much much higher than the actual financing cost of Treasury and S&P500 futures. There are several papers in this thread proving this to be the case.
Thanks. In the UK spreadbets levy something like LIBOR+2% for margin, Futures a lot less expensive. LIBOR can however spike at times, typically during periods of stress. TED spread (3 month LIBOR / 3 month T-Bill) for instance spiked to 4.5% during 2008 https://www.macrotrends.net/1447/ted-sp ... ical-chart
In 2008 the IRR (the financing cost) of treasury futures got as low as -4%. If I'm reading correctly, the basis traders were paying you to invest in treasury futures. In 2020, the opposite happened, and the IRR was about .3% above the 2-month T-Bill rate. In either case, the deviation from the 3-month T-Bill was short-lived. Some brief deviation (high or low) isn't going to cost very much in order to maintain a position.
DMoogle
Posts: 549
Joined: Sat Oct 31, 2020 10:24 pm

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by DMoogle »

Putting money where my mouth is. Bought a bunch of VT and ZN today - taxable portfolio now consists of 1.5x in equities and 3x in ITTs. Using IBKR margin at the moment, but have an open order for a box spread to get cheaper financing. Once I can get the single box to fill, I'll order more until my cash balance is no longer negative (or close).
klaus14
Posts: 953
Joined: Sun Nov 25, 2018 6:43 pm

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by klaus14 »

DMoogle wrote: Tue Sep 14, 2021 1:00 pm Putting money where my mouth is. Bought a bunch of VT and ZN today - taxable portfolio now consists of 1.5x in equities and 3x in ITTs. Using IBKR margin at the moment, but have an open order for a box spread to get cheaper financing. Once I can get the single box to fill, I'll order more until my cash balance is no longer negative (or close).
why not:
VXUS + /MES instead of VT ?

(even better: VXUS + VIOV + /MES)

this would be simpler than margin or box spread and would get better financing cost.
My investment algorithm: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=351899&p=6112869#p6112869
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

klaus14 wrote: Tue Sep 14, 2021 1:26 pm
DMoogle wrote: Tue Sep 14, 2021 1:00 pm Putting money where my mouth is. Bought a bunch of VT and ZN today - taxable portfolio now consists of 1.5x in equities and 3x in ITTs. Using IBKR margin at the moment, but have an open order for a box spread to get cheaper financing. Once I can get the single box to fill, I'll order more until my cash balance is no longer negative (or close).
why not:
VXUS + /MES instead of VT ?

(even better: VXUS + VIOV + /MES)

this would be simpler than margin or box spread and would get better financing cost.
Let's say we buy 100k of SPY with a box vs 100k with MES:

Let's say the MES grows 10% per year, but is taxed at ~20% on those gains. So that's the same as growing with a 8% annual growth rate. 100k*1.08^30 = 1006k.

The SPY is taxed at ~15% when it's sold in 30 years. (100k*1.1^30 -100k)*.85 + 100k = 1,498k

These are rough assumptions, and I've not included dividends, so the SPY will be taxed more than I've calculated. But I think the general idea is correct.

This doesn't include the interest deduction on the box, which would add to the SPY advantage. It assumes the financing cost is the same, which isn't correct, and would help out MES.
Last edited by skierincolorado on Tue Sep 14, 2021 1:56 pm, edited 2 times in total.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

skierincolorado wrote: Tue Sep 14, 2021 1:39 pm
klaus14 wrote: Tue Sep 14, 2021 1:26 pm
DMoogle wrote: Tue Sep 14, 2021 1:00 pm Putting money where my mouth is. Bought a bunch of VT and ZN today - taxable portfolio now consists of 1.5x in equities and 3x in ITTs. Using IBKR margin at the moment, but have an open order for a box spread to get cheaper financing. Once I can get the single box to fill, I'll order more until my cash balance is no longer negative (or close).
why not:
VXUS + /MES instead of VT ?

(even better: VXUS + VIOV + /MES)

this would be simpler than margin or box spread and would get better financing cost.
Let's say we buy 100k of SPY with a box vs 100k with MES:

Let's say the MES grows 10% per year, but is taxed at ~20% on those gains. So that's the same as growing with a 8% annual growth rate. 100k*1.08^30 = 1006k.

