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.
EfficientInvestor
Posts: 580
Joined: Thu Nov 01, 2018 7:02 pm
Location: Alabama

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

Post by EfficientInvestor »

skierincolorado wrote: Sat Dec 04, 2021 11:46 pm I think I see how this would be theta neutral. But isn't the return profile still very different from futures? If SPY drops, you have more downside risk than upside potential. Over time, the covered call premium should offset. So is the idea that by staying theta neutral, it will all average out much faster? Over a long enough time period, the covered call position basically is just earning theta.

But it still seems problematic when SPY goes down substantially in a month or two. You'll lose money on the covered call which could force you to change your AA. It seems like it might be better to do what I suggested and harvest IV when SPY drops.

Also I don't understand why it's 1 monthly covered call for every 3-4 LEAPS. A covered call would have a ~3% monthly premium. Which is much more than the ~3% annual premium of the LEAP (12x more, not 3-4x more).
It is definitely a different return profile than just using futures. I would say anything with LEAPS will be different because using futures is just straight leverage whereas LEAPS are leverage plus insurance. Yes, the idea is to stay theta neutral and things will average out over time.

I said 1 per 3 or 4 because I usually buy my LEAPS around 70 delta because I like having a larger hedge. In your case, with a LEAP so far in the money that is over 90 delta, it looks like you would only need to sell 1 call for every 9 LEAPS or so in order to be theta neutral. So your overall return profile isn't a whole lot different than not selling any calls because you are only selling a call against ~10% of your position. The other 90% of your position and return profile remains unchanged.

To your point about SPY going down...you will be in a better position by selling the call because it's an additional hedge. As SPY decreases, a covered call that starts at 50 delta will start to decrease in value. I usually choose to roll the option when it gets to 10 delta and reset it at 50 delta again. On the other end, if SPY goes up, you get a little less return. I roll this position when it gets to 90 delta. I don't roll all the way to 50 though. I roll it to 70 and give SPY a chance to come back down. In your case, this isn't a big deal. The other 8 LEAPS that are uncovered can continue to run without the return being capped by the covered call. If you start with 9 LEAPS at 95 delta apiece, total delta is 855. If you sell a 50 delta call against it, you get a total delta of 805. Let's pretend that you were aiming for a total SPY exposure of 800 delta to begin with, so your initial delta is close enough to target and your net theta, when selling a 50 delta call 1 month out, is actually slightly positive based on current pricing. If the short position goes up or down 40 delta, your total delta will go up to 845 or down to 765, not accounting for potential delta change of the LEAPS. So you have to be comfortable with this kind of delta variability. However, it's my outlook that this will even out over time and the average over time will have my close enough to the 800 delta target. In regard to the LEAPS...if SPY continues to go up, delta of LEAPS approaches 100 and your total LEAP delta increases to 900. If SPY crashes, your LEAP delta starts to decrease and as it decreases you start to receive more benefit from the hedge.

There is lots to unpack in all of this. Hopefully this makes sense. As I said, this approach isn't perfect. You have to be okay with some exposure variability.
Hfearless
Posts: 135
Joined: Sat Aug 14, 2021 9:53 am

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

Post by Hfearless »

Lock wrote: Sun Dec 05, 2021 5:29 pm Has anyone compared and contrasted the use of margin versus portfolio margin on IBKR? Any noticeable difference?
Well, portfolio margin is superior to Reg T margin in all aspects for all but the weirdest portfolios, so go for it if you qualify.
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 »

Lock wrote: Sun Dec 05, 2021 5:29 pm Has anyone compared and contrasted the use of margin versus portfolio margin on IBKR? Any noticeable difference?
Reg t is a pain with futures because it really limits your leverage because the sma can go negative when the equities are anywhere close to 2x leverage.

Portfolio is better in every regard from what I understand. Most of my money is in tax advantages so i don’t qualify. Reg t you just have to be more careful. I can explain more if anyone is using reg t.
Hfearless
Posts: 135
Joined: Sat Aug 14, 2021 9:53 am

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

Post by Hfearless »

Also box spreads only really work under portfolio margin, right?
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 »

Hfearless wrote: Mon Dec 06, 2021 5:24 pm Also box spreads only really work under portfolio margin, right?
I have one in reg t
L2F
Posts: 26
Joined: Fri Sep 24, 2021 9:33 am

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

Post by L2F »

skierincolorado wrote: Mon Dec 06, 2021 5:16 pm
Lock wrote: Sun Dec 05, 2021 5:29 pm Has anyone compared and contrasted the use of margin versus portfolio margin on IBKR? Any noticeable difference?
Reg t is a pain with futures because it really limits your leverage because the sma can go negative when the equities are anywhere close to 2x leverage.

Portfolio is better in every regard from what I understand. Most of my money is in tax advantages so i don’t qualify. Reg t you just have to be more careful. I can explain more if anyone is using reg t.
I'm using Reg T and want to try mHFEA starting next month, so would be thankful for any advice from you.
I'm still not decided on exact leverage ratio, but I'm leaning towards few /ZF contracts + VTI + S&P500 LETFs.
Last edited by L2F on Tue Dec 07, 2021 3:59 pm, edited 1 time in total.
parval
Posts: 153
Joined: Tue Oct 22, 2019 9:23 pm

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

Post by parval »

Anyone implement this on IBKR w/ futures? I have portfolio and box spreads setup, just waiting for new year to convert some UPRO/TMF into ES/ZN, anyone want to share their ratios/rough numbers?
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 »

parval wrote: Mon Dec 06, 2021 10:50 pm Anyone implement this on IBKR w/ futures? I have portfolio and box spreads setup, just waiting for new year to convert some UPRO/TMF into ES/ZN, anyone want to share their ratios/rough numbers?
I am. $463k net liquidation. $726k ZF, $392k ZN, $784k stocks/ETFs.
comeinvest
Posts: 2669
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 »

There are some other threads about series I and EE savings bonds; the I bonds currently pay 7.12%. Did anybody consider leveraging savings bonds with box spreads? To replace a portion of the treasuries that is unlikely to ever be needed for rebalancing.
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 »

comeinvest wrote: Tue Dec 07, 2021 1:17 am There are some other threads about series I and EE savings bonds; the I bonds currently pay 7.12%. Did anybody consider leveraging savings bonds with box spreads? To replace a portion of the treasuries that is unlikely to ever be needed for rebalancing.
They have 10k max per year per person limit. You should max it financing with box spreads or not.
My investment algorithm: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=351899&p=6112869#p6112869
comeinvest
Posts: 2669
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 »

klaus14 wrote: Tue Dec 07, 2021 1:46 am
comeinvest wrote: Tue Dec 07, 2021 1:17 am There are some other threads about series I and EE savings bonds; the I bonds currently pay 7.12%. Did anybody consider leveraging savings bonds with box spreads? To replace a portion of the treasuries that is unlikely to ever be needed for rebalancing.
They have 10k max per year per person limit. You should max it financing with box spreads or not.
What about EE bonds? Current interest you can forget; but you have to hold 20 years, which is when they will double in value -> about 3.5% p.a. IRR. Basically starting from the current interest rate level, you are probably "effectively" locked in for 20 years. Drawback vs. treasury futures: 1. The won't flood your account with cash during a macro crash; 2. no riding the yield curve (no rolldown yield). Benefits: 1. 3.5% significantly higher than current 20y or even ITT-equivalent yields; 2. no state tax; 3. tax only payable at the end. Thoughts?
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 »

comeinvest wrote: Tue Dec 07, 2021 2:52 am
klaus14 wrote: Tue Dec 07, 2021 1:46 am
comeinvest wrote: Tue Dec 07, 2021 1:17 am There are some other threads about series I and EE savings bonds; the I bonds currently pay 7.12%. Did anybody consider leveraging savings bonds with box spreads? To replace a portion of the treasuries that is unlikely to ever be needed for rebalancing.
They have 10k max per year per person limit. You should max it financing with box spreads or not.
What about EE bonds? Current interest you can forget; but you have to hold 20 years, which is when they will double in value -> about 3.5% p.a. IRR. Basically starting from the current interest rate level, you are probably "effectively" locked in for 20 years. Drawback vs. treasury futures: 1. The won't flood your account with cash during a macro crash; 2. no riding the yield curve (no rolldown yield). Benefits: 1. 3.5% significantly higher than current 20y or even ITT-equivalent yields; 2. no state tax; 3. tax only payable at the end. Thoughts?
If you buy EE bonds this year. 20y rising to 3.5% in 2024 won't be enough to catch up. it should go much higher since you accumulated 3 years.

They can complement ITT strategy. It is only 10k /y. If you buy both I Bonds and EE bonds for 20 years. They will create a nice income stream 20 years later. This can bridge one to social security @ 70 yo.

