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.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

skierincolorado wrote: Tue Jul 19, 2022 9:39 am
LazyOverthinker wrote: Tue Jul 19, 2022 3:20 am
skierincolorado wrote: Mon Jul 18, 2022 8:02 pm The market is basically betting that yields will fall in the medium term and that the 5 year will benefit more than the 2 year from this.
Ok, that makes sense. Does that same market-logic apply if we replace our 5-year treasuries with 60% more "duration-matched" 2-year treasuries?

In other words: even if we agree with the market's bet on 5-year rates, are we properly compensated for giving up the chance to hold 60% more-total "0.65% carry" treasuries?

Essentially I'm wondering if leveraged investors have a "bet against beta" advantage by being able to adapt total $ exposure, or if that is captured in the market's foresight.
That should be captured. For example let's say the equilibrium rate is 2% and the market is betting we will rall to that rate in 2 years. Let's say the FF rate is 1.5% the whole time for simplicity. Would we earn more by holding 100k of 5 year bonds with initial rate of 3% or 250k of 2 year bonds with initial yield of 2.5%? With the 2 year bonds we net 1% of 250k or 2.5k. With the 5 year bonds we net 1.5% of 100k, or 1.5k. But we also would see a capital appreciation of our 5 year bonds of 3k for total profit of 4.5k.

This example illustrates how the bond with higher carry per unit duration can easily have lower returns. Of course if this were actually the market expectation it would easily be arbitraged out of existence and would not happen in the first place.

Bet against beta is separate from this. In an efficient market there would be no advantage to investing in shorter durations with highe carry per unit risk even with leverage because the expected return above the risk free rate would be the same per unit of risk. So it would not matter which duration you leveraged or what the carry was. Higher carry per unit of risk would simply be a reflection of a duration that was expected to benefit less from expected rate changes. BAB is separate from this and causes there to be more return per unit of risk at shorter durations historically. But higher carry per unit risk, on its face, is not evidence of BAB or greater expected return above the risk free rate per unit risk.

It's possible that part of the higher carry of the 2 year is BAB but the majority is certainly that the market expects the 5 to benefit more from future rate changes (more on a per unit risk basis). The 5 year should capture most or all of BAB which is really about terrible risk adjusted returns beyond 10 years.
"Can someone help me understand the downside of always rolling contracts into the highest-carry treasuries between 2-yrs, 3-ys, and 5yrs? It's been mentioned positively in this thread, but I'm sensing some kind of market-timing catch (i.e. the 2-year carry is higher than 5-year carry right now because the market foresees more yield-inversion at that point of the curve)"

There are papers that show a significant excess return of "carry on the yield curve" investing. (There are other papers that say it's no actionable. I have not analyzed those papers yet.)
https://www.cfainstitute.org/en/researc ... 19-1628552
http://wp.lancs.ac.uk/fofi2018/files/20 ... tens-1.pdf

If true, one should be able to overlay a carry strategy over mHFEA by switching to the 2/3/5/7 or 3/5/7/10, etc. futures contract which has the highest carry. Of course everything is duration-adjusted, because you don't want to change your duration exposure. I think "carry" would be yield minus financing rate, or yield + current rolldown return - financing rate.

Of course it's not as simplistic as thinking you can pocket the entire difference between the current carries. Part of the difference reflects market expectations of rolldown and changes in rolldown returns, yield curve movements, etc. But there are significant excess returns as per the paper.

On another note, if you look at the current yield curve, the 10y point (TN) seems to be the absolute loser (in terms of yield i.e. for future expected returns), and I have a bit of a hard time reconciling this with the SOFR curve. treasury 7y is at ca. 3.13%, 10y is at ca. 3.02% i.e. below the 10y. But SOFR futures are in a range of 96.9 and 96.65 between 2029 and 2031, implying 3.1% - 3.35% future short-term rates in the 7-10 year range. The SOFR is the short-term interest rate forecast plus the term premium; per expectations theory, the treasury curve should be the average over the SOFR curve between now and the time in question, i.e. x-year treasury rate = average of short-term rates over the next x years + term premium, shouldn't it? So how can the 10-year treasury be below the 7-year, all the while the SOFR futures for the 7-10y period are above both the 7y and the 10y treasury?

Am I misunderstanding the relation between SOFR/LIBOR futures and the treasury yield curve?

Image

Image

Image
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Rebalancing bands

Post by comeinvest »

skierincolorado wrote: Wed Jul 06, 2022 9:54 am
Hfearless wrote: Wed Jul 06, 2022 9:25 am

Code: Select all

Period    SSO  UPRO/VOO CAGR, rebalanced:
         CAGR  never    ann. quart. monthly
2015-16  9.59%  8.99%  9.13%  9.29%  9.74%
2015-20 19.69% 18.70% 18.95% 20.32% 19.84%
2018-20 19.25% 16.36% 17.51% 20.53% 19.33%
So if you do rebalance, UPRO/VOO seems to come out on top, if only by a little. Perhaps what works in UPRO’s favor is that you simply need less of it so your exposure to its expense ratio and its volatility decay is lower?
Yeah the reduced expense ratio is significant. I would say to be sure don't just rebalance monthly... rebalance with fairly narrow bands.. like 5% moves in sp500 trigger rebalance. In high volatility like 2020 even monthly rebalancing produces underperformance for voo/upro compared to sso by 1% (20.5% vs 21.5%). This is no doubt mostly due to being underleveraged in mid March 2020 at the bottom and would have been fixed by rebalancing back into upro near the bottom.

Great data thank you. In those examples monthly and quarterly work out ok, but there is a risk that a sharp market bottom is reached mid month or mid quarter. Quarterly rebalancing worked in 2018-2020 because it is underleveraged for most of q1 2020 as the market dropped and then get lucky and buys back into upro near the bottom at the end of March. If the market had recovered more by the end of the quarter, upro/voo would have been very underleveraged for the rise back up and missed a lot of gains.
skierincolorado wrote: Wed Jul 06, 2022 11:33 am
Hfearless wrote: Wed Jul 06, 2022 10:56 am For that I had to write some code, for whatever reason the data does not align perfectly with what PV shows. SSO vs VOO/UPRO rebalanced monthly vs VOO/UPRO rebalanced when VOO strays beyond 95%–105% of its value at last rebalance:

Code: Select all

          SSO    UVm    UV5%
2015-16  9.53%  9.89% 10.23%
2015-20 19.68% 20.49% 20.81%
2018-20 18.73% 19.94% 20.27%
This all seems to suggest that if you want to implement mHFEA with futures and you want a modest leverage, say 1.2x, then VOO + UPRO is the way to go. Or VTI + UPRO. Or VTI + TQQQ, why not.
Preferably to SSO as long as you rebalance with narrow bands and/or monthly. But futures would still be preferable overall if available.


Are you doing jan2018 to dec2020? For Sso PV gives 19.25% a whole percent higher than what you got.

Reducing bands to 2% seems to eliminate most of the tracking error and behave similar to monthly rebalancing. But this is of course in PV which cannot rebalancd mid month if a band is crossed Mid month. I'd still prefer bands over monthly rebalancing in case of a big mid month crash and recovery. As your code shows, 5% bands actually seems to have some benefit and take advantage of momentum swings.

Is your code rebalancing only on monthly dates or can it rebalance mid month?
I'm trying to write my investment plan. I'm doing mHFEA with futures, not LETFs. The discussion about rebalancing on this page applied to LETF based rebalancing only, right?
With futures I don't need to counteract volatility decay, but I still want to benefit from the rebalancing bonus of a regular stocks/bonds portfolio.
The rebalancing bands refer to deviation from the target allocation percentage which can be monitored just looking at the current state of the portfolio, not to movements of the underlying which would be harder to keep track of as you would have to keep track of history.
Say I have a 140/200 equities/ITT target. The popular 5/25 rebalancing bands where the "5" applies to the larger position in a 60/40 portfolio, is triggered on a ca. +23% move in the equities market (i.e. they are rarely triggered), if bonds don't move: 60 * 1.23 = 74; increase in assets and equity: 74 - 60 = 14; 74 / (100 + 14) = 0.65; which would trigger the band (you would have 65% equities when the target was 60%).
For the 140/200 the same market movement would translate to 140 * 1.23 = 172; increase in assets and equity: 172 - 140 = 32; 172 / (100 + 32) = 1.3; i.e. the equivalent rebalancing band would be 10% deviation from target. Is my math correct? Does my whole thinking make sense?
Last edited by comeinvest on Wed Jul 20, 2022 4:53 am, edited 2 times in total.
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

comeinvest wrote: Wed Jul 20, 2022 12:34 am I think your alternatives are too complex. Not only implementation complexity, but also the increasing risk of parameter fitting i.e. "predicting the past, not the future".

What do you mean with "bond convexity has me unconcerned about rate hikes past 3%" - can you please elaborate?
I've been doing continued research and I agree. I sensed my approach was naive, and I was hoping someone could course-correct me. Backtesting a lot more time periods has revealed to me that "bet against beta" (namely consumer staples) and gold have strong correlations with bonds. They're all "defensive positions" that can get crowded at the same times.

I think I'll edit/delete those posts to not confuse any lurkers.

That being said, I think small-cap-value deserves further looking-into. If we interpret SCV's increased risk/return as "20% leverage", we can then lower our total equity allocation by that amount (e.g. 200% -> 160%). This means borrowing 40% less money, which can make a big difference during certain rate cycles.

However, the biggest benefit I see is the increased interaction between ITTs and SCV. SCV's increased volatility results in more rebalancing alpha, and SCV's idiosyncratic risks/rewards line up fantastically with bonds. In the 1973 oil crisis, the 1993 bond correction, and this recent 2022 inflation, SCV would have done wonders for drawdown protection (on top of it being a 20% smaller position).

What's really started to convince me is backtests like this: https://www.portfoliovisualizer.com/bac ... tion7_2=60

160% Global SCV + 400% 7-year-treasures (600% duration-matched ZF) vs. 200% Global market + 400% 7-year-treasuries. You'll see that the SCV portfolio currently has higher sharpe/sortino... but remove the VFITX + CASHX positions, use the gear icon to "Normalize" the portfolios, and you'll see that at this starting date, SCV had a lower sharpe/sortino than the global market.