The SPY is taxed at ~15% when it's sold in 30 years. (100k*1.1^30 -100k)*.85 + 100k = 1,498k

These are rough assumptions, and I've not included dividends, so the SPY will be taxed more than I've calculated. . But I think the general idea is correct.


This doesn't include the interest deduction on the box, which would add to the SPY advantage. It assumes the financing cost is the same, which isn't correct, and would help out MES.

SPY with dividiends:

8% annual growth + 2% annual dividends. Dividends taxed at 25%. Thus 9.5% annual growth rate. Guestimating 50k of dividends received over 30 years, thus 50k of basis.

(100k*1.095^30-150k)*.85+150k = 1316k

Again, this doesn't include the interest deduction on the box, which would add to SPY's advantage.
Last edited by skierincolorado on Tue Sep 14, 2021 1:57 pm, edited 2 times in total.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

skierincolorado wrote: Tue Sep 14, 2021 1:45 pm
skierincolorado wrote: Tue Sep 14, 2021 1:39 pm
klaus14 wrote: Tue Sep 14, 2021 1:26 pm
DMoogle wrote: Tue Sep 14, 2021 1:00 pm Putting money where my mouth is. Bought a bunch of VT and ZN today - taxable portfolio now consists of 1.5x in equities and 3x in ITTs. Using IBKR margin at the moment, but have an open order for a box spread to get cheaper financing. Once I can get the single box to fill, I'll order more until my cash balance is no longer negative (or close).
why not:
VXUS + /MES instead of VT ?

(even better: VXUS + VIOV + /MES)

this would be simpler than margin or box spread and would get better financing cost.
Let's say we buy 100k of SPY with a box vs 100k with MES:

Let's say the MES grows 10% per year, but is taxed at ~20% on those gains. So that's the same as growing with a 8% annual growth rate. 100k*1.08^30 = 1006k.

The SPY is taxed at ~15% when it's sold in 30 years. (100k*1.1^30 -100k)*.85 + 100k = 1,498k

These are rough assumptions, and I've not included dividends, so the SPY will be taxed more than I've calculated. . But I think the general idea is correct.

This doesn't include the interest deduction on the box, which would add to the SPY advantage.

SPY with dividiends:

8% annual growth + 2% annual dividends. Dividends taxed at 25%. Thus 9.5% annual growth rate. Guestimating 50k of dividends received over 30 years, thus 50k of basis.

(100k*1.095^30-150k)*.85+150k = 1316k

Again, this doesn't include the interest deduction on the box, which would add to SPY's advantage.
Same calcs for a more realistic 5% assumption of SPY growth with 1% dividends and deducting interest on the box:

For MES, growth rate is 4% after tax. 100k*1.04^30 = 324k

For SPY, growth rate is 4% cap gains, and .75% dividends after tax, and .1% from deduction on .4% of box-spread interest. All gains taxed at 15% at the end, assuming 30k in basis from dividends. (100k*1.0485^30-130)*.85+130 = 371k

If we assume that the financing cost of futures is .2% better, we can penalize the growth rate of SPY by .2%:

(100k*1.0465^30-130)*.85+130 = 351k

Remember, part of the reason financing costs on boxes are higher than futures at the present is because you are locking in a fixed rate for 1-3 years, and yields are expected to rise. 2 years from now, he could have a lower rate on his old box, than the you'll be paying on your futures.
DMoogle
Posts: 549
Joined: Sat Oct 31, 2020 10:24 pm

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by DMoogle »

Thanks for the analysis, skier, as usual. MES is... less intimidating than box spreads for sure, but even with a slightly lower implied financing rate, the tax efficiency of buy & hold equities + box spreads makes it worth it.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

Assuming my math is good!

One mistake I made is that I don't think I should have subtracted the box spread interest deduction. For futures, any loss from financing is already included. For boxes, the loss isn't included until we make it explicit on our tax form. For example SPY goes up 5%, but financing is 2%. For MES we would just get 3% and be taxed on 3%. For SPY, we would get 5% and then subtract 2% when we do our taxes, and again be taxed on the 3%. I guess technically the financing deduction is happening at 40/60 ST/LTCG for futures, and 100% at income tax rates for boxes, but this difference is on the order of .02% annually.