If 20y treasury goes above 3.5%, that year you can buy 20y treasury instead of EE Bonds.
My investment algorithm: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=351899&p=6112869#p6112869
comeinvest
Posts: 2669
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 »

klaus14 wrote: Tue Dec 07, 2021 2:59 am
comeinvest wrote: Tue Dec 07, 2021 2:52 am
klaus14 wrote: Tue Dec 07, 2021 1:46 am
comeinvest wrote: Tue Dec 07, 2021 1:17 am There are some other threads about series I and EE savings bonds; the I bonds currently pay 7.12%. Did anybody consider leveraging savings bonds with box spreads? To replace a portion of the treasuries that is unlikely to ever be needed for rebalancing.
They have 10k max per year per person limit. You should max it financing with box spreads or not.
What about EE bonds? Current interest you can forget; but you have to hold 20 years, which is when they will double in value -> about 3.5% p.a. IRR. Basically starting from the current interest rate level, you are probably "effectively" locked in for 20 years. Drawback vs. treasury futures: 1. The won't flood your account with cash during a macro crash; 2. no riding the yield curve (no rolldown yield). Benefits: 1. 3.5% significantly higher than current 20y or even ITT-equivalent yields; 2. no state tax; 3. tax only payable at the end. Thoughts?
If you buy EE bonds this year. 20y rising to 3.5% in 2024 won't be enough to catch up. it should go much higher since you accumulated 3 years.

They can complement ITT strategy. It is only 10k /y. If you buy both I Bonds and EE bonds for 20 years. They will create a nice income stream 20 years later. This can bridge one to social security @ 70 yo.

If 20y treasury goes above 3.5%, that year you can buy 20y treasury instead of EE Bonds.
"If you buy EE bonds this year. 20y rising to 3.5% in 2024 won't be enough to catch up." - I think it probably will, if rolldown returns are, for example, 0.5%. (I didn't do the math.)

"If 20y treasury goes above 3.5%, that year you can buy 20y treasury instead of EE Bonds." - no, because your existing EEs will have a higher YTM at that time, and new EEs will probably have a significantly higher base rate (as the doubling after 20 years would be not much incentive any more). Basically you are most likely locked in for 20 years if you buy now (unless you redeem at a relative loss for an emergency). I'd say the case for EE is not as clear as that for I.
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 »

comeinvest wrote: Tue Dec 07, 2021 3:16 am
"If you buy EE bonds this year. 20y rising to 3.5% in 2024 won't be enough to catch up." - I think probably it will, if rolldown returns are, for example, 0.5%. (I didn't do the math.)

"If 20y treasury goes above 3.5%, that year you can buy 20y treasury instead of EE Bonds." - no, because your existing EEs will have a higher YTM at that time, and new EEs will probably have a significantly higher base rate (as the doubling after 20 years would be not much incentive any more). Basically you are most likely locked in for 20 years if you buy now. I'd say the case for EE is not as clear as that for I.
Well, existing positions... If rates rise to 3.5% you already lost a ton due to price decrease of bonds.
Every year you should compare 20y rate vs EE bond rate and decide.
The downsides will only apply if you are capital constrained and your only bond holding is EE bonds. Otherwise i think it is no brainer.
My investment algorithm: https://www.bogleheads.org/forum/viewtopic.php?f=10&t=351899&p=6112869#p6112869
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

comeinvest wrote: Tue Dec 07, 2021 1:17 am There are some other threads about series I and EE savings bonds; the I bonds currently pay 7.12%. Did anybody consider leveraging savings bonds with box spreads? To replace a portion of the treasuries that is unlikely to ever be needed for rebalancing.
Series I and EE savings bonds do not have rate risk (not mark to market), so I don't consider them as an alternative to treasuries to achieve my target risk profile. They are a separate asset class from a risk perspective and something that can be added on top of the stock-bond portfolio.

Series I savings bonds, assuming their real return is lower-bounded at 0%, will be a profit as long as we can borrow at negative real interest rates. Given the 10 year TIPS yield is -1%, that is probably a safe bet. Also, if our borrowing rate exceeds 0% real it is only a minor penalty to close the position early. In addition, as the return is correlated with inflation, it helps to counteract negative inflation exposure in the stock-bond portfolio. So I-bonds are amazing. EE-bonds are also a better deal than any marketed security, but I think I-bonds are even better.

The downside to both is that saving bonds cannot be used as collateral in the taxable account and the funding cannot be moved back freely. So withdrawing box spread funded financing to buy saving bonds would increase the leverage of the account. The math gets trickier if there is a trade-off in investing in saving bonds vs having to increase the leverage of the stock-bond portfolio to compensate.

I have I-bonds but no EE-bonds.
parval
Posts: 153
Joined: Tue Oct 22, 2019 9:23 pm

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

Post by parval »

DMoogle wrote: Tue Dec 07, 2021 12:32 am
parval wrote: Mon Dec 06, 2021 10:50 pm Anyone implement this on IBKR w/ futures? I have portfolio and box spreads setup, just waiting for new year to convert some UPRO/TMF into ES/ZN, anyone want to share their ratios/rough numbers?
I am. $463k net liquidation. $726k ZF, $392k ZN, $784k stocks/ETFs.
Thanks so much for sharing! Super noob question but trying to understand how that works: I see on yahoo finance ZN is ~130, but on cme, it says face value at maturity is 100k, so when you have ~400k ZN, is that 4 ZN contracts?

Also I remember you mentioned you setup box spreads, how does your leverage look w/ this setup? I have portfolio margin with VTI/EDV for now but adding box spreads doubled my leverage for some reason.
comeinvest
Posts: 2669
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 »

zkn wrote: Tue Dec 07, 2021 9:51 am
comeinvest wrote: Tue Dec 07, 2021 1:17 am There are some other threads about series I and EE savings bonds; the I bonds currently pay 7.12%. Did anybody consider leveraging savings bonds with box spreads? To replace a portion of the treasuries that is unlikely to ever be needed for rebalancing.
Series I and EE savings bonds do not have rate risk (not mark to market), so I don't consider them as an alternative to treasuries to achieve my target risk profile. They are a separate asset class from a risk perspective and something that can be added on top of the stock-bond portfolio.
They are "separate" as they are inflation-protected bonds vs. treasuries which are nominal bonds. They are also "separate" in the sense that, as you say, they cannot be used to rebalance the stock/bond portfolio (unlikely that the totality of your fixed income would ever be rebalanced to stocks), and they won't go up on the market when interest rates go down. But other than that I think they should be considered as part of the "bond" allocation of the portfolio for purpose of estimating your probabilistic final portfolio outcome after x years, as they are fixed income - like treasuries - i.e. they produce a fixed income, in this case a *real* fixed income i.e. they move with *real* rates.
zkn wrote: Tue Dec 07, 2021 9:51 am Series I savings bonds, assuming their real return is lower-bounded at 0%, will be a profit as long as we can borrow at negative real interest rates. Given the 10 year TIPS yield is -1%, that is probably a safe bet. Also, if our borrowing rate exceeds 0% real it is only a minor penalty to close the position early. In addition, as the return is correlated with inflation, it helps to counteract negative inflation exposure in the stock-bond portfolio. So I-bonds are amazing. EE-bonds are also a better deal than any marketed security, but I think I-bonds are even better.

The downside to both is that saving bonds cannot be used as collateral in the taxable account and the funding cannot be moved back freely. So withdrawing box spread funded financing to buy saving bonds would increase the leverage of the account. The math gets trickier if there is a trade-off in investing in saving bonds vs having to increase the leverage of the stock-bond portfolio to compensate.

I have I-bonds but no EE-bonds.
Your sentence that I put in bold is a drawback that I forgot to mention in my initial post. There certainly will be a tradeoff between how much in I-Bonds vs. the amount of possible / reasonable leverage in the brokerage account disregarding the debt-financed I-Bonds. It's also what makes me question EE-Bonds, other than the probability that interest rates go above [3.5% - yield curve rolldown returns]. I intend to buy I-Bonds, but I'm uncertain about EE-Bonds.
Last edited by comeinvest on Tue Dec 07, 2021 2:40 pm, edited 3 times in total.
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 »

parval wrote: Tue Dec 07, 2021 1:05 pm
DMoogle wrote: Tue Dec 07, 2021 12:32 am
parval wrote: Mon Dec 06, 2021 10:50 pm Anyone implement this on IBKR w/ futures? I have portfolio and box spreads setup, just waiting for new year to convert some UPRO/TMF into ES/ZN, anyone want to share their ratios/rough numbers?
I am. $463k net liquidation. $726k ZF, $392k ZN, $784k stocks/ETFs.
Thanks so much for sharing! Super noob question but trying to understand how that works: I see on yahoo finance ZN is ~130, but on cme, it says face value at maturity is 100k, so when you have ~400k ZN, is that 4 ZN contracts?

Also I remember you mentioned you setup box spreads, how does your leverage look w/ this setup? I have portfolio margin with VTI/EDV for now but adding box spreads doubled my leverage for some reason.
Noob questions welcome; first time I bought a futures contract was a few months ago. I'm still learning myself.

Ignore the value at maturity; current value is what matters. My balance is made up of 3 ZN contracts and 6 ZF contracts.

I have two 5000-4000 boxes for an effective 2x $100k loan, and two 4600-4400 boxes for an effective 2x $20k loan. I also have a negative cash balance of $87k that's being financed by IBKR's margin. I have an open box spread order that I'm trying to get filled to return my cash balance to ~$0.
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 »

I'd like to pose a question, and maybe it's worth a new thread - what do people think about introducing leverage rebalancing bands?