Another example starting at 1985: https://www.portfoliovisualizer.com/bac ... ation4_1=5

1988: https://www.portfoliovisualizer.com/bac ... ation4_1=5

1995: https://www.portfoliovisualizer.com/bac ... ation4_1=5

etc.

Would love some more thoughts on this. It seems much more fundamentally sound/simple than what I had going before, and the sortino gets even higher at 600% VFITX (though I admit that's getting pretty steep)
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

LazyOverthinker wrote: Wed Jul 20, 2022 2:42 am
comeinvest wrote: Wed Jul 20, 2022 12:34 am I think your alternatives are too complex. Not only implementation complexity, but also the increasing risk of parameter fitting i.e. "predicting the past, not the future".

What do you mean with "bond convexity has me unconcerned about rate hikes past 3%" - can you please elaborate?
I've been doing continued research and I agree. I sensed my approach was naive, and I was hoping someone could course-correct me. Backtesting a lot more time periods has revealed to me that "bet against beta" (namely consumer staples) and gold have strong correlations with bonds. They're all "defensive positions" that can get crowded at the same times.

I think I'll edit/delete those posts to not confuse any lurkers.

That being said, I think small-cap-value deserves further looking-into. If we interpret SCV's increased risk/return as "20% leverage", we can then lower our total equity allocation by that amount (e.g. 200% -> 160%). This means borrowing 40% less money, which can make a big difference during certain rate cycles.

However, the biggest benefit I see is the increased interaction between ITTs and SCV. SCV's increased volatility results in more rebalancing alpha, and SCV's idiosyncratic risks/rewards line up fantastically with bonds. In the 1973 oil crisis, the 1993 bond correction, and this recent 2022 inflation, SCV would have done wonders for drawdown protection (on top of it being a 20% smaller position).

What's really started to convince me is backtests like this: https://www.portfoliovisualizer.com/bac ... tion7_2=60

160% Global SCV + 400% 7-year-treasures (600% duration-matched ZF) vs. 200% Global market + 400% 7-year-treasuries. You'll see that the SCV portfolio currently has higher sharpe/sortino - but remove the VFITX + CASHX position, use the gear icon to "Normalize" the portfolios, and you'll see that at this starting date, SCV had a lower sharpe/sortino than the global market.

Another example starting at 1985: https://www.portfoliovisualizer.com/bac ... ation4_1=5

1988: https://www.portfoliovisualizer.com/bac ... ation4_1=5

1995: https://www.portfoliovisualizer.com/bac ... ation4_1=5

etc.

Would love some more thoughts on this. It seems much more fundamentally sound/simple than what I had going before, and the sortino gets even higher at 600% VFITX (though I admit that's getting pretty steep)
I have to read your post in more detail later. There was another thread with some debate over whether SCV and the market are correlated, i.e. whether it just amplifies the market or whether it is an independent source of return. Not sure if it was this one viewtopic.php?t=369537
I know skierincolorado and myself independently try to add factor exposure to mHFEA, which has some implications and tradeoffs because there are no equity index futures with suitable factor exposure. I'm still working on my investment plan.
And here is the guy who uses no leverage, but a strategy that rebalances several factor tilted equities ETFs of different geographies, with godd historical outperformance over plain equity indexes: viewtopic.php?t=372156
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

comeinvest wrote: Wed Jul 20, 2022 2:58 am There was another thread with some debate over whether SCV and the market are correlated, i.e. whether it just amplifies the market or whether it is an independent source of return.
I've also heard this concern, but I think there are enough observable periods where SCV outperforms by doing its own thing. Particularly 1973, 1993, 2000 and 2022 show that whatever idiosyncratic risk/rewards SCV has are likely correlated with treasury rates going up. There is a very sneaky bond-correction from April-May 2000 that SCV would have greatly mitigated. Using google finances intra-month data shows that, if you were caught with your pants down holding a greedy 4:1 ITT:equities position @6.1% rates, it would have been the difference between seeing your liquid net worth go down 37% (with market stocks) or 22% (with SCV).
comeinvest wrote: Wed Jul 20, 2022 2:58 am I know skierincolorado and myself independently try to add factor exposure to mHFEA, which has some implications and tradeoffs because there are no equity index futures with suitable factor exposure. I'm still working on my investment plan.
Good point - one may have to continue planning around SP500 exposure in an IRA. I'm not sure what the consensus is on total IRA leverage, but if it's 150% equity exposure... yeah, it may be hard to have much more than a 20% SCV tilt. It wouldn't be too much of a headache though, you just have to consider the SCV as 1.2x leverage when calculating the total leverage of the IRA.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

LazyOverthinker wrote: Wed Jul 20, 2022 4:17 am
comeinvest wrote: Wed Jul 20, 2022 2:58 am There was another thread with some debate over whether SCV and the market are correlated, i.e. whether it just amplifies the market or whether it is an independent source of return.
I've also heard this concern, but I think there are enough observable periods where SCV outperforms by doing its own thing. Particularly 1973, 1993, 2000 and 2022 show that whatever idiosyncratic risk/rewards SCV has are likely correlated with treasury rates going up. There is a very sneaky bond-correction from April-May 2000 that SCV would have greatly mitigated. Using google finances intra-month data shows that, if you were caught with your pants down holding a greedy 4:1 ITT:equities position @6.1% rates, it would have been the difference between seeing your liquid net worth go down 37% (with market stocks) or 22% (with SCV).
comeinvest wrote: Wed Jul 20, 2022 2:58 am I know skierincolorado and myself independently try to add factor exposure to mHFEA, which has some implications and tradeoffs because there are no equity index futures with suitable factor exposure. I'm still working on my investment plan.
Good point - one may have to continue planning around SP500 exposure in an IRA. I'm not sure what the consensus is on total IRA leverage, but if it's 150% equity exposure... yeah, it may be hard to have much more than a 20% SCV tilt. It wouldn't be too much of a headache though, you just have to consider the SCV as 1.2x leverage when calculating the total leverage of the IRA.
Where do you get the 1.2 from? Source or calculation?
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

comeinvest wrote: Wed Jul 20, 2022 4:55 am Where do you get the 1.2 from? Source or calculation?
It's an estimation from different angles. PV has SCV data back to 1972, but that is a very favorable starting point that gives SCV 31% more CAGR than total-market. Starting at 1979, SCV has 20% more CAGR up to current month. It then supposedly only has 15% more standard deviation, but that's too simple - we should instead look at deflationary drawdowns. In 2007, US-SCV fell 25% more than the US market. Global SCV only fell 17% more than the global market. During the covid crash, US-SCV fell 25% harder, while global SCV fell 27%~ harder.

There are a few idiosyncratic crashes where SCV really shows its risk, like in 1998 where global markets fell 17.69% compared to SCV's whopping 34.15%. But 1998 is a good example of bonds going up in response to a stock crash. If you were to hold a greedy enough bond position (e.g. 4:1) and the variance of bonds lined up just right, you may not have experienced more than a 17%~ total drawdown even with SCV.

And of course, all of the above must be included with the fact that SCV had significantly less drawdowns in 1973, 1993, and 2022 when mHFEA would have wanted that the most.

To me, that justifies an estimate of 20% increased variance up and down. Thus, we lower our equity allocation 20% and reap the benefits of the implied 20% CASHX position. The backtests I linked seem to justify the estimate.

P.S. I asked around for a more academic estimate @ the RationalReminder forum, and I got a very CAPM response: the historical sharpe ratio of SCV should be roughly equal to the total stock market. So whatever increased CAGR you believe in for SCV, it's likely a good estimate for baked-in leverage.

P.P.S. US-SCV has -0.11 correlation to ITTs, while the US stock market has -0.05 correlation. Seems like the simplest explanation for why SCV+ITT could produce alpha
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

This seems too good to be true...

If on May 1st '07 you leveraged $100K as 140% VFINX, 60% VGTSX, and 600% ZF, you'd lose it all (if you didn't rebalance):

Image

But if you leveraged $100K as 112% DFSVX, 36% DISVX, 12% DFEVX and 600% ZF, you'd "only" lose 77%:

Image

Again, if your bond position was even greedier (1000% ZF), the SCV portfolio would "only" lose 71K, whereas the total-market + 1000% ZF still loses 95K.

I've checked my inputs thrice. Either I'm crazy, google finance's data is off, or SCV interacted with bonds much more desirably in this very important timespan. I'm also just realizing that 160% times 1.2x is 192%, so 166% total SCV exposure would be more appropriate when comparing to 200% market-equity. Accounting for that, the SCV portfolio goes 79% from peak to trough, so the point still stands.

If someone would like to double-check my inputs, here they are (I estimate VFITX as 1.555x the duration of ZF):

Image
HolyGrill
Posts: 52
Joined: Fri Feb 19, 2021 5:33 pm

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

Post by HolyGrill »

Skier,

Thanks for the detailed reply. I have access to options in Australia. I’m pretty new to option. For example I loook for MES and I can see I have few options on my screen:

MES SEP
MES DEC
MES MAR
MES JUN
MES ∞ SEP

Which one should I buy?
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

comeinvest wrote: Wed Jul 20, 2022 1:14 am
skierincolorado wrote: Tue Jul 19, 2022 9:39 am
LazyOverthinker wrote: Tue Jul 19, 2022 3:20 am
skierincolorado wrote: Mon Jul 18, 2022 8:02 pm The market is basically betting that yields will fall in the medium term and that the 5 year will benefit more than the 2 year from this.
Ok, that makes sense. Does that same market-logic apply if we replace our 5-year treasuries with 60% more "duration-matched" 2-year treasuries?

In other words: even if we agree with the market's bet on 5-year rates, are we properly compensated for giving up the chance to hold 60% more-total "0.65% carry" treasuries?