The box spread deduction was .1% in my calculation. If we subtract .1% from the SPY growth rate we'd get 4.55%. If we then add back in .02% because the box spread deduction is at income tax rates, we'd get 4.57% annual growth.

(100k*1.0455^30-130)*.85+130 = 344k

MES was 321k.

So boxes still have a slight lead unless I'm missing something (7% final difference). Assuming my math is right, it's probably close enough now that it's not worth the hassle for those that prefer futures and are close to my assumptions:

1) 20% tax on futures (assumes 28% marginal income tax rate - not an actual rate currently - some will be lower some will be higher)
2) 15% LTCG tax when funds are sold (if you think you could be in the 0% LTCG in retirement then box spreads are better)
3) 30 year time horizon. Longer will favor box-spreads
4) financing cost of box spread is .2% higher than futures, lower will favor boxes, higher will favor futures
5) Higher financing will give slight preference to box-spread because the deduction is at income tax rates instead of 40/60 ST/LTCG. 1% financing is a .05% annual growth rate benefit to boxes, assuming a 5% delta in tax rates between income tax and 40/60ST/LTCG.
6) The LTCG rate stays at 15% for most. An increase would favor futures.

I've also neglected the effect of tax-loss harvesting, and I'm not sure how this would differ beween futures and boxes.
MeanVarianceOptimal
Posts: 10
Joined: Sun Nov 15, 2020 12:36 pm

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by MeanVarianceOptimal »

All of the analysis here assumes that cash is the risk-free asset. However if the goal of my portfolio is to fund my retirement then arguably (inflation protected) long term bonds are really my risk-free asset. Running the portfolio optimization with LTT or LTPZ (15+ year TIPS) as the risk-free asset ends up putting a larger fraction of the portfolio in long term bonds. What do you think?
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

MeanVarianceOptimal wrote: Wed Sep 15, 2021 5:22 am All of the analysis here assumes that cash is the risk-free asset. However if the goal of my portfolio is to fund my retirement then arguably (inflation protected) long term bonds are really my risk-free asset. Running the portfolio optimization with LTT or LTPZ (15+ year TIPS) as the risk-free asset ends up putting a larger fraction of the portfolio in long term bonds. What do you think?
How so? What optimizer are we talking about? Efficient frontier? If using efficient frontier, then using LTT as the risk-free asset will allocate 100% to equities and nothing to bonds including LTT. Efficient frontier maximizes sharpe ratio. Sharpe ratio numerator is the excess annual return over risk free asset. If the risk free asset is LTT, then the excess annual return of LTT is 0%. Thus the sharpe ratio would be zero.

Now if instead of efficient frontier and sharpe ratio, what if our optimization problem was to maximize 30+ year returns with as little variance as possible? Maybe that's what you are suggesting? That's what millennialmillions and I are trying to do here: viewtopic.php?f=10&t=355441

So far we've found that a similar mix of stocks and LTT as the efficient frontier maximizes the 35 year return while minimizing variance. Testing using both the actual historical sequence of returns since 1955, and bootstrapping.
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

My current plan is to go 100/900 stocks/STT. 4xZT contracts which have a nominal value of 880,000 (I think) and then 98,000 VTI. I like that all the treasuries are futures and all the stock is ETF. I like the balance of 1:9. The math is easy. I like the backtesting. Best sharpe ratio since 1955 while keeping it simple. Minimal cash drag. Will sell VTI if margin called and rebalance every quarter as needed. I’d like to go over 100% VTI but I don’t think I can justify the increased volatility.

I’m going to road test in thinkorswim for a little while and then dive in.

Taking comments now!
55% VUG - 20% VEA - 20% EDV - 5% BNDX
AllomancerJack
Posts: 18
Joined: Fri May 21, 2021 11:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by AllomancerJack »

LTCM wrote: Thu Sep 16, 2021 3:01 am My current plan is to go 100/900 stocks/STT. 4xZT contracts which have a nominal value of 880,000 (I think) and then 98,000 VTI. I like that all the treasuries are futures and all the stock is ETF. I like the balance of 1:9. The math is easy. I like the backtesting. Best sharpe ratio since 1955 while keeping it simple. Minimal cash drag. Will sell VTI if margin called and rebalance every quarter as needed. I’d like to go over 100% VTI but I don’t think I can justify the increased volatility.