In HFEA, you're always at a constant 3x ratio, so the only rebalancing is between asset allocation and arguably not that big of a deal. However, with mHFEA, there is added risk in that your leverage ratio can vary wildly based on intraquarter performance. In the event of a crash, you can be way overleveraged vs. whatever your desired ratios are. Rebalancing bands seem like the logical way to deal with that risk. Thoughts?
Hfearless
Posts: 135
Joined: Sat Aug 14, 2021 9:53 am

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

Post by Hfearless »

parval wrote: Tue Dec 07, 2021 1:05 pm Thanks so much for sharing! Super noob question but trying to understand how that works: I see on yahoo finance ZN is ~130, but on cme, it says face value at maturity is 100k, so when you have ~400k ZN, is that 4 ZN contracts?
That’s $100k worth of hypothetical bonds yielding 6%. No wonder the market values that considerably higher than $100k.

Anyway what matters is the notional value, what Yahoo shows. If you buy one contract, and /ZN goes up or down by 0.1%, you gain or lose $130.
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

comeinvest wrote: Tue Dec 07, 2021 2:35 pm They are "separate" as they are inflation-protected bonds vs. treasuries which are nominal bonds. They are also "separate" in the sense that, as you say, they cannot be used to rebalance the stock/bond portfolio (unlikely that the totality of your fixed income would ever be rebalanced to stocks), and they won't go up on the market when interest rates go down. But other than that I think they should be considered as part of the "bond" allocation of the portfolio for purpose of estimating your probabilistic final portfolio outcome after x years, as they are fixed income - like treasuries - i.e. they produce a fixed income, in this case a *real* fixed income i.e. they move with *real* rates.
The distinction I was referring to is that the value of the bond does not vary with interest rates. If rates go up you can still withdraw the original amount. So they are totally different from a risk perspective to rolling a bond/bond fund/bond future where the value can go down which exposes the portfolio to term risk. In other words, removing treasuries from the portfolio and adding I-bonds will reduce the term risk beta of the portfolio. There is only rate risk associated with the early withdrawal penalty if real rates exceeded zero before the bond was five years old such that we could no longer borrow for less than zero and wanted to withdraw the money.

Similarly, I would make a distinction between rolling a bond/bond fund/future and committing to hold a bond until maturity that you did not need (or use) as collateral. The former would be intended to harvest the term premium whereas the latter would be about locking in a risk-free rate and would not contribute risk. (The price would still vary but that would not matter in the long run unless it is collateral.)

Another way to think about it is you could be short a nominal treasury and long a TIPS bond of the same maturity to create a position that moves with CPI but no term risk. Going long I-bonds instead of TIPS is even better, because the real return on I-bonds is a minimum of 0%, so you get to harvest the difference between 0% and the real rate as an added bonus in addition to the long CPI exposure. We should also consider ~.4% additional financing cost on the box spread.
comeinvest wrote: Tue Dec 07, 2021 2:35 pm Your sentence that I put in bold is a drawback that I forgot to mention in my initial post. There certainly will be a tradeoff between how much in I-Bonds vs. the amount of possible / reasonable leverage in the brokerage account disregarding the debt-financed I-Bonds. It's also what makes me question EE-Bonds, other than the probability that interest rates go above [3.5% - yield curve rolldown returns]. I intend to buy I-Bonds, but I'm uncertain about EE-Bonds.
I am in the same boat. If I was very much lower than the "reasonable leverage" threshold, I would buy EE-bonds too, but otherwise I intend to stick with I-bonds for now. I sleep better at night knowing I have collateral in the account or available to move into the account and locking up the money for 20 years is less appealing.
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 »

L2F wrote: Mon Dec 06, 2021 7:04 pm
skierincolorado wrote: Mon Dec 06, 2021 5:16 pm
Lock wrote: Sun Dec 05, 2021 5:29 pm Has anyone compared and contrasted the use of margin versus portfolio margin on IBKR? Any noticeable difference?
Reg t is a pain with futures because it really limits your leverage because the sma can go negative when the equities are anywhere close to 2x leverage.

Portfolio is better in every regard from what I understand. Most of my money is in tax advantages so i don’t qualify. Reg t you just have to be more careful. I can explain more if anyone is using reg t.
I'm using Reg T and planning to try mHFEA starting next month, so would be thankful for any advice from you.
I'm still not decided on exact leverage ratio, but most likely I will use few /ZF contracts + VTI + S&P500 LETFs.
It's easiest to run through with an example. Say you have 50k account value. You buy 80k of VTI with a 30k margin loan (refinanced with box spread). That's within the Reg T requirement of up to 50% which would have allowed you to borrow up to 50k. If the market drops 30%, you'd lose 24k of equity, and have 26k of equity remaining and 60k of stock over 2x leverage - but this would still be fine with Reg T.

Reg T has the initial 50% requirement, but also the SMA (special memorandem agreement) must stay positive.

SMA is simply the greater of either 1) equity - reg t requirement or 2) previous day SMA +/- change in cash +/- initial reg t margin requirements of new trades.

In the above example, there is no change in cash balance no matter how much VTI goes down - it's not affecting cash. The SMA would be zero once the 50% requirement is exceeded but would not go negative because the cash is never changing. So the limiting factor will become the broker's margin requirement (usually 25%). SInce 25% margin requirement allows up to 4x leverage, which is more than any of us should possibly need, there is no risk of a margin call. If the market continued to go down in the above example, one might need to sell some VTI and de-lever in order to stay near 2x leverage, but it wasn't forced by the broker and the initial 25% market drop was not problematic.

But now let's look what would happen if we held a single ZF in the same account. Once the SMA hit 0, any negative change in the cash balance due to ZF will cause the SMA to go negative and a margin call. If the market fell ~22% the SMA would hit zero. 62k of stock with 32k of equity. The reg t requirement would be .5*62 + 1k for the ZF = 32k. The SMA is now zero. Any chance drop in cash balance due to the ZF will cause the SMA to go negative and a margin call. So while in the first example the market fell 25% we were still no where near a margin call, but holding a future in the same account, we'd get margin called around 22% market drop.

I didn't understand SMA initially and got a margin call because the SMA went negative. Portfolio eliminates this problem because there is no SMA. Obviously you still need to keep your leverage from getting out of control in a large market drop, but there is much more flexibility. With Reg T you basically get margin called at 2x leverage if you hold futures in the account, whereas if you don't hold futures in the account or have portfolio margin you won't get margin called until ~4x leverage depending on the broker. So a lot more breathing room.

The example here is very helpful:
https://www.clientam.com.hk/en/index.ph ... =overview3
comeinvest
Posts: 2669
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 »

zkn wrote: Tue Dec 07, 2021 3:29 pm
comeinvest wrote: Tue Dec 07, 2021 2:35 pm They are "separate" as they are inflation-protected bonds vs. treasuries which are nominal bonds. They are also "separate" in the sense that, as you say, they cannot be used to rebalance the stock/bond portfolio (unlikely that the totality of your fixed income would ever be rebalanced to stocks), and they won't go up on the market when interest rates go down. But other than that I think they should be considered as part of the "bond" allocation of the portfolio for purpose of estimating your probabilistic final portfolio outcome after x years, as they are fixed income - like treasuries - i.e. they produce a fixed income, in this case a *real* fixed income i.e. they move with *real* rates.
The distinction I was referring to is that the value of the bond does not vary with interest rates. If rates go up you can still withdraw the original amount. So they are totally different from a risk perspective to rolling a bond/bond fund/bond future where the value can go down which exposes the portfolio to term risk. In other words, removing treasuries from the portfolio and adding I-bonds will reduce the term risk beta of the portfolio. There is only rate risk associated with the early withdrawal penalty if real rates exceeded zero before the bond was five years old such that we could no longer borrow for less than zero and wanted to withdraw the money.

Similarly, I would make a distinction between rolling a bond/bond fund/future and committing to hold a bond until maturity that you did not need (or use) as collateral. The former would be intended to harvest the term premium whereas the latter would be about locking in a risk-free rate and would not contribute risk. (The price would still vary but that would not matter in the long run unless it is collateral.)

Another way to think about it is you could be short a nominal treasury and long a TIPS bond of the same maturity to create a position that moves with CPI but no term risk. Going long I-bonds instead of TIPS is even better, because the real return on I-bonds is a minimum of 0%, so you get to harvest the difference between 0% and the real rate as an added bonus in addition to the long CPI exposure. We should also consider ~.4% additional financing cost on the box spread.
comeinvest wrote: Tue Dec 07, 2021 2:35 pm Your sentence that I put in bold is a drawback that I forgot to mention in my initial post. There certainly will be a tradeoff between how much in I-Bonds vs. the amount of possible / reasonable leverage in the brokerage account disregarding the debt-financed I-Bonds. It's also what makes me question EE-Bonds, other than the probability that interest rates go above [3.5% - yield curve rolldown returns]. I intend to buy I-Bonds, but I'm uncertain about EE-Bonds.
I am in the same boat. If I was very much lower than the "reasonable leverage" threshold, I would buy EE-bonds too, but otherwise I intend to stick with I-bonds for now. I sleep better at night knowing I have collateral in the account or available to move into the account and locking up the money for 20 years is less appealing.
I revise my previous statement. Leveraged I-Bonds are not part of the bond portion of the asset allocation, but a pure arbitrage play. That's why usually there would be no positive return, as there is (almost) no risk - neither credit nor term risk. It's currently an anomaly, or more precisely, I-Bonds are mispriced. That's why the amount that can be purchased is limited.
Last edited by comeinvest on Tue Dec 07, 2021 6:09 pm, edited 2 times in total.
L2F
Posts: 26
Joined: Fri Sep 24, 2021 9:33 am

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

Post by L2F »

skierincolorado wrote: Tue Dec 07, 2021 4:04 pm
L2F wrote: Mon Dec 06, 2021 7:04 pm
skierincolorado wrote: Mon Dec 06, 2021 5:16 pm
Lock wrote: Sun Dec 05, 2021 5:29 pm Has anyone compared and contrasted the use of margin versus portfolio margin on IBKR? Any noticeable difference?
Reg t is a pain with futures because it really limits your leverage because the sma can go negative when the equities are anywhere close to 2x leverage.