Essentially I'm wondering if leveraged investors have a "bet against beta" advantage by being able to adapt total $ exposure, or if that is captured in the market's foresight.
That should be captured. For example let's say the equilibrium rate is 2% and the market is betting we will rall to that rate in 2 years. Let's say the FF rate is 1.5% the whole time for simplicity. Would we earn more by holding 100k of 5 year bonds with initial rate of 3% or 250k of 2 year bonds with initial yield of 2.5%? With the 2 year bonds we net 1% of 250k or 2.5k. With the 5 year bonds we net 1.5% of 100k, or 1.5k. But we also would see a capital appreciation of our 5 year bonds of 3k for total profit of 4.5k.

This example illustrates how the bond with higher carry per unit duration can easily have lower returns. Of course if this were actually the market expectation it would easily be arbitraged out of existence and would not happen in the first place.

Bet against beta is separate from this. In an efficient market there would be no advantage to investing in shorter durations with highe carry per unit risk even with leverage because the expected return above the risk free rate would be the same per unit of risk. So it would not matter which duration you leveraged or what the carry was. Higher carry per unit of risk would simply be a reflection of a duration that was expected to benefit less from expected rate changes. BAB is separate from this and causes there to be more return per unit of risk at shorter durations historically. But higher carry per unit risk, on its face, is not evidence of BAB or greater expected return above the risk free rate per unit risk.

It's possible that part of the higher carry of the 2 year is BAB but the majority is certainly that the market expects the 5 to benefit more from future rate changes (more on a per unit risk basis). The 5 year should capture most or all of BAB which is really about terrible risk adjusted returns beyond 10 years.
"Can someone help me understand the downside of always rolling contracts into the highest-carry treasuries between 2-yrs, 3-ys, and 5yrs? It's been mentioned positively in this thread, but I'm sensing some kind of market-timing catch (i.e. the 2-year carry is higher than 5-year carry right now because the market foresees more yield-inversion at that point of the curve)"

There are papers that show a significant excess return of "carry on the yield curve" investing. (There are other papers that say it's no actionable. I have not analyzed those papers yet.)
https://www.cfainstitute.org/en/researc ... 19-1628552
http://wp.lancs.ac.uk/fofi2018/files/20 ... tens-1.pdf

If true, one should be able to overlay a carry strategy over mHFEA by switching to the 2/3/5/7 or 3/5/7/10, etc. futures contract which has the highest carry. Of course everything is duration-adjusted, because you don't want to change your duration exposure. I think "carry" would be yield minus financing rate, or yield + current rolldown return - financing rate.

Of course it's not as simplistic as thinking you can pocket the entire difference between the current carries. Part of the difference reflects market expectations of rolldown and changes in rolldown returns, yield curve movements, etc. But there are significant excess returns as per the paper.

On another note, if you look at the current yield curve, the 10y point (TN) seems to be the absolute loser (in terms of yield i.e. for future expected returns), and I have a bit of a hard time reconciling this with the SOFR curve. treasury 7y is at ca. 3.13%, 10y is at ca. 3.02% i.e. below the 10y. But SOFR futures are in a range of 96.9 and 96.65 between 2029 and 2031, implying 3.1% - 3.35% future short-term rates in the 7-10 year range. The SOFR is the short-term interest rate forecast plus the term premium; per expectations theory, the treasury curve should be the average over the SOFR curve between now and the time in question, i.e. x-year treasury rate = average of short-term rates over the next x years + term premium, shouldn't it? So how can the 10-year treasury be below the 7-year, all the while the SOFR futures for the 7-10y period are above both the 7y and the 10y treasury?

Am I misunderstanding the relation between SOFR/LIBOR futures and the treasury yield curve?

Image

Image

Image
THREEFF7 is the instant rate 7 years hence and is at 2.8% currently. That's more consistent with the inversion we see. Not sure what the difference with the sofr rates is due to. I usually use forward rates from the fed.
Googling THREEFF7 should bring it up or just fed forward rates.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Rebalancing bands

Post by skierincolorado »

comeinvest wrote: Wed Jul 20, 2022 2:40 am
skierincolorado wrote: Wed Jul 06, 2022 9:54 am
Hfearless wrote: Wed Jul 06, 2022 9:25 am

Code: Select all

Period    SSO  UPRO/VOO CAGR, rebalanced:
         CAGR  never    ann. quart. monthly
2015-16  9.59%  8.99%  9.13%  9.29%  9.74%
2015-20 19.69% 18.70% 18.95% 20.32% 19.84%
2018-20 19.25% 16.36% 17.51% 20.53% 19.33%
So if you do rebalance, UPRO/VOO seems to come out on top, if only by a little. Perhaps what works in UPRO’s favor is that you simply need less of it so your exposure to its expense ratio and its volatility decay is lower?
Yeah the reduced expense ratio is significant. I would say to be sure don't just rebalance monthly... rebalance with fairly narrow bands.. like 5% moves in sp500 trigger rebalance. In high volatility like 2020 even monthly rebalancing produces underperformance for voo/upro compared to sso by 1% (20.5% vs 21.5%). This is no doubt mostly due to being underleveraged in mid March 2020 at the bottom and would have been fixed by rebalancing back into upro near the bottom.

Great data thank you. In those examples monthly and quarterly work out ok, but there is a risk that a sharp market bottom is reached mid month or mid quarter. Quarterly rebalancing worked in 2018-2020 because it is underleveraged for most of q1 2020 as the market dropped and then get lucky and buys back into upro near the bottom at the end of March. If the market had recovered more by the end of the quarter, upro/voo would have been very underleveraged for the rise back up and missed a lot of gains.
skierincolorado wrote: Wed Jul 06, 2022 11:33 am
Hfearless wrote: Wed Jul 06, 2022 10:56 am For that I had to write some code, for whatever reason the data does not align perfectly with what PV shows. SSO vs VOO/UPRO rebalanced monthly vs VOO/UPRO rebalanced when VOO strays beyond 95%–105% of its value at last rebalance:

Code: Select all

          SSO    UVm    UV5%
2015-16  9.53%  9.89% 10.23%
2015-20 19.68% 20.49% 20.81%
2018-20 18.73% 19.94% 20.27%
This all seems to suggest that if you want to implement mHFEA with futures and you want a modest leverage, say 1.2x, then VOO + UPRO is the way to go. Or VTI + UPRO. Or VTI + TQQQ, why not.
Preferably to SSO as long as you rebalance with narrow bands and/or monthly. But futures would still be preferable overall if available.


Are you doing jan2018 to dec2020? For Sso PV gives 19.25% a whole percent higher than what you got.

Reducing bands to 2% seems to eliminate most of the tracking error and behave similar to monthly rebalancing. But this is of course in PV which cannot rebalancd mid month if a band is crossed Mid month. I'd still prefer bands over monthly rebalancing in case of a big mid month crash and recovery. As your code shows, 5% bands actually seems to have some benefit and take advantage of momentum swings.

Is your code rebalancing only on monthly dates or can it rebalance mid month?
I'm trying to write my investment plan. I'm doing mHFEA with futures, not LETFs. The discussion about rebalancing on this page applied to LETF based rebalancing only, right?
With futures I don't need to counteract volatility decay, but I still want to benefit from the rebalancing bonus of a regular stocks/bonds portfolio.
The rebalancing bands refer to deviation from the target allocation percentage which can be monitored just looking at the current state of the portfolio, not to movements of the underlying which would be harder to keep track of as you would have to keep track of history.
Say I have a 140/200 equities/ITT target. The popular 5/25 rebalancing bands where the "5" applies to the larger position in a 60/40 portfolio, is triggered on a ca. +23% move in the equities market (i.e. they are rarely triggered), if bonds don't move: 60 * 1.23 = 74; increase in assets and equity: 74 - 60 = 14; 74 / (100 + 14) = 0.65; which would trigger the band (you would have 65% equities when the target was 60%).
For the 140/200 the same market movement would translate to 140 * 1.23 = 172; increase in assets and equity: 172 - 140 = 32; 172 / (100 + 32) = 1.3; i.e. the equivalent rebalancing band would be 10% deviation from target. Is my math correct? Does my whole thinking make sense?
The strategy you have described is more similar to LETF because you are maintaining a fixed leverage ratio but with less frequent rebalancing. It's less consistent with lifecycle investing and more subject to volatility decay. It can work for very high risk tolerance but there there is a point beyond which increasing leverage does not increase returns and can decrease them.
unemployed_pysicist
Posts: 220
Joined: Sat Oct 09, 2021 2:32 pm
Location: Amsterdam

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

Post by unemployed_pysicist »

comeinvest wrote: Wed Jul 20, 2022 1:14 am
On another note, if you look at the current yield curve, the 10y point (TN) seems to be the absolute loser (in terms of yield i.e. for future expected returns), and I have a bit of a hard time reconciling this with the SOFR curve. treasury 7y is at ca. 3.13%, 10y is at ca. 3.02% i.e. below the 10y. But SOFR futures are in a range of 96.9 and 96.65 between 2029 and 2031, implying 3.1% - 3.35% future short-term rates in the 7-10 year range. The SOFR is the short-term interest rate forecast plus the term premium; per expectations theory, the treasury curve should be the average over the SOFR curve between now and the time in question, i.e. x-year treasury rate = average of short-term rates over the next x years + term premium, shouldn't it? So how can the 10-year treasury be below the 7-year, all the while the SOFR futures for the 7-10y period are above both the 7y and the 10y treasury?
It is possible for the SOFR curve to have a different shape than the treasury curve, but you have to compare them on equal footing (treasury forward curve vs sofr forward curve, treasury zero coupon curve vs sofr zero coupon curve, etc.) You need to apply the convexity adjustment to the SOFR futures rates in order to get a forward rate, just like with Eurodollar futures.
couldn't afford the h | | BUY BONDS | WEAR DIAMONDS
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

Re: Rebalancing bands

Post by LazyOverthinker »

comeinvest wrote: Wed Jul 20, 2022 2:40 am The popular 5/25 rebalancing bands
I looked into this today for my own portfolio, and found quarterly rebalancing to almost always win-out in sortino ratio. I tried it for your portfolio's equity:treasury ratio at a few different time periods: https://www.portfoliovisualizer.com/bac ... n2_1=58.83

Whether quarterly's success is due to lucky timings or "capturing momentum," it takes some very funky "rebalancing bands" to match quarterly's sortino ratio, making it feel even more like overfitting to me. I plan to put monthly contributions into the lagging assets and then re-balance any remaining differences quarterly, avoiding this if possible for the first year (capital gains).
comeinvest wrote: Wed Jul 20, 2022 2:40 am but I still want to benefit from the rebalancing bonus of a regular stocks/bonds portfolio.
Then you really should consider my posts above. SCV + ITT captures more rebalancing benefits than Total Market + ITT, and the interaction is strong enough that I feel secure holding more ITT than I would with Total Market, which just further increases returns.