I’m going to road test in thinkorswim for a little while and then dive in.

Taking comments now!
Would be interesting to know your risk management policy (if any), i.e. how much cash you are going to keep as collateral for ZT contracts?
Also maintaining exactly 100% of stocks seems impossible when using non-leveraged VTI (because you still need some cash in collateral for ZT).
Had you considered adding at least one S&P500 future contract to minimize margin calls/rebalancing on prolonged STT bear market?
User avatar
Bentonkb
Posts: 83
Joined: Sun Mar 07, 2021 5:47 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by Bentonkb »

LTCM wrote: Thu Sep 16, 2021 3:01 am My current plan is to go 100/900 stocks/STT. 4xZT contracts which have a nominal value of 880,000 (I think) and then 98,000 VTI. I like that all the treasuries are futures and all the stock is ETF. I like the balance of 1:9. The math is easy. I like the backtesting. Best sharpe ratio since 1955 while keeping it simple. Minimal cash drag. Will sell VTI if margin called and rebalance every quarter as needed. I’d like to go over 100% VTI but I don’t think I can justify the increased volatility.

I’m going to road test in thinkorswim for a little while and then dive in.

Taking comments now!
The invoice price for 4 /ZT would be $836,041 if the contract expired today. That is a futures price of 110-04.625 * $2000 * TCF. The current TCF is 0.9488.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

LTCM wrote: Thu Sep 16, 2021 3:01 am My current plan is to go 100/900 stocks/STT. 4xZT contracts which have a nominal value of 880,000 (I think) and then 98,000 VTI. I like that all the treasuries are futures and all the stock is ETF. I like the balance of 1:9. The math is easy. I like the backtesting. Best sharpe ratio since 1955 while keeping it simple. Minimal cash drag. Will sell VTI if margin called and rebalance every quarter as needed. I’d like to go over 100% VTI but I don’t think I can justify the increased volatility.

I’m going to road test in thinkorswim for a little while and then dive in.

Taking comments now!
A couple thoughts

1) Check out the Lifecycle vs HFEA thread for the discussion on variance in final outcome, which is really what we are trying to optimize. I'm somewhat concerned by how poorly bonds do in a portfolio during a major rate increase cycle like 1955-1981. Higher stock allocations, and lower bond allocations, seem to produce less variance in the final outcome and higher 10th percentiles. I think you'd have to argue that the 1955-1981 bond cycle is not repeatable to go as heavy as you are in bonds. Stocks seem to be faster mean reverting than leveraged bonds, and require a shorter time horizon. The minimum time horizon for investing in bonds seems to be 30+ years. I'm still thinking this all through.

2) Without getting into the details of implementation, just generally speaking ITT are more consistent with a max-carry or max-roll strategy. The sharpe ratio of using ITT tested back to the 1970s is functionally the same as using STT so we can say either are on the efficient frontier. Given they are equally efficient, might as well take the one with better carry and roll currently. What are the reasons for taking STT over ITT?

3) Most of our peers *younger retail investors) are in stocks, not bonds. Being so heavily in bonds exposes you to the unique risk of returning less than the average retail investor. For example, below market returns could price you out of real-estate that other people in your cohort can afford. For this reason, I favor stocks over bonds a little more than the efficient frontier would suggest.

Personally I am 125/200 stock/ITT.. was thinking of going 125/250 but now I'm thinking of going 130 or 135/200 (taking more risk in stocks), based on the analysis in the Lifecycle vs HFEA thread. I'm going to try to do a more thorough analysis and post it in that thread in the coming days.

I've also been thinking about how I'd psychologically handle a crash in treasuries. I feel confident that psychologically I could handle a stock market crash, as I handled March 2020 just fine. But a crash in treasuries I worry I might feel differently and panic sell. Treasuries aren't diversified the way stocks are. It's one issuer... would the world really end if the U.S. had a managed partial default? Is a 2012 Eurozone crisis happening here impossible? I know the odds of this are super low, but the results of the Lifecycle vs HFEA thread make me question if this miniscule risk is worth it. It's worth thinkng about the "what-ifs" at least.