Portfolio is better in every regard from what I understand. Most of my money is in tax advantages so i don’t qualify. Reg t you just have to be more careful. I can explain more if anyone is using reg t.
I'm using Reg T and planning to try mHFEA starting next month, so would be thankful for any advice from you.
I'm still not decided on exact leverage ratio, but most likely I will use few /ZF contracts + VTI + S&P500 LETFs.
It's easiest to run through with an example. Say you have 50k account value. You buy 80k of VTI with a 30k margin loan (refinanced with box spread). That's within the Reg T requirement of up to 50% which would have allowed you to borrow up to 50k. If the market drops 30%, you'd lose 24k of equity, and have 26k of equity remaining and 60k of stock over 2x leverage - but this would still be fine with Reg T.

Reg T has the initial 50% requirement, but also the SMA (special memorandem agreement) must stay positive.

SMA is simply the greater of either 1) equity - reg t requirement or 2) previous day SMA +/- change in cash +/- initial reg t margin requirements of new trades.

In the above example, there is no change in cash balance no matter how much VTI goes down - it's not affecting cash. The SMA would be zero once the 50% requirement is exceeded but would not go negative because the cash is never changing. So the limiting factor will become the broker's margin requirement (usually 25%). SInce 25% margin requirement allows up to 4x leverage, which is more than any of us should possibly need, there is no risk of a margin call. If the market continued to go down in the above example, one might need to sell some VTI and de-lever in order to stay near 2x leverage, but it wasn't forced by the broker and the initial 25% market drop was not problematic.

But now let's look what would happen if we held a single ZF in the same account. Once the SMA hit 0, any negative change in the cash balance due to ZF will cause the SMA to go negative and a margin call. If the market fell ~22% the SMA would hit zero. 62k of stock with 32k of equity. The reg t requirement would be .5*62 + 1k for the ZF = 32k. The SMA is now zero. Any chance drop in cash balance due to the ZF will cause the SMA to go negative and a margin call. So while in the first example the market fell 25% we were still no where near a margin call, but holding a future in the same account, we'd get margin called around 22% market drop.

I didn't understand SMA initially and got a margin call because the SMA went negative. Portfolio eliminates this problem because there is no SMA. Obviously you still need to keep your leverage from getting out of control in a large market drop, but there is much more flexibility. With Reg T you basically get margin called at 2x leverage if you hold futures in the account, whereas if you don't hold futures in the account or have portfolio margin you won't get margin called until ~4x leverage depending on the broker. So a lot more breathing room.

The example here is very helpful:
https://www.clientam.com.hk/en/index.ph ... =overview3
Thanks for explanation, I think I should be safe in this regard as I'm not planning to take so much leverage (using margin).
Actually I have switched to Reg T from cash account to just minimize cash drag as you recommended earlier, so most of the time my cash balance would be slightly negative (far far away even from 50% margin requirement).
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

comeinvest wrote: Tue Dec 07, 2021 6:04 pm I revise my previous statement. Leveraged I-Bonds are not part of the bond portion of the asset allocation, but a pure arbitrage play. That's why usually there would be no positive return, as there is (almost) no risk - neither credit nor term risk. It's currently an anomaly, or more precisely, I-Bonds are mispriced. That's why the amount that can be purchased is limited.
Negative real yields are not all bad :sharebeer
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 »

L2F wrote: Tue Dec 07, 2021 6:06 pm
skierincolorado wrote: Tue Dec 07, 2021 4:04 pm
L2F wrote: Mon Dec 06, 2021 7:04 pm
skierincolorado wrote: Mon Dec 06, 2021 5:16 pm
Lock wrote: Sun Dec 05, 2021 5:29 pm Has anyone compared and contrasted the use of margin versus portfolio margin on IBKR? Any noticeable difference?
Reg t is a pain with futures because it really limits your leverage because the sma can go negative when the equities are anywhere close to 2x leverage.

Portfolio is better in every regard from what I understand. Most of my money is in tax advantages so i don’t qualify. Reg t you just have to be more careful. I can explain more if anyone is using reg t.
I'm using Reg T and planning to try mHFEA starting next month, so would be thankful for any advice from you.
I'm still not decided on exact leverage ratio, but most likely I will use few /ZF contracts + VTI + S&P500 LETFs.
It's easiest to run through with an example. Say you have 50k account value. You buy 80k of VTI with a 30k margin loan (refinanced with box spread). That's within the Reg T requirement of up to 50% which would have allowed you to borrow up to 50k. If the market drops 30%, you'd lose 24k of equity, and have 26k of equity remaining and 60k of stock over 2x leverage - but this would still be fine with Reg T.

Reg T has the initial 50% requirement, but also the SMA (special memorandem agreement) must stay positive.

SMA is simply the greater of either 1) equity - reg t requirement or 2) previous day SMA +/- change in cash +/- initial reg t margin requirements of new trades.

In the above example, there is no change in cash balance no matter how much VTI goes down - it's not affecting cash. The SMA would be zero once the 50% requirement is exceeded but would not go negative because the cash is never changing. So the limiting factor will become the broker's margin requirement (usually 25%). SInce 25% margin requirement allows up to 4x leverage, which is more than any of us should possibly need, there is no risk of a margin call. If the market continued to go down in the above example, one might need to sell some VTI and de-lever in order to stay near 2x leverage, but it wasn't forced by the broker and the initial 25% market drop was not problematic.

But now let's look what would happen if we held a single ZF in the same account. Once the SMA hit 0, any negative change in the cash balance due to ZF will cause the SMA to go negative and a margin call. If the market fell ~22% the SMA would hit zero. 62k of stock with 32k of equity. The reg t requirement would be .5*62 + 1k for the ZF = 32k. The SMA is now zero. Any chance drop in cash balance due to the ZF will cause the SMA to go negative and a margin call. So while in the first example the market fell 25% we were still no where near a margin call, but holding a future in the same account, we'd get margin called around 22% market drop.

I didn't understand SMA initially and got a margin call because the SMA went negative. Portfolio eliminates this problem because there is no SMA. Obviously you still need to keep your leverage from getting out of control in a large market drop, but there is much more flexibility. With Reg T you basically get margin called at 2x leverage if you hold futures in the account, whereas if you don't hold futures in the account or have portfolio margin you won't get margin called until ~4x leverage depending on the broker. So a lot more breathing room.

The example here is very helpful:
https://www.clientam.com.hk/en/index.ph ... =overview3
Thanks for explanation, I think I should be safe in this regard as I'm not planning to take so much leverage (using margin).
Actually I have switched to Reg T from cash account to just minimize cash drag as you recommended earlier, so most of the time my cash balance would be slightly negative (far far away even from 50% margin requirement).
Have you confirmed the reg t requirement on your upro is still only 50%?

Yes holding modest leverage like 1.3x in equities plus some treasury futures should be fine in reg t. I run into problems because I’m like 1.5x in that particular account (because less leverage in other accounts) so the sma is pretty low. I have to keep a close eye on it because once I hit ~1.85x the sma is zero and I get margin called (due to the futures - would be fine without the futures).
parval
Posts: 153
Joined: Tue Oct 22, 2019 9:23 pm

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

Post by parval »

millennialmillions wrote: Wed Dec 08, 2021 7:57 pm Reminder to everyone using MES futures: tomorrow is the conventional roll data for equity index futures.
How many months out should we be using for 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 »

parval wrote: Mon Dec 13, 2021 1:46 pmHow many months out should we be using for futures?
Whatever the next expiration period is. Beyond that, the volume is unfortunately too low, so you risk getting a bad spread. So you're going to need to roll over the futures 4x/year.
parval
Posts: 153
Joined: Tue Oct 22, 2019 9:23 pm

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

Post by parval »

DMoogle wrote: Mon Dec 13, 2021 1:51 pm
parval wrote: Mon Dec 13, 2021 1:46 pmHow many months out should we be using for futures?
Whatever the next expiration period is. Beyond that, the volume is unfortunately too low, so you risk getting a bad spread. So you're going to need to roll over the futures 4x/year.
Thanks, this will force me to rebalance more than the ETFs :?

Somewhat related is if there's a way to optimize rebalancing, I skimmed this paper https://deliverypdf.ssrn.com/delivery.p ... INDEX=TRUE

And my main take away is if we have some prediction of future volatility (which is suppose to be more predictable than returns), we can skew the proportions when we rebalance, wdyt?

Otherwise I might just do ~33/33/33 es/zn/zf to keep it simple
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

parval wrote: Tue Dec 14, 2021 1:55 pm
DMoogle wrote: Mon Dec 13, 2021 1:51 pm
parval wrote: Mon Dec 13, 2021 1:46 pmHow many months out should we be using for futures?
Whatever the next expiration period is. Beyond that, the volume is unfortunately too low, so you risk getting a bad spread. So you're going to need to roll over the futures 4x/year.
Thanks, this will force me to rebalance more than the ETFs :?