Even while SCV underperforms the market, the above can make up for it. Here's an example with real funds where, because of the decreased correlation, the investor in portfolio 1 feels comfortable holding 20% more total treasuries than the VT investor in portfolio 3 does: https://www.portfoliovisualizer.com/bac ... n10_3=-175

I use 25% SHY for funding to simulate the 6-month treasury bill, and as previously mentioned you have to decrease your equity position by 20% if switching to fully SCV (which saves money on borrowing).

Starting at Feb. 2007 there (IEI inception date), the SCV portfolio has a comparable Sortino ratio, despite global SCV equities underperforming by 0.82% CAGR and +4.7% SD in this timespan.

Move to the less-hectic starting date of 2010 and the ratios remain respectable despite SCV underperforming by a similar degree. Speaking of rebalancing, if you switch to "quarterly," SCV catches up quite a bit to the total-market portfolio. I made the link "Monthly" since that's more accurate for comparing two different leverage ratios.

Finally, note that the SCV portfolio had lower drawdowns in 2008 and 2022.
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

Just a thought I had, for 2x VT to match 1.66x SCV's 2022 drawdown, you'd actually need to hold 33% less bonds than the SCV portfolio.

And that makes a huge difference when your starting date is 2007: https://www.portfoliovisualizer.com/bac ... n10_3=-175
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: Rebalancing bands

Post by comeinvest »

LazyOverthinker wrote: Thu Jul 21, 2022 4:44 pm
comeinvest wrote: Wed Jul 20, 2022 2:40 am The popular 5/25 rebalancing bands
I looked into this today for my own portfolio, and found quarterly rebalancing to almost always win-out in sortino ratio. I tried it for your portfolio's equity:treasury ratio at a few different time periods: https://www.portfoliovisualizer.com/bac ... n2_1=58.83

Whether quarterly's success is due to lucky timings or "capturing momentum," it takes some very funky "rebalancing bands" to match quarterly's sortino ratio, making it feel even more like overfitting to me. I plan to put monthly contributions into the lagging assets and then re-balance any remaining differences quarterly, avoiding this if possible for the first year (capital gains).
comeinvest wrote: Wed Jul 20, 2022 2:40 am but I still want to benefit from the rebalancing bonus of a regular stocks/bonds portfolio.
Then you really should consider my posts above. SCV + ITT captures more rebalancing benefits than Total Market + ITT, and the interaction is strong enough that I feel secure holding more ITT than I would with Total Market, which just further increases returns.

Even while SCV underperforms the market, the above can make up for it. Here's an example with real funds where, because of the decreased correlation, the investor in portfolio 1 feels comfortable holding 20% more total treasuries than the VT investor in portfolio 3 does: https://www.portfoliovisualizer.com/bac ... n10_3=-175

I use 25% SHY for funding to simulate the 6-month treasury bill, and as previously mentioned you have to decrease your equity position by 20% if switching to fully SCV (which saves money on borrowing).

Starting at Feb. 2007 there (IEI inception date), the SCV portfolio has a comparable Sortino ratio, despite global SCV equities underperforming by 0.82% CAGR and +4.7% SD in this timespan.

Move to the less-hectic starting date of 2010 and the ratios remain respectable despite SCV underperforming by a similar degree. Speaking of rebalancing, if you switch to "quarterly," SCV catches up quite a bit to the total-market portfolio. I made the link "Monthly" since that's more accurate for comparing two different leverage ratios.

Finally, note that the SCV portfolio had lower drawdowns in 2008 and 2022.
Interesting observations. But the timeframe of 2007-2022 is too short to draw conclusions. I think mHFEA was tested 1955-2021.
I personally have mostly international (ex-U.S.) equities for my mHFEA, which I think are more valuey in the first place. But how to implement mHFEA with international equities is a whole different topic. Whether to currency-hedge or not, and which leverage ratio vs. a U.S.-only mHFEA implementation.
strategist37
Posts: 6
Joined: Sat Jul 23, 2022 10:34 am

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

Post by strategist37 »

LazyOverthinker wrote: Wed Jul 20, 2022 2:42 am
That being said, I think small-cap-value deserves further looking-into. If we interpret SCV's increased risk/return as "20% leverage", we can then lower our total equity allocation by that amount (e.g. 200% -> 160%). This means borrowing 40% less money, which can make a big difference during certain rate cycles.

However, the biggest benefit I see is the increased interaction between ITTs and SCV. SCV's increased volatility results in more rebalancing alpha, and SCV's idiosyncratic risks/rewards line up fantastically with bonds. In the 1973 oil crisis, the 1993 bond correction, and this recent 2022 inflation, SCV would have done wonders for drawdown protection (on top of it being a 20% smaller position).

What's really started to convince me is backtests like this: https://www.portfoliovisualizer.com/bac ... tion7_2=60

160% Global SCV + 400% 7-year-treasures (600% duration-matched ZF) vs. 200% Global market + 400% 7-year-treasuries. You'll see that the SCV portfolio currently has higher sharpe/sortino... but remove the VFITX + CASHX positions, use the gear icon to "Normalize" the portfolios, and you'll see that at this starting date, SCV had a lower sharpe/sortino than the global market.

..................

Would love some more thoughts on this. It seems much more fundamentally sound/simple than what I had going before, and the sortino gets even higher at 600% VFITX (though I admit that's getting pretty steep)
From my point of view, I am convinced that adding SCV to stock mix will improve performance for reasons others mentioned before -- and I think counting SCV at 1.2 leverage is reasonable. I also believe having _all_ of stock portion in SCV is suboptimal -- partly because of potential for not-in-backtest behavior that can crush SCV (I assume detailed discussion of that is not in scope - But basically, since we are backtesting indexes, rather than individual stocks, all we have is a handful of decades of data for _one_ sample path, so we need to supplement backtest with fundamental analysis, and my fundamental analysis tells me 1.2 (= 1.45/1.2) leverage on SCV is too risky, bonds or not). Secondary reasons to be partially but not fully in SCV include diversification (I agree with those who pointed out before that SCV performance is fundamentally different from US market, not just effectively 1.2 leveraged US market, thus providing diversification benefit). This is assuming one can achieve that leverage inexpensively -- with futures for bond and overall US market exposure, this is very doable just by holding VIOV, as long as desired SCV leverage is not too large.

Right now my SCV is about 1/5th of US stock allocation, and I plan to increase it to ~1/3rd shortly -- I can see a reason to bump it to 1/2 or go lower to 1/6th (and I'd love to discuss pros and cons of that if someone has thoughts on it), but I'm guessing optimal SCV fraction is between those two values.

One disclaimer is that what I am running is not exactly mHFEA, but long term leverage is very similar, so under normal market conditions (which we are not in now, but are ~90% of the time) optimal SCV fraction should be the same.
Last edited by strategist37 on Sat Jul 23, 2022 3:22 pm, edited 4 times in total.
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

Re: Rebalancing bands

Post by LazyOverthinker »

comeinvest wrote: Fri Jul 22, 2022 8:13 am But the timeframe of 2007-2022 is too short to draw conclusions.
Agreed, the 2007 starting date is just to feature IEI. I'm not sure if Simba's backtest has SCV, but PV's "asset allocation" allows you to test SCV+ITT back to 1972~ and see how incredibly helpful it was to be in SCV during the 70s oil/bond crisis (and the 1993-1994 bond drawdowns, and the year 2002~ bond dip)
strategist37 wrote: Sat Jul 23, 2022 3:10 pm Right now my SCV is about 1/5th of US stock allocation, and I plan to increase it to ~1/3rd shortly -- I can see a reason to bump it to 1/2 or go lower to 1/6th (and I'd love to discuss pros and cons of that if someone has thoughts on it), but I'm guessing optimal SCV fraction is between those two values.
Your whole post is valid, I just want to add that there's a lot of discussion about how much a "long-only" SCV position actually captures the SCV-premium. By some estimates, you're only 33% "in" SCV when your position is 100%, long-only, diversified SCV. The other 67% is essentially market-beta. So it could be that a 33% long-SCV position in an otherwise market-cap-weighted portfolio is more like 11% pure exposure to SCV.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: Rebalancing bands

Post by comeinvest »

LazyOverthinker wrote: Mon Jul 25, 2022 12:43 am Your whole post is valid, I just want to add that there's a lot of discussion about how much a "long-only" SCV position actually captures the SCV-premium. By some estimates, you're only 33% "in" SCV when your position is 100%, long-only, diversified SCV. The other 67% is essentially market-beta. So it could be that a 33% long-SCV position in an otherwise market-cap-weighted portfolio is more like 11% pure exposure to SCV.
There are other long threads where people deliberate which is the optimal SCV percentage. I think there was no conclusive result, which will be difficult to obtain given that the utility of overweighting SCV in the first place is highly controversial. But if you want to combine factor tilts with mHFEA with >100% equities in the form of index futures, the question is largely moot as much of the equities will be in the index. That is for IRAs and maybe for 401(k)s. In taxable accounts you have more options for the borrowing part and you can factor tilt all your equities.
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

Re: Rebalancing bands

Post by LazyOverthinker »

comeinvest wrote: Mon Jul 25, 2022 8:11 pm That is for IRAs and maybe for 401(k)s. In taxable accounts you have more options for the borrowing part and you can factor tilt all your equities.
100%, this is what I've landed on. I'm keeping my IRA simple between MES and ZT/ZF until it grows enough to allow wiggle-room in international futures. Meanwhile, my taxable is entirely global SCV with 55% US allocation. If the jury is still out on how much SCV is the right amount, then I prefer to lean fully in the direction that captures SCV's lower correlation with bonds.