Just ballparking it the total annual return on 1M in ZT is 3-4k if rates don't go up. Closer to zero if rates go up as forwads project.
Last edited by skierincolorado on Thu Sep 16, 2021 1:36 pm, edited 1 time in total.
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

Bentonkb wrote: Thu Sep 16, 2021 9:38 am
LTCM wrote: Thu Sep 16, 2021 3:01 am My current plan is to go 100/900 stocks/STT. 4xZT contracts which have a nominal value of 880,000 (I think) and then 98,000 VTI. I like that all the treasuries are futures and all the stock is ETF. I like the balance of 1:9. The math is easy. I like the backtesting. Best sharpe ratio since 1955 while keeping it simple. Minimal cash drag. Will sell VTI if margin called and rebalance every quarter as needed. I’d like to go over 100% VTI but I don’t think I can justify the increased volatility.

I’m going to road test in thinkorswim for a little while and then dive in.

Taking comments now!
The invoice price for 4 /ZT would be $836,041 if the contract expired today. That is a futures price of 110-04.625 * $2000 * TCF. The current TCF is 0.9488.
Thanks. I went to CME to find where this is displayed. https://www.cmegroup.com/trading/intere ... es/TCF.xls
The CTD is always the bond with the highest TCF? Is the TCF displayed in thinkorswim anywhere?
55% VUG - 20% VEA - 20% EDV - 5% BNDX
User avatar
Bentonkb
Posts: 83
Joined: Sun Mar 07, 2021 5:47 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by Bentonkb »

LTCM wrote: Thu Sep 16, 2021 1:29 pm
Bentonkb wrote: Thu Sep 16, 2021 9:38 am
LTCM wrote: Thu Sep 16, 2021 3:01 am My current plan is to go 100/900 stocks/STT. 4xZT contracts which have a nominal value of 880,000 (I think) and then 98,000 VTI. I like that all the treasuries are futures and all the stock is ETF. I like the balance of 1:9. The math is easy. I like the backtesting. Best sharpe ratio since 1955 while keeping it simple. Minimal cash drag. Will sell VTI if margin called and rebalance every quarter as needed. I’d like to go over 100% VTI but I don’t think I can justify the increased volatility.

I’m going to road test in thinkorswim for a little while and then dive in.

Taking comments now!
The invoice price for 4 /ZT would be $836,041 if the contract expired today. That is a futures price of 110-04.625 * $2000 * TCF. The current TCF is 0.9488.
Thanks. I went to CME to find where this is displayed. https://www.cmegroup.com/trading/intere ... es/TCF.xls
The CTD is always the bond with the highest TCF? Is the TCF displayed in thinkorswim anywhere?
The CME website is terrible to navigate. https://www.cmegroup.com/tools-informat ... ytics.html

The CTD for each contract is given in the link above. It is the bond with the highest implied repo rate. The same table has the TCF for each bond. Right now the CTD happens to have the highest TCF, but I don't think that is always the case. In fact, the CTD for /ZN is not the highest TCF right now.
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

Bentonkb wrote: Thu Sep 16, 2021 1:50 pm The CME website is terrible to navigate. https://www.cmegroup.com/tools-informat ... ytics.html
Thanks. Very helpful link. That should be in the first post.
55% VUG - 20% VEA - 20% EDV - 5% BNDX
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

AllomancerJack wrote: Thu Sep 16, 2021 3:56 am Would be interesting to know your risk management policy (if any), i.e. how much cash you are going to keep as collateral for ZT contracts?
Also maintaining exactly 100% of stocks seems impossible when using non-leveraged VTI (because you still need some cash in collateral for ZT).
Had you considered adding at least one S&P500 future contract to minimize margin calls/rebalancing on prolonged STT bear market?
As little cash as possible ideally. Since the 4 x /ZT is worth $840,000. I've bought $93,000 VTI to keep the ratio at 1:9. The available balance is $5800. The "not available" balance is $7600 ($1800 collateral for the /ZT).

Since I want to push it and find out how close to the bone I can get and what happens I've now bought an additional 2 x VTI and 73 x VGSH. Available balance is down to $850.