Somewhat related is if there's a way to optimize rebalancing, I skimmed this paper https://deliverypdf.ssrn.com/delivery.p ... INDEX=TRUE

And my main take away is if we have some prediction of future volatility (which is suppose to be more predictable than returns), we can skew the proportions when we rebalance, wdyt?

Otherwise I might just do ~33/33/33 es/zn/zf to keep it simple
That's an interesting paper. There are a lot of results, which are you referring to?

The paper considers the case of rebalancing between two risky assets. If the mean returns are equal, rebalanced minimum variance portfolio is best. I understand this as the arithmetic mean is independent to the weights, so you would want to minimize portfolio variance to maximize the portfolio geometric mean towards the upper limit of the portfolio arithmetic mean. If the asset mean returns are unequal, there is a penalty to rebalance into the lower returning asset.

I'm not sure if any of the results are immediately relevant here because it is not sensible to assume stocks and bonds (let alone ITTs) have the same mean return. Furthermore, we are rebalancing between two risky assets and one risk-free asset (i.e., rebalancing between stocks and bonds as well as resetting the leverage ratio), not just between two risky assets. I think it could be shown that if you assume stocks and ITTs have the same Sharpe, then (two asset) rebalanced risk parity would be the mean variance optimal portfolio and to max geometric returns, I think it would be rebalanced risk parity leveraged to the Kelly criterion.

I've thought about using short-term historical volatility or option implied volatility when rebalancing, but I haven't been convinced yet. My concern is that real-world results are often dominated by unexpected events, so you could get lulled into a false sense of security and be over-exposed to stocks or bonds in a period of low volatility and get wiped out. Plus there is so much work on volatility and its relationships I have not had a chance to read... And many other areas (real job getting in the way)...
parval
Posts: 153
Joined: Tue Oct 22, 2019 9:23 pm

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

Post by parval »

O the paper goes through rebalancing across any set of assets.

The Shannon-Demon example is balancing risky asset w/ cash: they assert the optimal weight (of risky asset) = mean / variance^2, this is equation 7. They also conclude selecting the max Sharp asset for this.

The Parrondo’s Paradox shows the non-intuitive case of 2 risky assets that are both -ev but through optimal rebalancing can still grow wealth.

The end of the paper goes through the more complex case for multi-asset portfolio, my eyes kinda glazed over at this point.

For me, the main thing it shows is theoretically how powerful rebalancing can be. And my conclusion (could 100% be wrong) is to adjust the weights in proportion to expected volatility.
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

parval wrote: Tue Dec 14, 2021 7:48 pm O the paper goes through rebalancing across any set of assets.

The Shannon-Demon example is balancing risky asset w/ cash: they assert the optimal weight (of risky asset) = mean / variance^2, this is equation 7. They also conclude selecting the max Sharp asset for this.

The Parrondo’s Paradox shows the non-intuitive case of 2 risky assets that are both -ev but through optimal rebalancing can still grow wealth.

The end of the paper goes through the more complex case for multi-asset portfolio, my eyes kinda glazed over at this point.

For me, the main thing it shows is theoretically how powerful rebalancing can be. And my conclusion (could 100% be wrong) is to adjust the weights in proportion to expected volatility.
To relate your comments to my earlier post by translating some of the words: equation 7 is the Kelly criterion under their set of assumptions. A more general expression without assuming zero risk-free rate is (M-Rf)/SD^2, where Rf is the borrowing rate and SD is standard deviation (not variance). I think you would still need a positive expected (arithmetic) return or mean reversion to have benefits from rebalancing. Usually the rebalancing effect works by decreasing the volatility of the portfolio, which decreases the volatility drag on geometric or compounding returns. Inverse volatility weighting is mean-variance optimal (max Sharpe) if the assets have the same Sharpe and all the pairwise correlations are the same (or there are only two assets). I must have missed it if they talked about the assets having equal Sharpe in the paper, I thought they only talked about equal means leading to minimum variance portfolio which I don't think this relevant where the assets are stocks and bonds.

I agree that rebalancing is important. If you have a weighting that is optimal in some sense, the longer you go between rebalancing the more you can deviate from the optimal weighting which can increase volatility of the portfolio -> create volatility drag -> reduce compounding returns as in the paper.
Hydromod
Posts: 1048
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod »

I use a type of minimum variance approach with LETFs. I use the previous 2 months for volatility and previous 3 months for correlations, but I don't use returns. Instead I use a risk budget approach, which assigns a fraction of the portfolio volatility to each asset. In practice, I typically set the risk budget for the equities to be 3 times the risk budget for long-term treasuries, which would have worked pretty well since 1986. This would have averaged about 55/45 for UPRO/TMF, but with swings between 80/20 and 20/80. The risk ratio needs to be much larger for TYD instead of TMF.

I find that the adaptive allocations are increasingly effective as the leverage goes up.
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

Hydromod wrote: Tue Dec 14, 2021 9:28 pm I use a type of minimum variance approach with LETFs. I use the previous 2 months for volatility and previous 3 months for correlations, but I don't use returns. Instead I use a risk budget approach, which assigns a fraction of the portfolio volatility to each asset. In practice, I typically set the risk budget for the equities to be 3 times the risk budget for long-term treasuries, which would have worked pretty well since 1986. This would have averaged about 55/45 for UPRO/TMF, but with swings between 80/20 and 20/80. The risk ratio needs to be much larger for TYD instead of TMF.

I find that the adaptive allocations are increasingly effective as the leverage goes up.
Why not use implied volatility from option prices (e.g., for options on SPX/SPX and TLT)? Do you use the volatility and correlation estimates to set the leverage on the whole portfolio (targeting a level of volatility for the whole portfolio), or just the proportion of URPO vs TMF?
comeinvest
Posts: 2669
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 »

zkn wrote: Tue Dec 14, 2021 9:45 pm
Hydromod wrote: Tue Dec 14, 2021 9:28 pm I use a type of minimum variance approach with LETFs. I use the previous 2 months for volatility and previous 3 months for correlations, but I don't use returns. Instead I use a risk budget approach, which assigns a fraction of the portfolio volatility to each asset. In practice, I typically set the risk budget for the equities to be 3 times the risk budget for long-term treasuries, which would have worked pretty well since 1986. This would have averaged about 55/45 for UPRO/TMF, but with swings between 80/20 and 20/80. The risk ratio needs to be much larger for TYD instead of TMF.

I find that the adaptive allocations are increasingly effective as the leverage goes up.
Why not use implied volatility from option prices (e.g., for options on SPX/SPX and TLT)? Do you use the volatility and correlation estimates to set the leverage on the whole portfolio (targeting a level of volatility for the whole portfolio), or just the proportion of URPO vs TMF?
I didn't read the paper cited above, but you would have a hard time convincing me that volatility during some past period should determine the asset allocation of a future period, where the markets are unknown by definition, where market crashes or stagflation can happen or not.

Something unrelated. Parval mentioned 33/33/33 es/zn/zf. So assuming I want constant exposure to interest rate changes via treasury futures. I personally would take the duration into account, for example by normalizing it to 25 year duration (or any other duration). The duration of a specific futures symbol can change from quarter to quarter based on the CTD, which is an arbitrary artifact to me, as I want to have constant exposure to interest rate changes. Since the allocation of the portfolio to the treasury futures is "choppy" because of the contract sizes, I would also look to keep the total duration exposure as percentage of NAV constant, not the total equity per NAV allocated to the futures in total, when adding or removing funds to/from the individual contracts on rebalancing. So basically I would specify units of duration as a percentage of NAV in my asset allocation, not asset value allocated to treasuries.

Now, when the interest rate regime changes across the curve, like the level of interest rates is much higher at some point than it is now. The durations of the futures products will decrease, I think. Should I adjust for that effect in my asset allocation over time, or not, i.e. keep the total duration (per NAV) or the allocated capital (per NAV) constant? And how would I technically separate that effect from duration changes of futures contracts based on CTD changes. I have not made up my mind.

Comments are welcome.
Hydromod
Posts: 1048
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod »

comeinvest wrote: Wed Dec 15, 2021 3:41 am
zkn wrote: Tue Dec 14, 2021 9:45 pm
Hydromod wrote: Tue Dec 14, 2021 9:28 pm I use a type of minimum variance approach with LETFs. I use the previous 2 months for volatility and previous 3 months for correlations, but I don't use returns. Instead I use a risk budget approach, which assigns a fraction of the portfolio volatility to each asset. In practice, I typically set the risk budget for the equities to be 3 times the risk budget for long-term treasuries, which would have worked pretty well since 1986. This would have averaged about 55/45 for UPRO/TMF, but with swings between 80/20 and 20/80. The risk ratio needs to be much larger for TYD instead of TMF.

I find that the adaptive allocations are increasingly effective as the leverage goes up.
Why not use implied volatility from option prices (e.g., for options on SPX/SPX and TLT)? Do you use the volatility and correlation estimates to set the leverage on the whole portfolio (targeting a level of volatility for the whole portfolio), or just the proportion of URPO vs TMF?

In this case, I'm running four or five 3x LETFs together as a portfolio in two smallish siloed accounts; I'm under my desired overall leverage, but cannot transfer funds from my main portfolio into accounts that have access to leverage of any sort. So these are being treated as a silo until they grow enough to worry about rebalancing the overall portfolio.