Here's hoping we get that SCV futures contract one day!
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

LazyOverthinker wrote: Tue Jul 26, 2022 3:07 pm
comeinvest wrote: Mon Jul 25, 2022 8:11 pm That is for IRAs and maybe for 401(k)s. In taxable accounts you have more options for the borrowing part and you can factor tilt all your equities.
100%, this is what I've landed on. I'm keeping my IRA simple between MES and ZT/ZF until it grows enough to allow wiggle-room in international futures. Meanwhile, my taxable is entirely global SCV with 55% US allocation. If the jury is still out on how much SCV is the right amount, then I prefer to lean fully in the direction that captures SCV's lower correlation with bonds.

Here's hoping we get that SCV futures contract one day!
There is 1 factor futures contract with enough volume: /RLV (based on the Russel 1000 Value Index). I decided not to use it, because the Russell factor indexes seem to have significant issues.

I use ZF and ZN, while you use ZT and ZF.

My taxable account is in individual, international equities (basically index sampling), because this gives me almost guaranteed excess returns from tax loss management, whereas the excess returns (or lack thereof) of factor tilting are heavily debated. I also don't want to be effectively locked into any ETF product for life.
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

comeinvest wrote: Wed Jul 27, 2022 1:07 am There is 1 factor futures contract with enough volume: /RLV (based on the Russel 1000 Value Index). I decided not to use it, because the Russell factor indexes seem to have significant issues.

I use ZF and ZN, while you use ZT and ZF.

My taxable account is in individual, international equities (basically index sampling), because this gives me almost guaranteed excess returns from tax loss management, whereas the excess returns (or lack thereof) of factor tilting are heavily debated. I also don't want to be effectively locked into any ETF product for life.
Thanks for the tip about /RLV - yeah, I've heard similar misgivings about the Russel 1000. Plus, DFA's/Avantis' aggressive methodology is a significant driver of bond uncorrelation opposed to the standard SCV index.

Currently only using ZT because it has less duration exposure in one contract than ZF, and my IRA is that small right now (I was late to start contributing to it). I agree with your diversification more - ZN-duration is less vulnerable to yield compression (2022), while ZF-duration seems more responsive to equity crashes.

Interesting, I've not read about index sampling as a tax strategy. I intended to swap Avantis funds for DFA funds if there was ever an enormous drawdown. Index sampling sounds more complicated/hands-on
gtwhitegold
Posts: 673
Joined: Fri Sep 21, 2012 1:55 pm

Re: Rebalancing bands

Post by gtwhitegold »

LazyOverthinker wrote: Tue Jul 26, 2022 3:07 pm
comeinvest wrote: Mon Jul 25, 2022 8:11 pm That is for IRAs and maybe for 401(k)s. In taxable accounts you have more options for the borrowing part and you can factor tilt all your equities.
100%, this is what I've landed on. I'm keeping my IRA simple between MES and ZT/ZF until it grows enough to allow wiggle-room in international futures. Meanwhile, my taxable is entirely global SCV with 55% US allocation. If the jury is still out on how much SCV is the right amount, then I prefer to lean fully in the direction that captures SCV's lower correlation with bonds.

Here's hoping we get that SCV futures contract one day!
From what I have seen, there aren't any good small cap value future contract options available. From what I have seen, your best bet in the small cap space is SMC, which tracks the S&P 600 index, so it has a quality tilt vice a junk tilt like the Russell 2000. I have considered using it, but I have refrained from using it since I have much of my international exposure in my Roth IRA and I'm not going to change that just to allow for more leverage. This may change however once I separate from my current job and I can roll over my current retirement plan. I'm slightly hesitant about it since I don't want to spend time rolling over futures contracts, but I don't get that it's too big of a deal.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: Rebalancing bands

Post by comeinvest »

gtwhitegold wrote: Wed Jul 27, 2022 9:47 pm From what I have seen, there aren't any good small cap value future contract options available. From what I have seen, your best bet in the small cap space is SMC, which tracks the S&P 600 index, so it has a quality tilt vice a junk tilt like the Russell 2000.
The S&P 600 follows the same methodology as the S&P500 just small cap instead of large cap, doesn't it? If yes, then it would have practically no factor tilt except very minimal filtering out of very low quality companies.
User avatar
cos
Posts: 506
Joined: Fri Aug 23, 2019 7:34 pm
Location: Boston
Contact:

Re: Rebalancing bands

Post by cos »

comeinvest wrote: Thu Jul 28, 2022 6:45 pm
gtwhitegold wrote: Wed Jul 27, 2022 9:47 pm From what I have seen, there aren't any good small cap value future contract options available. From what I have seen, your best bet in the small cap space is SMC, which tracks the S&P 600 index, so it has a quality tilt vice a junk tilt like the Russell 2000.
The S&P 600 follows the same methodology as the S&P500 just small cap instead of large cap, doesn't it? If yes, then it would have practically no factor tilt except very minimal filtering out of very low quality companies.
You'd be surprised. The S&P SmallCap 600 gets you about halfway to the actual value indices, and when you're leveraging that exposure, that ain't too shabby.
gtwhitegold
Posts: 673
Joined: Fri Sep 21, 2012 1:55 pm

Re: Rebalancing bands

Post by gtwhitegold »

cos wrote: Fri Jul 29, 2022 9:13 am
comeinvest wrote: Thu Jul 28, 2022 6:45 pm
gtwhitegold wrote: Wed Jul 27, 2022 9:47 pm From what I have seen, there aren't any good small cap value future contract options available. From what I have seen, your best bet in the small cap space is SMC, which tracks the S&P 600 index, so it has a quality tilt vice a junk tilt like the Russell 2000.
The S&P 600 follows the same methodology as the S&P500 just small cap instead of large cap, doesn't it? If yes, then it would have practically no factor tilt except very minimal filtering out of very low quality companies.
You'd be surprised. The S&P SmallCap 600 gets you about halfway to the actual value indices, and when you're leveraging that exposure, that ain't too shabby.
The factor loads are below, so it looks pretty good, especially if you are leveraging it.

https://www.portfoliovisualizer.com/fac ... sion=false
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

gtwhitegold wrote: Fri Jul 29, 2022 8:36 pm
cos wrote: Fri Jul 29, 2022 9:13 am
comeinvest wrote: Thu Jul 28, 2022 6:45 pm
gtwhitegold wrote: Wed Jul 27, 2022 9:47 pm From what I have seen, there aren't any good small cap value future contract options available. From what I have seen, your best bet in the small cap space is SMC, which tracks the S&P 600 index, so it has a quality tilt vice a junk tilt like the Russell 2000.
The S&P 600 follows the same methodology as the S&P500 just small cap instead of large cap, doesn't it? If yes, then it would have practically no factor tilt except very minimal filtering out of very low quality companies.
You'd be surprised. The S&P SmallCap 600 gets you about halfway to the actual value indices, and when you're leveraging that exposure, that ain't too shabby.
The factor loads are below, so it looks pretty good, especially if you are leveraging it.

https://www.portfoliovisualizer.com/fac ... sion=false
What does this tell us? Nowhere in the IJR description can I find that it tracks the S&P 600 small cap index.
SLY would be a S&P 600 ETF. I put it into PV and it has positive value and quality factors, but not as high as IJR.
For my education, what is the "reference point" of the PV factor loading calcs, i.e. what portfolio would have all factors zero?
Interesting enough, SLY outperformed IJR since 2006. SLY had 9.67% CAGR, IJR had 9.25%. I guess the value underperformance in that period shows even for small tilts, and the quality exposure couldn't fix it. So be careful what you wish for. EDIT: IJR is also a higher "small" factor than SLY. So we would need to consider crosscorrelations between the factors during the time frame contemplated, to arrive at conclusions about value.
mmKay
Posts: 3
Joined: Sat Jun 04, 2022 6:51 pm

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

Post by mmKay »

mmKay wrote: Sat Jun 04, 2022 7:34 pm I'm wondering if anyone using treasury futures for ITTs has considered adding short positions on other treasury futures (including longer term US and international) that are in contango (i.e., futures prices above cash / spot price) as a kind of a hedge?
mmKay wrote: Mon Jun 27, 2022 3:03 am Any expected price change from roll from peaks and troughs / zigs and zags should be included in the difference between the spot price and the price of the futures contract, right? (That is, along with carry and cost of borrowing.)
I reduced my exposure this strategy of shorting treasuries that are in contango as a hedge after I saw a post from skierincolorado in the HFEA thread late last year saying that expected changes in future yields could also be priced into current futures prices (and also mentioned in the post by comeinvest below, which I guess could be a cause of this contango. Having this extra unknown variable makes this strategy seem less likely to work. I'm keen to learn more about futures pricing and what causes this contango, but it seems like the way the academic literature goes of calculating carry (as outlined in this thread) is a better way to go if one wants to take short positions, like in the papers recently quoted by comeinvest:
comeinvest wrote: Wed Jul 20, 2022 1:14 am There are papers that show a significant excess return of "carry on the yield curve" investing. (There are other papers that say it's no actionable. I have not analyzed those papers yet.)
https://www.cfainstitute.org/en/researc ... 19-1628552
http://wp.lancs.ac.uk/fofi2018/files/20 ... tens-1.pdf

If true, one should be able to overlay a carry strategy over mHFEA by switching to the 2/3/5/7 or 3/5/7/10, etc. futures contract which has the highest carry. Of course everything is duration-adjusted, because you don't want to change your duration exposure. I think "carry" would be yield minus financing rate, or yield + current rolldown return - financing rate.