(still paper trades)
55% VUG - 20% VEA - 20% EDV - 5% BNDX
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

skierincolorado wrote: Thu Sep 16, 2021 10:46 am 1) Check out the Lifecycle vs HFEA thread for the discussion on variance in final outcome, which is really what we are trying to optimize. I'm somewhat concerned by how poorly bonds do in a portfolio during a major rate increase cycle like 1955-1981. Higher stock allocations, and lower bond allocations, seem to produce less variance in the final outcome and higher 10th percentiles. I think you'd have to argue that the 1955-1981 bond cycle is not repeatable to go as heavy as you are in bonds. Stocks seem to be faster mean reverting than leveraged bonds, and require a shorter time horizon. The minimum time horizon for investing in bonds seems to be 30+ years. I'm still thinking this all through.

2) Without getting into the details of implementation, just generally speaking ITT are more consistent with a max-carry or max-roll strategy. The sharpe ratio of using ITT tested back to the 1970s is functionally the same as using STT so we can say either are on the efficient frontier. Given they are equally efficient, might as well take the one with better carry and roll currently. What are the reasons for taking STT over ITT?

3) Most of our peers *younger retail investors) are in stocks, not bonds. Being so heavily in bonds exposes you to the unique risk of returning less than the average retail investor. For example, below market returns could price you out of real-estate that other people in your cohort can afford. For this reason, I favor stocks over bonds a little more than the efficient frontier would suggest.
1) I'm having trouble with the spreadsheet supplied in the lifecycle thread. I changed the data to VGSH returns from the simba sheet but I'm not sure where its supposed to spit out the percentile returns. If you want to run it I'd appreciate seeing what happens. When I tested the low percentiles in the simba sheet I wasn't too upset with the returns. 10th percentile still beat 100% stock and Sharpe maximized stock + ITT or LTT portfolios. I didn't test dialing them back to the levels you typically advocate though.

2) Returns. ITT got crushed in the bond crash compared to STT. Back test to mid 80s and you get 1:9 stock/STT. Back to the 70s and you get 1:4 stock/STT. Back to the 50s and you're back to 1:9. If you equalize the ratios on all 3 to get to approx 30% stdev then STT wins on Sharpe, downside, upside, drawdown and ulcer. (I used 100:900 stock/STT 100:350 stock/ITT and 150:150 stock/LTT)

3) I'll try not to think about this! I do currently own more VUG than VTI for this reason. I live in a "tech-wealth" city so if I want a bigger house I'm somewhat correlated with tech success. But I believe in this strategy, own a paid off house and live pretty frugally I think. This is a swing for the fences for my kids and to pass on wealth. My parents generation got their wealth from leverage (in the property market) and I'd like to do the same. This is the cheapest most efficient way to borrow and invest I can think of.
55% VUG - 20% VEA - 20% EDV - 5% BNDX
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

LTCM wrote: Thu Sep 16, 2021 2:53 pm
skierincolorado wrote: Thu Sep 16, 2021 10:46 am 1) Check out the Lifecycle vs HFEA thread for the discussion on variance in final outcome, which is really what we are trying to optimize. I'm somewhat concerned by how poorly bonds do in a portfolio during a major rate increase cycle like 1955-1981. Higher stock allocations, and lower bond allocations, seem to produce less variance in the final outcome and higher 10th percentiles. I think you'd have to argue that the 1955-1981 bond cycle is not repeatable to go as heavy as you are in bonds. Stocks seem to be faster mean reverting than leveraged bonds, and require a shorter time horizon. The minimum time horizon for investing in bonds seems to be 30+ years. I'm still thinking this all through.

2) Without getting into the details of implementation, just generally speaking ITT are more consistent with a max-carry or max-roll strategy. The sharpe ratio of using ITT tested back to the 1970s is functionally the same as using STT so we can say either are on the efficient frontier. Given they are equally efficient, might as well take the one with better carry and roll currently. What are the reasons for taking STT over ITT?