I run the minimum variance on all of the assets simultaneously to capture the correlations properly.
I didn't read the paper cited above, but you would have a hard time convincing me that volatility during some past period should determine the asset allocation of a future period, where the markets are unknown by definition, where market crashes or stagflation can happen or not.

The key here is the time scale between rebalancing.

The volatility clustering that I rely on is typically statistically significant (at a level I am content is actionable for me) for future periods of at least several weeks, based on my backtesting. I update weekly, and decide on rebalancing based on a bands approach. I'm prepared to act each week in Roth given dramatic swings, although I anticipate that rebalancing within a week will be quite rarely necessary. In taxable, I'm not going to rebalance more frequently than 30 days and will be targeting six weeks once the portfolio assets are all long term.

I agree that projecting more than this time scale is not warranted; even six weeks is really pushing the boundaries, although does fine in backtesting. The nice thing about frequent updating, though, is that calculated allocations are continually updated, which tends to be self correcting.
comeinvest
Posts: 2669
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 »

Hydromod wrote: Wed Dec 15, 2021 8:04 am
comeinvest wrote: Wed Dec 15, 2021 3:41 am
zkn wrote: Tue Dec 14, 2021 9:45 pm
Hydromod wrote: Tue Dec 14, 2021 9:28 pm I use a type of minimum variance approach with LETFs. I use the previous 2 months for volatility and previous 3 months for correlations, but I don't use returns. Instead I use a risk budget approach, which assigns a fraction of the portfolio volatility to each asset. In practice, I typically set the risk budget for the equities to be 3 times the risk budget for long-term treasuries, which would have worked pretty well since 1986. This would have averaged about 55/45 for UPRO/TMF, but with swings between 80/20 and 20/80. The risk ratio needs to be much larger for TYD instead of TMF.

I find that the adaptive allocations are increasingly effective as the leverage goes up.
Why not use implied volatility from option prices (e.g., for options on SPX/SPX and TLT)? Do you use the volatility and correlation estimates to set the leverage on the whole portfolio (targeting a level of volatility for the whole portfolio), or just the proportion of URPO vs TMF?

In this case, I'm running four or five 3x LETFs together as a portfolio in two smallish siloed accounts; I'm under my desired overall leverage, but cannot transfer funds from my main portfolio into accounts that have access to leverage of any sort. So these are being treated as a silo until they grow enough to worry about rebalancing the overall portfolio.

I run the minimum variance on all of the assets simultaneously to capture the correlations properly.
I didn't read the paper cited above, but you would have a hard time convincing me that volatility during some past period should determine the asset allocation of a future period, where the markets are unknown by definition, where market crashes or stagflation can happen or not.

The key here is the time scale between rebalancing.

The volatility clustering that I rely on is typically statistically significant (at a level I am content is actionable for me) for future periods of at least several weeks, based on my backtesting. I update weekly, and decide on rebalancing based on a bands approach. I'm prepared to act each week in Roth given dramatic swings, although I anticipate that rebalancing within a week will be quite rarely necessary. In taxable, I'm not going to rebalance more frequently than 30 days and will be targeting six weeks once the portfolio assets are all long term.

I agree that projecting more than this time scale is not warranted; even six weeks is really pushing the boundaries, although does fine in backtesting. The nice thing about frequent updating, though, is that calculated allocations are continually updated, which tends to be self correcting.
So per your research and analysis, what is the all-in excess return and/or risk from doing all that, vs. a fixed percentage asset allocation rebalanced quarterly or with rebalancing bands, and what was your backtesting timeframe? Possible tail risks unaccounted for and not showing in the backtests, etc.? It seems to significantly add operational complexity, especially for people how have multiple equity subdivisions like international, small cap value, etc. Please convince me reading the paper ;)
Hydromod
Posts: 1048
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod »

comeinvest wrote: Wed Dec 15, 2021 7:23 pm So per your research and analysis, what is the all-in excess return and/or risk from doing all that, vs. a fixed percentage asset allocation rebalanced quarterly or with rebalancing bands, and what was your backtesting timeframe? Possible tail risks unaccounted for and not showing in the backtests, etc.? It seems to significantly add operational complexity, especially for people how have multiple equity subdivisions like international, small cap value, etc. Please convince me reading the paper ;)
I haven't looked at the paper, so I can't speak to that.

I don't recommend moving away from a fixed allocation and fixed rebalance period unless you have the temperament and desire to track the portfolio fairly frequently (e.g., once every week or two). I also find that it is almost certainly not worth the effort for 1x or even 2x portfolios.

When I do backtesting, I typically use 1986 - present or 1991 - present, depending on the available assets. Trades account for slippage and random timing luck during the day. The synthetic parts of the funds use the same methodology as Siamond, but I think that they still provide optimistic returns prior to 2010.

As an example, I ran a UPRO/TQQQ/TMF portfolio from 1986 to present, assigning a risk budget of 37.5/37.5/25 and with monthly rebalancing (equities get 3/4 of the risk budget). That ended up as a time-averaged allocation of 30/22/48 UPRO/TQQQ/TMF, then ran the same funds with a fixed allocation and quarterly rebalancing. Rebalancing was NOT aligned to months or quarters.

Overall CAGR was only nominally better with the adaptive allocation, and overall volatility dropped about 8 percent. Sharpe ratio increased by about 10 percent. Most of the outperformance was associated with 2000 - 2002.

So incremental improvements, providing that you already know the allocation to aim for.

The thing that I like is that I don't have to guess about the long-term allocations for the next 10 or 20 years, where the markets are unknown by definition, where market crashes or stagflation can happen or not; the allocation continually adjusts to recent market behavior.

I also like the emphasis on reducing portfolio volatility, which will be important for increasing safe withdrawal rates during decumulation.

You pays your money and you takes your choice.
comeinvest
Posts: 2669
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 »

Hydromod wrote: Wed Dec 15, 2021 10:03 pm
comeinvest wrote: Wed Dec 15, 2021 7:23 pm So per your research and analysis, what is the all-in excess return and/or risk from doing all that, vs. a fixed percentage asset allocation rebalanced quarterly or with rebalancing bands, and what was your backtesting timeframe? Possible tail risks unaccounted for and not showing in the backtests, etc.? It seems to significantly add operational complexity, especially for people how have multiple equity subdivisions like international, small cap value, etc. Please convince me reading the paper ;)
I haven't looked at the paper, so I can't speak to that.

I don't recommend moving away from a fixed allocation and fixed rebalance period unless you have the temperament and desire to track the portfolio fairly frequently (e.g., once every week or two). I also find that it is almost certainly not worth the effort for 1x or even 2x portfolios.

When I do backtesting, I typically use 1986 - present or 1991 - present, depending on the available assets. Trades account for slippage and random timing luck during the day. The synthetic parts of the funds use the same methodology as Siamond, but I think that they still provide optimistic returns prior to 2010.

As an example, I ran a UPRO/TQQQ/TMF portfolio from 1986 to present, assigning a risk budget of 37.5/37.5/25 and with monthly rebalancing (equities get 3/4 of the risk budget). That ended up as a time-averaged allocation of 30/22/48 UPRO/TQQQ/TMF, then ran the same funds with a fixed allocation and quarterly rebalancing. Rebalancing was NOT aligned to months or quarters.

Overall CAGR was only nominally better with the adaptive allocation, and overall volatility dropped about 8 percent. Sharpe ratio increased by about 10 percent. Most of the outperformance was associated with 2000 - 2002.

So incremental improvements, providing that you already know the allocation to aim for.

The thing that I like is that I don't have to guess about the long-term allocations for the next 10 or 20 years, where the markets are unknown by definition, where market crashes or stagflation can happen or not; the allocation continually adjusts to recent market behavior.

I also like the emphasis on reducing portfolio volatility, which will be important for increasing safe withdrawal rates during decumulation.

You pays your money and you takes your choice.
Thanks for sharing! Interesting, but probably too engineered and unpredictable for me. Sorry if I missed it, but I think your strategy makes no assumptions of expected returns. I personally probably feel safer just reducing the treasuries when interest rates approach zero or below, or when they significantly deviate from the Fed target rate in either direction.
Hydromod
Posts: 1048
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod »

comeinvest wrote: Wed Dec 15, 2021 10:21 pm Thanks for sharing! Interesting, but probably too engineered and unpredictable for me. Sorry if I missed it, but I think your strategy makes no assumptions of expected returns. I personally probably feel safer just reducing the treasuries when interest rates approach zero or below, or when they significantly deviate from the Fed target rate in either direction.
That's the temperament thing in investing. I'm an engineer by training...

Not quite right on assumptions of expected returns. The assumption is that expected returns are independent of volatility.

I did some backtesting 1986 to present, breaking the daily sequence into every possible segment of a fixed duration (e.g., a month), calculating the volatility for the segment and the returns for the next week, two weeks, or month, and binning the volatility/future return pairs by volatility.

It turns out that the mean future returns were very similar across all bins.

This is liberating, because it is very hard to estimate returns.