Of course it's not as simplistic as thinking you can pocket the entire difference between the current carries. Part of the difference reflects market expectations of rolldown and changes in rolldown returns, yield curve movements, etc. But there are significant excess returns as per the paper.
comeinvest - I would be interested to read the other papers that say it's not actionable if you had time to cite them. :happy
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

mmKay wrote: Sun Jul 31, 2022 1:39 am comeinvest - I would be interested to read the other papers that say it's not actionable if you had time to cite them. :happy
Can't find them right now, but I will post when I find them.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

mmKay wrote: Sun Jul 31, 2022 1:39 am
mmKay wrote: Sat Jun 04, 2022 7:34 pm I'm wondering if anyone using treasury futures for ITTs has considered adding short positions on other treasury futures (including longer term US and international) that are in contango (i.e., futures prices above cash / spot price) as a kind of a hedge?
mmKay wrote: Mon Jun 27, 2022 3:03 am Any expected price change from roll from peaks and troughs / zigs and zags should be included in the difference between the spot price and the price of the futures contract, right? (That is, along with carry and cost of borrowing.)
I reduced my exposure this strategy of shorting treasuries that are in contango as a hedge after I saw a post from skierincolorado in the HFEA thread late last year saying that expected changes in future yields could also be priced into current futures prices (and also mentioned in the post by comeinvest below, which I guess could be a cause of this contango. Having this extra unknown variable makes this strategy seem less likely to work. I'm keen to learn more about futures pricing and what causes this contango, but it seems like the way the academic literature goes of calculating carry (as outlined in this thread) is a better way to go if one wants to take short positions, like in the papers recently quoted by comeinvest:
comeinvest wrote: Wed Jul 20, 2022 1:14 am There are papers that show a significant excess return of "carry on the yield curve" investing. (There are other papers that say it's no actionable. I have not analyzed those papers yet.)
https://www.cfainstitute.org/en/researc ... 19-1628552
http://wp.lancs.ac.uk/fofi2018/files/20 ... tens-1.pdf

If true, one should be able to overlay a carry strategy over mHFEA by switching to the 2/3/5/7 or 3/5/7/10, etc. futures contract which has the highest carry. Of course everything is duration-adjusted, because you don't want to change your duration exposure. I think "carry" would be yield minus financing rate, or yield + current rolldown return - financing rate.

Of course it's not as simplistic as thinking you can pocket the entire difference between the current carries. Part of the difference reflects market expectations of rolldown and changes in rolldown returns, yield curve movements, etc. But there are significant excess returns as per the paper.
comeinvest - I would be interested to read the other papers that say it's not actionable if you had time to cite them. :happy
Another article on carry investing: https://seekingalpha.com/article/452821 ... strategies
I find the time frame that they tested too short, and the overall theoretical framework presented in this article not too convincing. The time period was a period of generally falling interest rates - not too surprising that you could have made money by buying forward contracts. The dynamic carry strategy sounds good, but isn't the fair price of most futures contracts in part based on short-term interest rates? At least those that have an asset as underlying. Not sure about VIX.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

comeinvest wrote: Sat Jul 30, 2022 5:15 am
gtwhitegold wrote: Fri Jul 29, 2022 8:36 pm
cos wrote: Fri Jul 29, 2022 9:13 am
comeinvest wrote: Thu Jul 28, 2022 6:45 pm
gtwhitegold wrote: Wed Jul 27, 2022 9:47 pm From what I have seen, there aren't any good small cap value future contract options available. From what I have seen, your best bet in the small cap space is SMC, which tracks the S&P 600 index, so it has a quality tilt vice a junk tilt like the Russell 2000.
The S&P 600 follows the same methodology as the S&P500 just small cap instead of large cap, doesn't it? If yes, then it would have practically no factor tilt except very minimal filtering out of very low quality companies.
You'd be surprised. The S&P SmallCap 600 gets you about halfway to the actual value indices, and when you're leveraging that exposure, that ain't too shabby.
The factor loads are below, so it looks pretty good, especially if you are leveraging it.

https://www.portfoliovisualizer.com/fac ... sion=false
What does this tell us? Nowhere in the IJR description can I find that it tracks the S&P 600 small cap index.
SLY would be a S&P 600 ETF. I put it into PV and it has positive value and quality factors, but not as high as IJR.
For my education, what is the "reference point" of the PV factor loading calcs, i.e. what portfolio would have all factors zero?
Interesting enough, SLY outperformed IJR since 2006. SLY had 9.67% CAGR, IJR had 9.25%. I guess the value underperformance in that period shows even for small tilts, and the quality exposure couldn't fix it. So be careful what you wish for. EDIT: IJR is also a higher "small" factor than SLY. So we would need to consider crosscorrelations between the factors during the time frame contemplated, to arrive at conclusions about value.
Nobody followed up on this subthread. I'm still thinking about using SMC within my mHFEA, in combination with /ES and international equity index futures. Thoughts?

I found this regarding risk-adjusted returns of SCV: viewtopic.php?p=6462128#p6462128 which compares SCV + bonds to S&P500 + bonds for the period from 1994 to now, but that is hardly representative because of the "value" disaster during that exact period.

Also: https://www.msci.com/research-and-insig ... volatility

Image
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

comeinvest wrote: Tue Aug 02, 2022 12:33 am
I found this regarding risk-adjusted returns of SCV: viewtopic.php?p=6462128#p6462128 which compares SCV + bonds to S&P500 + bonds for the period from 1994 to now, but that is hardly representative because of the "value" disaster during that exact period.
Bad backtest imo. The SCV:Market ratio expects SCV to outperform by 35% (67/50), it's not globally diversified, and if you even just change PV to "monthly" so that it includes all available data (March 1993), the portfolios' sharpe/sortino ratios only differ by 0.01.

AND the top 3 drawdowns for the SCV portfolio are significantly lower. That poster seems to have proven the opposite of what they meant to
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

Also sorry but, I'm struggling to find the part in this thread where ZF's exact duration is discussed.

I assume we get it from the delivery conditions:
The settlement is by delivery of the underlying asset — which is a U.S. Treasury note with an original term to maturity of not more than five years and three months and a remaining term to maturity of not less than four years and two months as of the first day of the delivery month.
Averaging those two durations, I get 4.7 years duration, but that assumes that "original term to maturity" means the beginning of the futures contract.

Interesting to see that Z3N has a much tighter spread of "remaining maturity at delivery" of 2 years + 9 months to 3 years.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

LazyOverthinker wrote: Tue Aug 02, 2022 3:34 pm Also sorry but, I'm struggling to find the part in this thread where ZF's exact duration is discussed.

I assume we get it from the delivery conditions:
The settlement is by delivery of the underlying asset — which is a U.S. Treasury note with an original term to maturity of not more than five years and three months and a remaining term to maturity of not less than four years and two months as of the first day of the delivery month.
Averaging those two durations, I get 4.7 years duration, but that assumes that "original term to maturity" means the beginning of the futures contract.

Interesting to see that Z3N has a much tighter spread of "remaining maturity at delivery" of 2 years + 9 months to 3 years.
I think you got it wrong.
- "Treasury note with an original term to maturity" is pretty explicit language. It has nothing to do with "the beginning of the futures contract".
- Your averaging calc does not make sense to me.
- I would concur that "The settlement is by delivery of the underlying asset". That means the interest rate duration you are exposed to by holding a futures contract is approximately the duration of the underlying CTD (cheapest to deliver) treasury bill, note, or bond. There is some optionality involved in the futures contract which means the CTD can sometimes change before the end of the contract, but I think the effect of that on effective duration of the futures contract is small to negligible (and would be very complicated to calculate).
- For durations of the futures contracts, go to https://www.cmegroup.com/tools-informat ... ytics.html and click on the individual contracts on the left side, and you will see the duration of the current CTD.
- There is some confusion as to the meaning of "futures yield", which does not seem to be the yield of the CTD. I have not wrapped my head around that yet.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Factor tilts with mHFEA

Post by comeinvest »

comeinvest wrote: Tue Aug 02, 2022 12:33 am
comeinvest wrote: Sat Jul 30, 2022 5:15 am
gtwhitegold wrote: Fri Jul 29, 2022 8:36 pm
cos wrote: Fri Jul 29, 2022 9:13 am
comeinvest wrote: Thu Jul 28, 2022 6:45 pm

The S&P 600 follows the same methodology as the S&P500 just small cap instead of large cap, doesn't it? If yes, then it would have practically no factor tilt except very minimal filtering out of very low quality companies.
You'd be surprised. The S&P SmallCap 600 gets you about halfway to the actual value indices, and when you're leveraging that exposure, that ain't too shabby.
The factor loads are below, so it looks pretty good, especially if you are leveraging it.

https://www.portfoliovisualizer.com/fac ... sion=false
What does this tell us? Nowhere in the IJR description can I find that it tracks the S&P 600 small cap index.
SLY would be a S&P 600 ETF. I put it into PV and it has positive value and quality factors, but not as high as IJR.
For my education, what is the "reference point" of the PV factor loading calcs, i.e. what portfolio would have all factors zero?
Interesting enough, SLY outperformed IJR since 2006. SLY had 9.67% CAGR, IJR had 9.25%. I guess the value underperformance in that period shows even for small tilts, and the quality exposure couldn't fix it. So be careful what you wish for. EDIT: IJR is also a higher "small" factor than SLY. So we would need to consider crosscorrelations between the factors during the time frame contemplated, to arrive at conclusions about value.
Nobody followed up on this subthread. I'm still thinking about using SMC within my mHFEA, in combination with /ES and international equity index futures. Thoughts?

I found this regarding risk-adjusted returns of SCV: viewtopic.php?p=6462128#p6462128 which compares SCV + bonds to S&P500 + bonds for the period from 1994 to now, but that is hardly representative because of the "value" disaster during that exact period.

Also: https://www.msci.com/research-and-insig ... volatility

https://i.imgur.com/lG2HbS7.png
UPDATE: The SMC futures seem to have no volume.
As an alternative to the /ES futures, the S&P Midcap 400 futures (EMD) have sufficient volume, and very favorable implied rates with negative spread to LIBOR:

https://www.cmegroup.com/trading/equity ... /main.html

Image

I think the midcap index is less tech heavy and more diversified in some sense. I think part of the reason for the low or negative spread is the opportunity cost of missed securities lending income that one would benefit from if one invested in the securities, and that the hedging counterparty benefits from. But the spread looks still surprisingly low to me. I don't know how much securities lending income the mid cap ETFs generate, but at almost 1% lower financing cost than the /ES futures (based on dividend estimates), it almost looks like a "free lunch" risk-free alpha opportunity. All that of course assumes that the dividend estimates are comparable.