3) Most of our peers *younger retail investors) are in stocks, not bonds. Being so heavily in bonds exposes you to the unique risk of returning less than the average retail investor. For example, below market returns could price you out of real-estate that other people in your cohort can afford. For this reason, I favor stocks over bonds a little more than the efficient frontier would suggest.
1) I'm having trouble with the spreadsheet supplied in the lifecycle thread. I changed the data to VGSH returns from the simba sheet but I'm not sure where its supposed to spit out the percentile returns. If you want to run it I'd appreciate seeing what happens. When I tested the low percentiles in the simba sheet I wasn't too upset with the returns. 10th percentile still beat 100% stock and Sharpe maximized stock + ITT or LTT portfolios. I didn't test dialing them back to the levels you typically advocate though.

2) Returns. ITT got crushed in the bond crash compared to STT. Back test to mid 80s and you get 1:9 stock/STT. Back to the 70s and you get 1:4 stock/STT. Back to the 50s and you're back to 1:9. If you equalize the ratios on all 3 to get to approx 30% stdev then STT wins on Sharpe, downside, upside, drawdown and ulcer. (I used 100:900 stock/STT 100:350 stock/ITT and 150:150 stock/LTT)

3) I'll try not to think about this! I do currently own more VUG than VTI for this reason. I live in a "tech-wealth" city so if I want a bigger house I'm somewhat correlated with tech success. But I believe in this strategy, own a paid off house and live pretty frugally I think. This is a swing for the fences for my kids and to pass on wealth. My parents generation got their wealth from leverage (in the property market) and I'd like to do the same. This is the cheapest most efficient way to borrow and invest I can think of.
1) will post something comprehensive soon

2) when you say stt wins, are you looking at by how much? The tests back to the 1970s showed stt and itt to be very very close in terms of sharpe - functionally tied. Stt has usually had more roll and carry than it does currently so all else equal I’d go with the max roll or max carry. If you’ve got something that shows a statistically meaningful lead for stt over itt in terms of sharpe or drawdown could you point me to it or post?

3) it’s not so much the psychology of falling relative to peers but the real financial consequence. This certainly isn’t the most important factor but just one more reason to maybe lean a little more towards equities. I think reason #1 is the biggest reason
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

skierincolorado wrote: Thu Sep 16, 2021 3:17 pm 2) when you say stt wins, are you looking at by how much? The tests back to the 1970s showed stt and itt to be very very close in terms of sharpe - functionally tied. Stt has usually had more roll and carry than it does currently so all else equal I’d go with the max roll or max carry. If you’ve got something that shows a statistically meaningful lead for stt over itt in terms of sharpe or drawdown could you point me to it or post?
I'm going as far back as I can get. Since 1955 ITT scores 0.26 sharpe. STT scores 0.37. That seems pretty significant to me. ITT is still good. Just from what I can tell STT is better.

Image
55% VUG - 20% VEA - 20% EDV - 5% BNDX
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

LTCM wrote: Thu Sep 16, 2021 3:29 pm
skierincolorado wrote: Thu Sep 16, 2021 3:17 pm 2) when you say stt wins, are you looking at by how much? The tests back to the 1970s showed stt and itt to be very very close in terms of sharpe - functionally tied. Stt has usually had more roll and carry than it does currently so all else equal I’d go with the max roll or max carry. If you’ve got something that shows a statistically meaningful lead for stt over itt in terms of sharpe or drawdown could you point me to it or post?
I'm going as far back as I can get. Since 1955 ITT scores 0.26 sharpe. STT scores 0.37. That seems pretty significant to me. ITT is still good. Just from what I can tell STT is better.

Image
Interesting thanks will check it out also in combination with equities which can change things especially given the higher leverage stt require. Not sure 200% vs 100% is the correct comparison could be more like 300%
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

LTCM wrote: Thu Sep 16, 2021 3:29 pm
skierincolorado wrote: Thu Sep 16, 2021 3:17 pm 2) when you say stt wins, are you looking at by how much? The tests back to the 1970s showed stt and itt to be very very close in terms of sharpe - functionally tied. Stt has usually had more roll and carry than it does currently so all else equal I’d go with the max roll or max carry. If you’ve got something that shows a statistically meaningful lead for stt over itt in terms of sharpe or drawdown could you point me to it or post?
I'm going as far back as I can get. Since 1955 ITT scores 0.26 sharpe. STT scores 0.37. That seems pretty significant to me. ITT is still good. Just from what I can tell STT is better.