Estimated returns can be tacitly included by making the risk budget unequal, however. I tend to assign greater risk budge to the equity portion than to the treasury portion, typically 3 to 1 or so, mainly to increase returns.
parval
Posts: 153
Joined: Tue Oct 22, 2019 9:23 pm

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

Post by parval »

Hydromod wrote: Thu Dec 16, 2021 7:33 am
comeinvest wrote: Wed Dec 15, 2021 10:21 pm Thanks for sharing! Interesting, but probably too engineered and unpredictable for me. Sorry if I missed it, but I think your strategy makes no assumptions of expected returns. I personally probably feel safer just reducing the treasuries when interest rates approach zero or below, or when they significantly deviate from the Fed target rate in either direction.
That's the temperament thing in investing. I'm an engineer by training...

Not quite right on assumptions of expected returns. The assumption is that expected returns are independent of volatility.

I did some backtesting 1986 to present, breaking the daily sequence into every possible segment of a fixed duration (e.g., a month), calculating the volatility for the segment and the returns for the next week, two weeks, or month, and binning the volatility/future return pairs by volatility.

It turns out that the mean future returns were very similar across all bins.

This is liberating, because it is very hard to estimate returns.

Estimated returns can be tacitly included by making the risk budget unequal, however. I tend to assign greater risk budge to the equity portion than to the treasury portion, typically 3 to 1 or so, mainly to increase returns.
Sorry if I'm mixing concepts, but does that assumption related to actual returns and realized volatility?

I thought especially w/ leverage (but even without), as realized volatility increases, actual returns decreases due to decay, and the two are not iid.
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

parval wrote: Thu Dec 16, 2021 8:52 am
Hydromod wrote: Thu Dec 16, 2021 7:33 am
comeinvest wrote: Wed Dec 15, 2021 10:21 pm Thanks for sharing! Interesting, but probably too engineered and unpredictable for me. Sorry if I missed it, but I think your strategy makes no assumptions of expected returns. I personally probably feel safer just reducing the treasuries when interest rates approach zero or below, or when they significantly deviate from the Fed target rate in either direction.
That's the temperament thing in investing. I'm an engineer by training...

Not quite right on assumptions of expected returns. The assumption is that expected returns are independent of volatility.

I did some backtesting 1986 to present, breaking the daily sequence into every possible segment of a fixed duration (e.g., a month), calculating the volatility for the segment and the returns for the next week, two weeks, or month, and binning the volatility/future return pairs by volatility.

It turns out that the mean future returns were very similar across all bins.

This is liberating, because it is very hard to estimate returns.

Estimated returns can be tacitly included by making the risk budget unequal, however. I tend to assign greater risk budge to the equity portion than to the treasury portion, typically 3 to 1 or so, mainly to increase returns.
Sorry if I'm mixing concepts, but does that assumption related to actual returns and realized volatility?

I thought especially w/ leverage (but even without), as realized volatility increases, actual returns decreases due to decay, and the two are not iid.
Hydromod is talking about arithmetic mean returns, you are talking about geometric mean returns. If volatility is predictable but mean returns are not, predicted geometric returns decrease with increased predicted volatility.
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

At the risk of meandering this thread in yet another direction: one issue that seems critical is the sensitivity of the portfolio risk to the stock-bond correlation going positive. Portfolio volatility is a function of asset volatility and the correlation between assets, portfolio max-drawdown is a function of asset large drawdowns and their propensity to co-occur. So an increase in stock-bond correlation implies an increase in the riskiness of the portfolio.

My thinking is that since the correlation is dynamic (varies over time), we should evaluate the maximum risk of the portfolio depending on the maximum plausible correlation (i.e., positive), not the recent (negative) or average (zero) correlation. So if we construct the portfolio by maximizing Sharpe then leveraging to a desired maximum risk, correlation is relevant to both steps (higher correlation -> lower Sharpe; higher max correlation -> lower leverage and hence lower returns).

Correlations are standardized measures of co-movement. The best conceptual framework I am aware of to explain short-term (co)movements in investments are the macro factors of growth and inflation expectations (and volatility of growth and inflation expectations). Changes in inflation expectations move stock and bond prices in the same direction so the stock-bond correlation will increase in times of inflation uncertainty and volatility. In this framework we can see that stock-bond portfolios are optimal for the recent past (low growth, low inflation, low volatility), but not for other periods (e.g., high inflation volatility of the 1970s).

Modifications to a stock-bond portfolio to reduce the average and maximum stock-bond correlation or off-set this covariance can be categorized as three types:

First, reduce the sensitivity of the stock part to inflation expectations (i.e., reduce disinflation beta):
  • Global diversification (not US-only). This should help to the extent inflation is not perfectly positively correlated across countries.
  • Small-value tilt. Value is a) slightly more pro-inflationary vs growth (at least, it overweighs pro-inflation sectors like energy); b) shorter duration so less embedded rate sensitivity; and c) to the extent value risk reflects greater economic growth risk, this is "good" risk because it can be hedged with bonds (in contrast to inflation risk).
  • EM tilt. EM may do better with high inflation (commodity exporters), and high growth (which is relevant to the extent high growth co-occurs with high inflation). This effect may be stronger if EM stocks are not currency-hedged back to USD due to the EM currency exposure.
Second, reduce the sensitivity of the bond part to inflation expectations. To the extent bond yields reflect growth and inflation expectations, nominal bonds have an embedded short position on inflation. Certainly back tests show bonds are very negatively affected by unexpected inflation and more than stocks.
  • Tilt shorter duration. See this graph from the book Expected Returns.
  • Tilt inflation-linked. Kind of obvious: reduce inflation sensitivity by removing the embedded short inflation position in nominal bonds by replacing them with TIPS. Unfortunately there are less ways to access TIPS than nominal treasuries. Leveraging ITT TIPS ETFs with options or margin is expensive. Long duration TIPS with LTPZ may be better, but ER is a little high (.20%). Note the play here is borrowing at nominal rates to invest in something that moves with inflation; inflation inflates away the debt while the asset maintains real value.
  • Global diversification. I'd like to do this but managing futures in different currencies and exchanges seems difficult and maybe inefficient. Australian bonds are a good choice (AAA rated, commodity currency -> positive inflation exposure, Australian housing bubble may imply a limit on how high Australian rates can go).
  • EM bonds. We have three flavors: EM USD-denominated; EM local currency denominated, currency hedged; EM local currency denominated, currency unhedged. The third is probably the most effective inflation hedge. Great in theory but has a large downside relative to upside (i.e., they crash hard). Also a little expensive (my preferred ETF, LEMB, has ER = .30%). Note EM bonds effectively have EM equity exposure so contribute to stock exposure too.
  • Underweight bonds. This seems to be the approach a lot of people take after observing the challenge with inflation in back tests.
Third, add an asset that moves positively with inflation expectations to counteract negative inflation exposure in stocks and bonds. Let's call them "alts". These are way too complex to summarize or evaluate concisely, but just to note those that seem more promising:
  • Commodities. According to some, an inflation hedge with positive risk premium. Kind of "too good to be true" but the third leg in many risk-parity portfolios.
  • Trend following/time series momentum. Maybe another risk factor (providing general diversification benefits) with "crisis alpha".
  • Gold. Maybe the "flee the monetary system asset" during periods of inflation uncertainty.
  • EM currency/currency carry trade. There is a premium in investing in EM/high interest rate currencies, but there should not be one under no arbitrage and the trade tends to aggressively reverse during stock market crashes. AUD, CAD, NZD, KRW could be relevant here too, and other currencies besides USD on the short side of the trade.
  • I-bonds. I-bonds move with CPI and provide a positive return with very little risk (funded with box spreads). See recent posts in this thread.
  • Real estate.
It might not be a good idea to completely neutralize your portfolio to inflation, especially for young people. First, being modestly short inflation helps hedge against deflationary crashes. Much of the hedging capabilities of nominal bonds exhibited in back tests is actually coming from being short inflation during a deflationary shock (in addition to being short real rates and nominal treasuries being "the safe asset"). A portfolio less negatively affected by inflation would probably have crashed harder in the GFC because it would be less positively affected by deflation. Second, having a modestly short-inflation portfolio balances a long-inflation human capital. I invest partly to hedge against governments and central banks continuing to kick the can down the road towards Japanification, lower real rates and capital over labor that would reduce opportunities to make money. If some type of inflation and growth spurt knocks us off that path, there could be other opportunities and human capital could be worth more. But I will tilt my portfolio towards disinflation in case, as most likely will be the case, we continue along the current disinflationary road to even lower real rates.

Looking at what the professionals do, AQR's multi-asset fund seems the most transparent. They overweight (global developed) stocks over bonds (43% of risk allocation versus 23.5% of nominal bonds), assign just 4.4% to global inflation-linked bonds, 22.3% to commodities, and 6.6% to EM currency with forward contracts. Gold seems to be treated just like any other commodity and EM debt and equity is not present. Bridgewater is less transparent, but also have commodities. In contrast to AQR, Bridgewater seem to put more emphasis on inflation-linked bonds, gold, EM debt, and EM equity (at least China).

My stocks are globally diversified, small-value tilt (especially in US), slight EM overweight. Almost all my bonds are US treasuries except a small amount in EM bonds. I don't have separate currency positions but I don't currency hedge; carry trade seems fair so I don't want to pay to try to be on one or the other side of it and let my stock and bond allocations decide my currency positions. I've considered taking about a third of my bond risk in TIPS but I would not take more to avoid neutralizing the portfolio to inflation or making it positively related to inflation. I-bonds is by far the best alt but is limited by amount and cannot be used as collateral in your brokerage account. I am reluctant with the risk premium story for commodities, trend following, and gold, and think readily available funds for the first two are poorly constructed or too expensive. I think it would be useful to have at least some commodities (maybe 5-10%... not 20% of risk) but I am still looking for a good fund. Ultimately I think reasonable tilts in the stock-bond portfolio plus I-bonds should be the first step, and to the extent more protection is required, replacing some nominal treasuries with TIPS is the next step. Some alts (commodities in particular) could make sense but I do not have the conviction to dedicate more than a little risk to them especially when they cannot be accessed cheaply.