Related thread examining SCV with bonds 1970 to 2016: viewtopic.php?t=226504
The risk-adjusted returns of SCV (but not rebalanced with bonds) are examined in this post: viewtopic.php?p=6572285#p6572285
Another thread related to combining SCV and bonds, but I think they don't consider the possibility of leveraged bonds: viewtopic.php?t=364950

Also referencing the posts from LazyOverthinker and strategist37 regarding SCV with mHFEA, so we have everything in one place:
viewtopic.php?p=6783453#p6783453
viewtopic.php?p=6783464#p6783464
viewtopic.php?p=6783483#p6783483
viewtopic.php?p=6783523#p6783523
viewtopic.php?p=6786209#p6786209
viewtopic.php?p=6786268#p6786268
viewtopic.php?p=6788947#p6788947
viewtopic.php?p=6790912#p6790912
viewtopic.php?p=6793594#p6793594
viewtopic.php?p=6793594#p6793594

I named the title of this subthread "Factor tilts with mHFEA".
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

comeinvest wrote: Tue Aug 02, 2022 6:21 pm - "Treasury note with an original term to maturity" is pretty explicit language. It has nothing to do with "the beginning of the futures contract".
- For durations of the futures contracts, go to https://www.cmegroup.com/tools-informat ... ytics.html and click on the individual contracts on the left side, and you will see the duration of the current CTD.
Ah I see yes it's quite clear in retrospect. I understood this for Z3N which allows for "original term to maturity" up to 7 years, but for some reason I was straining to interpret that duration-phrasing as part of the basket in ZF. That's how I got my average calculation of "(5.25 + 4.2~) / 2 = 4.7"
comeinvest wrote: Tue Aug 02, 2022 8:47 pm I think the midcap index is less tech heavy and more diversified in some sense. I think part of the reason for the low or negative spread is the opportunity cost of missed securities lending income that one would benefit from if one invested in the securities, and that the hedging counterparty benefits from... at almost 1% lower financing cost than the /ES futures
Very interesting. Securities-lending is a smart guess for the lower borrowing cost, and it's good to be skeptical of how free this looks
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: Factor tilts with mHFEA

Post by comeinvest »

I was curious what is the financing cost of equity index futures other than the /ES. I finally got around looking at the /EMD (S&P 400 midcap index).

I picked the period from 06/30/2020 to 07/29/2022, as that is about the longest time frame available at IB for the /EMD continuous futures chart.

Comparable ETF (IVOO) during the period from 06/30/2020 to 07/29/2022:

Image

Continuous futures contract as of 06/29/2020 pinpointed on the IB futures chart: 1741.44

Image

Continuous futures contract as of 08/01/2022 pinpointed on the IB futures chart: 2509.20

Image

1% adjustment needed for market movement from 06/29/2020 to 06/30/2020 for the future contract:

Image

No adjustment needed for market movement from 07/29/2022 to 08/01/2022:

Image

Capital growth of unfunded futures contract: 2509.2 / (1741.44 * 1.01) = 1.4266 i.e. $10,000 became $14,266

Compare to final balance of IVOO ($14,498)

=> Implied financing cost ca. 0.85% p.a. (rough linear interpolation, not a precise calculation)

Compare to CASHX CAGR of 0.24% => spread ca. 0.6%
This would be the all-in effective financing cost spread with respect to a comparable product that is readily available to an investor, reflecting i.e. not adjusted for the 0.1% expense ratio of IVOO and whatever IVOO earns from securities lending, and with respect to CASHX as proxy for the risk-free rate.

Potential sources of error:
- Confidence in IB continuous futures chart
- ETF premium or discount
- what exactly is the definition of CASHX


EDIT:
I was able to get a chart with daily continuous futures values, showing a capital growth of 1.4348 for the time span.
With the revised data point, I estimate a financing spread of ca. 0.5% - 0.6%.
Somehow I am unable to capture the intraday data at the end of the ETF trading day (3 p.m. Chicago time) from the futures chart, so there is probably some error left.

Image

Image
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

Re: Factor tilts with mHFEA

Post by skierincolorado »

comeinvest wrote: Thu Aug 04, 2022 5:07 am I was curious what is the financing cost of equity index futures other than the /ES. I finally got around looking at the /EMD (S&P 400 midcap index).

I picked the period from 06/30/2020 to 07/29/2022, as that is about the longest time frame available at IB for the /EMD continuous futures chart.

Comparable ETF (IVOO) during the period from 06/30/2020 to 07/29/2022:

Image

Continuous futures contract as of 06/29/2020 pinpointed on the IB futures chart: 1741.44

Image

Continuous futures contract as of 08/01/2022 pinpointed on the IB futures chart: 2509.20

Image

1% adjustment needed for market movement from 06/29/2020 to 06/30/2020 for the future contract:

Image

No adjustment needed for market movement from 07/29/2022 to 08/01/2022:

Image

Capital growth of unfunded futures contract: 2509.2 / (1741.44 * 1.01) = 1.4266 i.e. $10,000 became $14,266

Compare to final balance of IVOO ($14,498)

=> Implied financing cost ca. 0.85% p.a. (rough linear interpolation, not a precise calculation)

Compare to CASHX CAGR of 0.24% => spread ca. 0.6%
This would be the all-in effective financing cost spread with respect to a comparable product that is readily available to an investor, reflecting i.e. not adjusted for the 0.1% expense ratio of IVOO and whatever IVOO earns from securities lending, and with respect to CASHX as proxy for the risk-free rate.

Potential sources of error:
- Confidence in IB continuous futures chart
- ETF premium or discount
- what exactly is the definition of CASHX


EDIT:
I was able to get a chart with daily continuous futures values, showing a capital growth of 1.4348 for the time span.
With the revised data point, I estimate a financing spread of ca. 0.5% - 0.6%.
Somehow I am unable to capture the intraday data at the end of the ETF trading day (3 p.m. Chicago time) from the futures chart, so there is probably some error left.

Image

Image
This is great thank you.

Is it not worth doing this analysis for the small caps because the data does not necessarily capture the wide spreads?

How does the .6% spread reconcile with the negative spread you found on the current contract?
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: Factor tilts with mHFEA

Post by comeinvest »

skierincolorado wrote: Thu Aug 04, 2022 10:20 am This is great thank you.

Is it not worth doing this analysis for the small caps because the data does not necessarily capture the wide spreads?

How does the .6% spread reconcile with the negative spread you found on the current contract?
The /SMC (E-mini S&P 600 SmallCap) futures have zero volume, so I think they are not an option. I think they are for institutional traders who pre-arrange large trades with counterparties that are settled through CME, but don't take my word for it, I'm not a professional trader. I have not looked at the bid/ask spread during business hours, or at the spread of the calendar roll which at IB you can only check during roll time.

Regarding the negative spread that the roll data suggest: I don't know how to reconcile the data. The roll data suggests measurably lower financing cost for the /EMD than for the /ES. Perhaps the consensus dividend estimates that were the basis of the calendar rolls during my 2 year backtest were not aligned with the realized dividends. Part of the period was the Covid aftermath, maybe consensus dividend estimates were too low for mid caps. (Although I think the /ES had a very consistent 0.5% financing spread in recent years, last time I checked on the spglogal.com site. I was not able to find the /EMD continuous futures index on the spglobal site.) I think the small and mid caps have higher securities lending income potential, which would justify lower financing cost (before accounting for securities lending income). I would not necessarily trust this limited 2 year backtest.

I need free historical continuous futures data to do a longer backtest. There are some sites that have free historical futures data, but not backward-adjusted or forward-adjusted, and I'm too lazy to calculate the performance of the spliced contracts. I'm not even sure when the front month contract is switched in the data series, although one could probably find out by analyzing the data with enough time on hand.

The results are based on the assumption that the IB continuous futures charts are backward-adjusted and spliced correctly based on VWAP of the calendar rolls, or something similar. I have not verified that, and I know IB continuous charts used to have some issues that I came across in the past.

I also was not able to find a free historical continuous futures data source (spliced and backward or forward adjusted) for the Stoxx Europe 600 and the Nikkei 225 futures that I also use in my mHFEA implementation. I think you can use the Google Finance, Yahoo, or investing.com futures data and splice the contracts manually, but that is quite some work. I just verified that the Google Finance /ES chart https://www.google.com/finance/quote/ESY00:CME_EMINIS switched to the Sep 2022 contract on Jun 15th, but I can't find the historical contracts on Google Finance, so no way to determine the roll yields. (You would have to compare the raw data of two contracts on a day where both have volume.) Please let me know if somebody can find continuous historical data. Raw historical data is pretty useless for backtesting.
Maybe a TradeStation customer would have better data. Their methodology is described here: https://help.tradestation.com/09_01/tra ... tracts.htm
It looks like some brokers and charting software have continuous futures data.
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

Re: Factor tilts with mHFEA

Post by LazyOverthinker »

comeinvest wrote: Thu Aug 04, 2022 5:07 am
- what exactly is the definition of CASHX
Pretty sure CASHX is 1-month T-bill, as in the 2nd post here: viewtopic.php?t=244339.

I sympathize with your efforts here - so much data to juggle from retail-trader sources, and one or two sources of error could change the conclusion you arrive at. Sorry I can't be more help, I wouldn't trust myself with it
impatientInv
Posts: 349
Joined: Fri Sep 03, 2021 1:26 pm

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

Post by impatientInv »

Today was a tough day for Treasury future holders. :(
No individual stocks.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: Factor tilts with mHFEA

Post by comeinvest »

LazyOverthinker wrote: Fri Aug 05, 2022 5:15 am
comeinvest wrote: Thu Aug 04, 2022 5:07 am
- what exactly is the definition of CASHX
Pretty sure CASHX is 1-month T-bill, as in the 2nd post here: viewtopic.php?t=244339.