Image
Something interesting I've found maybe you know why it's happening:

I'm getting pretty different sharpe ratios between PV and simba's spreadsheet. .01 or .03 I could understand.. but it's .06 different. PV only goes back to 1977 for the efficient frontier with STT, so doing the same in Simba's spreadhseet I get:

sharpe ratio for TSM+STT: .66
for TSM+ITT: .60

using PV I get .655 for TSM+ITT and .650 for TSM+STT - that was the basis of my initial statement that they are functionally the same. But the caveat is 1) using PV 2) 1977-present

Going back to 1955 in simba's makes both do worse by about the same amount. So it's not the year that's the difference. 1977 vs 1955 isn't the cause of our different conclusion. It's simba vs PV. We'll need to figure out the difference between their timeseries. Why does PV like ITT more than Simba?
Last edited by skierincolorado on Thu Sep 16, 2021 3:58 pm, edited 1 time in total.
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

Simba uses annual returns and PV uses monthly returns I think. Would that account for the difference?
55% VUG - 20% VEA - 20% EDV - 5% BNDX
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

LTCM wrote: Thu Sep 16, 2021 3:58 pm Simba uses annual returns and PV uses monthly returns I think. Would that account for the difference?
I don't think so

I posed the question in the simba backsheet thread
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by LTCM »

Do they use the same funds for the entire period?
55% VUG - 20% VEA - 20% EDV - 5% BNDX
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

LTCM wrote: Thu Sep 16, 2021 4:09 pm Do they use the same funds for the entire period?
The simba spreadsheet calculates it synthetically using a bond return simulator prior to whenever a fund was available. PV does the same.
DMoogle
Posts: 549
Joined: Sat Oct 31, 2020 10:24 pm

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by DMoogle »

Practically speaking, I wonder if the cost of leverage weighs largely in favor of ITTs. Sure, the implied financing rate of STT should be lower, but given you need more than 2x as much STT compared to ITT to get the same return, the levered Sharpe may end up being lower when that's put into consideration.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

DMoogle wrote: Thu Sep 16, 2021 4:19 pm Practically speaking, I wonder if the cost of leverage weighs largely in favor of ITTs. Sure, the implied financing rate of STT should be lower, but given you need more than 2x as much STT compared to ITT to get the same return, the levered Sharpe may end up being lower when that's put into consideration.
I don't think leverage affects sharpe ratios because sharpe ratios are calculated as return over the risk free rate. It might a little bit in PV because of the monthly rebalancing. But if borrowing costs affected the efficient frontier or sharpe ratios, then you'd see a much bigger drop off as you add leverage. The drop off we see is likely due to misfortunate rebalancing times.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Modified versions of HFEA with ITT and Futures / Lifecycle Investing with Modern Portfolio Theory

Post by skierincolorado »

LTCM wrote: Thu Sep 16, 2021 4:09 pm Do they use the same funds for the entire period?
Think the answer in the simba thread explains it. They are laddering treasuries from 10 years to 3 years. So it's equal weight in treasuries from 3-10 years. Given the drop off in risk adjusted returns after 5 years and the flattening of the curve, it makes sense that the simba index would have lower risk-adjusted returns than the 5-yr constant maturity index used in PV. The fact that it owns 3 yr duration bonds doesn't make up for owning 10yr duration bonds, because the sharpe ratio from 2-5 years is very very similar, and then it drops off. So they're investing in the flatter part of the curve with lower sharpe ratios.

I feel confident that PV is a good proxy for ZF (5 years vs 4.5). PV gives a hair higher sharpe ratio since 1977 to TSM+ITT than it does to TSM+STT. If we could backtest synthetic ZF to 1955 in Simba, I'm confident that the sharpe would be similar to or higher than STT. My guess is higher, but even if we go with equal, there are several other reason to go with ITT right now. More roll and carry. Less explicit leverage.

1M of ZT has roll and carry of 3-4k right now
300k of ZF has roll and carry of 6k right now

So if I have to pick one I pick ZF. But comeinvest and I have discussed how diversifying across the curve is probably best, like ZT/ZF/ZN/TN
Post Reply