Thoughts/comments/criticisms? Any asset I have missed or something I overlooked?
Last edited by zkn on Sat Dec 18, 2021 9:26 pm, edited 2 times in total.
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 »

zkn wrote: Sat Dec 18, 2021 3:16 pmThoughts/comments/criticisms? Any asset I have missed or something I overlooked?
It's a really good post, zkn. Honestly, all the points you made are just as relevant to a non-leveraged portfolio as they are a leveraged portfolio. They just become that much more important with a leveraged portfolio.

One thing I want to touch on: near the start of your post you talk about how correlations are dynamic and are the standardized measures of co-movement. Both true. However, there's one big flaw with correlations that I think is extremely often overlooked: expected returns aside, correlations only really matter during times of high volatility, yet they're calculated using all data points (in and out of crisis). Due to the dynamic nature of them, I think using out-of-crisis correlations is a big mistake.

Put another way, if the market is stable and performing well overall (say, returning their expected value), then I don't care if my assets have a correlation of 1, -1, or 0. Doesn't matter, because everything is stable. However, once that period ends, then that's when the risk management becomes relevant. Moreover, the dynamic nature of those correlations may mean that the in-crisis correlation may be different from the out-of-crisis correlation.

In another thread, someone mentioned "in a crisis, all correlations go to 1." Well, that's pretty patently false, but we need to understand the risks that can make something like that happen between, say, ITTs and stocks.
zkn
Posts: 67
Joined: Thu Oct 14, 2021 12:45 pm

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

Post by zkn »

DMoogle wrote: Sat Dec 18, 2021 6:41 pm
zkn wrote: Sat Dec 18, 2021 3:16 pmThoughts/comments/criticisms? Any asset I have missed or something I overlooked?
It's a really good post, zkn. Honestly, all the points you made are just as relevant to a non-leveraged portfolio as they are a leveraged portfolio. They just become that much more important with a leveraged portfolio.

One thing I want to touch on: near the start of your post you talk about how correlations are dynamic and are the standardized measures of co-movement. Both true. However, there's one big flaw with correlations that I think is extremely often overlooked: expected returns aside, correlations only really matter during times of high volatility, yet they're calculated using all data points (in and out of crisis). Due to the dynamic nature of them, I think using out-of-crisis correlations is a big mistake.

Put another way, if the market is stable and performing well overall (say, returning their expected value), then I don't care if my assets have a correlation of 1, -1, or 0. Doesn't matter, because everything is stable. However, once that period ends, then that's when the risk management becomes relevant. Moreover, the dynamic nature of those correlations may mean that the in-crisis correlation may be different from the out-of-crisis correlation.

In another thread, someone mentioned "in a crisis, all correlations go to 1." Well, that's pretty patently false, but we need to understand the risks that can make something like that happen between, say, ITTs and stocks.
Thanks. Good point on the correlations. I was trying to express a similar idea with reference to maximum correlation, but correlation during periods of high asset volatility/risk/drawdowns that define the maximum portfolio volatility/risk/drawdown is a better way to express the idea.
User avatar
hillclimber
Posts: 136
Joined: Wed Mar 10, 2021 2:32 pm

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

Post by hillclimber »

zkn wrote: Sat Dec 18, 2021 3:16 pm Thoughts/comments/criticisms? Any asset I have missed or something I overlooked?
That's a good overview of the situation we're in. I've been researching alts recently and haven't been impressed. With stocks, managers have to compete to earn alpha, while commodity managers have to fight just to earn a positive return. Even if the theory behind them works, the fees will eat you alive. I think an interesting case study is FUT, the ProShares Managed Futures Strategy ETF. It tracks the S&P Strategic Futures Index, which has returned 2.5% annualized since 2016 (The S&P site is acting weird, so here's a wayback version). FUT has returned 0.6% annualized. Some of that's because of the 0.76% expense ratio, but I think the rest is implementation issues.
If you just look at the index, the returns look good, but I think there are a lot of issues in implementing managed futures strategies.
Phyneas
Posts: 336
Joined: Tue Apr 27, 2021 9:10 pm

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

Post by Phyneas »

skierincolorado wrote: Fri Sep 03, 2021 10:27 pm https://www.portfoliovisualizer.com/bac ... on4_3=-465

Personally, HFEA is too much risk for me. But can we achieve the returns of HFEA with lower risk by diversifying across higher returning assets like ITT instead of LTT? A modified HFEA strategy using futures and ITT is essentially an optimization and fusion of several important threads on this forum. Lifecycle investing is about finding an optimal leverage glidepath over the course of one's life to diversify one's investments across time. HFEA diversifies across asset classes. While HFEA used TMF long-term treasuries for the bond allocation, this was because leveraging shorter durations requires more leverage than a 3x LETF can provide. Historically 1960-2018 the diversification benefit of ITT was higher, and the return per unit of risk is much greater. We just need to own more to achieve the same level of risk/variance and diversification as TMF. If we already have our whole portfolio allocated to UPRO and TMF, we can't achieve any more leverage without using futures contracts. Neither HFEA, nor the Lifecycle Investing book's 100% equity AA, have been on the efficient frontier historically, nor are they likely to ever be.

...

ITT are an improvement over LTT:
1) A portfolio of ITT+stocks has been less dependent on the performance of stocks than a portfolio of LTT+stocks. The benefit from diversification has been better. See exhibit 17 here: https://www.pimco.com/handlers/displayd ... d%2BfxA%3D
2) ITT have much better risk adjusted returns historically.
3) The ITT market is much larger than the LTT market, and owning ITT puts us much closer to market weight across asset classes

...

Now for the backtest. HFEA is a 165/135 stock/LTT AA. We can have a much less risky AA of 125/270 stock/ITT and achieve the same return since 1991. The max-drawdown is reduced from 66.6% to 47.8%. The 270% allocation to ITT is substantially less risky than the 135% allocation to LTT. The ITT in the backtest have a duration of 5 years, and the LTT have a duration of 20 years, making the max-drawdown in a bond crash nearly 4x larger for LTT. The duration risk of LTT is 4x ITT. Instead of replacing the LTT in HFEA with 4:1 ITT, we could do a modest 2:1 and still get the same returns as HFEA, with much less risk. We can even reduce the stock leverage from 165% to 125%, and still get the same return as HFEA. If we wanted to take the same level of risk as HFEA, we would do an AA of 165/400 stock/ITT. Since 1991, this has actually been much less risky, with a max-drawdown of 59.2% (vs 66.6% for HFEA).
t's a fascinating idea; I've been interested in using leverage for a while now and I agree about ITTs being the better long-term hold than LTTs for this kind of thing, especially during periods of economic stress (STTs actually being the best). However, the issue I've come across is that unless you are comfortable using futures (I'm not), you're essentially stuck with the 2x and 3x ITT LETFs that have tiny AUMs.

I've tried approximating leveraged ITT performance with short and long-term leveraged treasuries, but it doesn't seem to work as it becomes very AA and start and end-date dependent. Is there any way around this? For instance, if you wanted to build a 2x portfolio but wanted control over the AA and didn't want LTTs or corporates (like PSLDX offers), are you either going to have to use futures, or a low-AUM ITT ETF?
60% AVGE | 20 Year TIPS LMP | 5% Cash
parval
Posts: 153
Joined: Tue Oct 22, 2019 9:23 pm

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

Post by parval »

Yeah my one takeaways from researching leverage this past year is that simple leverage via LETFs is expensive, and one of the best ways to reduce that cost is via futures.

I'm also very uncomfortable w/ futures, since there's liquidation risk compared to LETFs, so I'm trying to read up more on ES/ZN/ZF, but honestly I just want a tool to sim liquidation %s.

Another route is via options (LEAPs) if you want to look into that. The main idea is if you want 2x leverage, just go w/ ITM leaps with 50% strike, I'm not sure how good the liquidity is for ITT options though.
bling
Posts: 1452
Joined: Sat Jan 21, 2012 11:49 am

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

Post by bling »

parval wrote: Sun Dec 19, 2021 9:34 am Yeah my one takeaways from researching leverage this past year is that simple leverage via LETFs is expensive, and one of the best ways to reduce that cost is via futures.

I'm also very uncomfortable w/ futures, since there's liquidation risk compared to LETFs, so I'm trying to read up more on ES/ZN/ZF, but honestly I just want a tool to sim liquidation %s.

Another route is via options (LEAPs) if you want to look into that. The main idea is if you want 2x leverage, just go w/ ITM leaps with 50% strike, I'm not sure how good the liquidity is for ITT options though.
i have deep ITMs LEAPs in my Roth IRA. you would think that liquidity is bad given how far into the future these options are, but i've also never had a problem getting them filled at the mid price. SPY's spread is actually really good, even 2 years out the spread is less than 1% of the current price. unfortunately, international indices are not as good...the spread for EFA is a whopping 6%.
Post Reply