I sympathize with your efforts here - so much data to juggle from retail-trader sources, and one or two sources of error could change the conclusion you arrive at. Sorry I can't be more help, I wouldn't trust myself with it
I will eventually monitor performance and financing cost of my various futures based on my own calendar rolls. I started about 2 years ago. But it would be nice to have a longer data series of backward-adjusted futures data.
vsmt
Posts: 3
Joined: Tue Jun 08, 2021 12:10 pm

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

Post by vsmt »

Hi,

long time lurker and first timer poster here.

I'm running 30% international equity (VT) and 70% ITT (ZN / ZF) at 3 times leverage, i.e., 90%/210% at IBKR in a portfolio margin account.

I'm thinking about replacing the ZN and ZF futures with Vanguard's BNDW ETF to diversify internationally. I used IB's Risk Navigator tool to simulate the margin impact and it is about 40% margin to equity.

Could you please spare a moment and let me know your thoughts? Thanks
Last edited by vsmt on Tue Aug 09, 2022 2:57 pm, edited 1 time in total.
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

vsmt wrote: Tue Aug 09, 2022 2:10 pm I'm thinking about replacing the ZN and ZF futures with Vanguard's BNDX ETF to diversify internationally. I used IB's Risk Navigator tool to simulate the margin impact and it is about 40% margin to equity.

Could you please spare a moment and let me know your thoughts? Thanks
Hi, thanks for posting a new idea. Unfortunately, I would strongly advise you against it. First of all, BNDX is ex-US bond market, were you aware of that? Not only that, it includes corporate bonds - I'm even seeing Apple and ATT bonds in here on Vanguard's page (strange for something described as ex-US... maybe they mean it only excludes US treasuries?).

Corporate bonds are correlated with the stock market, making you less diversified, not more:

Image

Also BNDX has almost twice the duration-risk of ZF, but that you may already know. This duration-adjusted backtest shows BNDX has a notably lower sortino ratio because of the correlated drawdowns: https://www.portfoliovisualizer.com/bac ... mbol10=SHY

US treasury-bonds are a unique asset because of the US-dollar's position as the hegemonic reserve currency. I'd try not to overcomplicate it, plus the financing rate on futures are some of the fairest a retail investor can get.

If much of this information is new to you, I would strongly advise you do more reading before increasing your leverage. You're at a respectable level right now so it's fine, but understanding is key here. For instance, if you had just switched to 210% BNDX you'd have almost doubled your duration risk and drawdown risk

All that said, we're in unprecedented times and I have no idea what will happen if the US-dollar slowly loses its position as the hegemonic reserve currency due to international tension. Fun!
vsmt
Posts: 3
Joined: Tue Jun 08, 2021 12:10 pm

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

Post by vsmt »

LazyOverthinker wrote: Tue Aug 09, 2022 2:56 pm
vsmt wrote: Tue Aug 09, 2022 2:10 pm I'm thinking about replacing the ZN and ZF futures with Vanguard's BNDX ETF to diversify internationally. I used IB's Risk Navigator tool to simulate the margin impact and it is about 40% margin to equity.

Could you please spare a moment and let me know your thoughts? Thanks
Hi, thanks for posting a new idea. Unfortunately, I would strongly advise you against it. First of all, BNDX is ex-US bond market, were you aware of that? Not only that, it includes corporate bonds - I'm even seeing Apple and ATT bonds in here on Vanguard's page (strange for something described as ex-US... maybe they mean it only excludes US treasuries?).

Corporate bonds are correlated with the stock market, making you less diversified, not more:

Image

Also BNDX has almost twice the duration-risk of ZF, but that you may already know. This duration-adjusted backtest shows BNDX has a notably lower sortino ratio because of the correlated drawdowns: https://www.portfoliovisualizer.com/bac ... mbol10=SHY

US treasury-bonds are a unique asset because of the US-dollar's position as the hegemonic reserve currency. I'd try not to overcomplicate it, plus the financing rate on futures are some of the fairest a retail investor can get.

If much of this information is new to you, I would strongly advise you do more reading before increasing your leverage. You're at a respectable level right now so it's fine, but understanding is key here. For instance, if you had just switched to 210% BNDX you'd have almost doubled your duration risk and drawdown risk

All that said, we're in unprecedented times and I have no idea what will happen if the US-dollar slowly loses its position as the hegemonic reserve currency due to international tension. Fun!
Thanks for your prompt reply. Sorry, I did a typo and meant BNDW instead of BNDX.

Edit: I updated your link with BNDW. Unfortunately, the history is short: https://www.portfoliovisualizer.com/bac ... ol10=FRSHY
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

Also I wanted to share my conclusion about re-leveraging thresholds. I used Google Finance's intra-month data to simulate holding my aggressive "160% global SCV + 600% ZF (backtested as IEI)" idea since mid-2009, with frequent rebalancing and everything. I would only re-leverage upwards if the portfolio was up on the 22nd of the month (my payday). Simulating $100K leveraged to $760K, I found that re-leveraging when the 760K falls 2% is a nice, simple sweetspot for avoiding volatility decay. At 7.6x total leverage, that means re-leveraging if I lose 15%~ of the liquidity in my account.

Since mid-2009, with rebalancing, the leveraged portfolio fell 1% 30~ times, whereas it only fell 2%+ 7-10~ times, depending how you count extended crash events.

I then simulated the October 2021-June2022 drawdown with the recent daily data and found that re-leveraging at every 2% (15.2%) loss or every 1% (7.6%) loss would only shift your max-drawdown from 60% to 58%. My conclusion is that having any respectable "re-leverage on the way down" rule is more important than how neurotic/strict it is.

Repeating this experiment for a simple SPY+IEI or VT+IEI would probably be even easier and give a slightly different answer, if you feel so inclined. Just thought I'd share what it did for me.
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

vsmt wrote: Tue Aug 09, 2022 2:59 pm
Thanks for your prompt reply. Sorry, I did a typo and meant BNDW instead of BNDX.

Edit: I updated your link with BNDW. Unfortunately, the history is short: https://www.portfoliovisualizer.com/bac ... ol10=FRSHY
Oh I see, that's ok. The answers are still the same, BNDW includes corporate bonds and has a 0.35 correlation to stocks in that time period, whereas IEI has -0.15

Check out the "Assets" tab to see this

Image
LazyOverthinker
Posts: 133
Joined: Wed Jun 08, 2022 5:03 am

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

Post by LazyOverthinker »

Before I take a break from investing forums, I thought I'd share what I've learned about assets uncorrelated to equities/treasuries from Larry Swedroe's books on Factor Investing and "Reducing the Risk of Black Swans."

"Time-series momentum" is a risk premium uncorrelated (and often negatively correlated) to both stocks and treasuries, and it has become more and more democratized via ETFs like DBMF, KMLM, and CTA. Longer backtests can be run via mutual funds like QMHIX (AQR rules-based), PQTIX (more active), ASFYX and others. To get data past 2010 you have to reach out to the companies themselves, but they've done their "job" since 2000.

^This is the most interesting finding for me, since adding time-series-momentum to a mHFEA portfolio as a pure overlay increases CAGR multiplicatively while protecting against the 2022 drawdown. Past performance is not indicative of future returns; it could be an overcrowded trade now or any other number of things. There's a great extended discussion about this on the RationalReminder forum.

The "Re-insurance" premium is uncorrelated to stocks/bonds because natural disasters don't obey the stock market. Unfortunately, it's yet to be democratized and can be difficult to get into the SHRIX/SRRIX funds, especially on IBKR ($500K minimum).

Long-volatility via puts can be used to mitigate tail-events. It bleeds a little money most of the time but spikes tremendously in the most volatile times (Covid March 2020, GFC). More or less helpful depending on your leverage ratio and de-leveraging algorithm, since it could help prevent some of the worst de-leveraging events, but can increase tiny drawdowns not caused by extreme volatility. The only currently recommended fund is AVOLX - the CEO of AlphaArchitect uses it.
^EDIT: AVOLX seems to have interest-rate risk (2018 and 2022) because it has a strong focus on efficient collateral like box-spreads. Still, it could have a use as a capital-efficient risk-parity asset. In other words, with 12% SD since inception, perhaps it could "replace" a treasury allocation of the same volatility. Since it's a black box, I wouldn't trust it to replace more than 1/4th of my treasury position. The fund also suffers when volatility is "medium/choppy" but not extreme, since it goes long volatility via deep puts.

Lastly, there are quirky long-short strategies like the recently democratized FLSP ETF. No one really recommends these since they're vague black boxes with unproven risk limits.

All the above strategies are uncorrelated to stocks, treasuries, and each other.

Of course this all adds complexity, but their past performance has increased the sortino ratio of mHFEA. Your mileage may vary, but I found it very interesting.
Last edited by LazyOverthinker on Mon Aug 22, 2022 5:17 pm, edited 1 time in total.
HolyGrill
Posts: 52
Joined: Fri Feb 19, 2021 5:33 pm

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

Post by HolyGrill »

Hi everyone, after a read of the recommended article/book I noticed the ideal leverage ratio should be around or at 2x. From my understanding, they mentioned 2x for stock only. Currently, I am on 33/67 UPRO/TYD, is this considered 3x(300 in total) or 1x(If I only count stock)?

Thanks for any advice :happy
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

LazyOverthinker wrote: Fri Aug 19, 2022 4:39 pm The "Re-insurance" premium is uncorrelated to stocks/bonds because natural disasters don't obey the stock market. Unfortunately, it's yet to be democratized and can be difficult to get into the SHRIX/SRRIX funds, especially on IBKR ($500K minimum).
Scientists found that the Covid virus also doesn't follow the stock market ;)
(They didn't mention that the stock market followed the virus.)
Last edited by comeinvest on Sun Aug 21, 2022 3:56 am, edited 1 time in total.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

LazyOverthinker wrote: Fri Aug 19, 2022 4:39 pm Before I take a break from investing forums, I thought I'd share what I've learned about assets uncorrelated to equities/treasuries from Larry Swedroe's books on Factor Investing and "Reducing the Risk of Black Swans."

"Time-series momentum" is a risk premium uncorrelated (and often negatively correlated) to both stocks and treasuries, and it has become more and more democratized via ETFs like DBMF, KMLM, and CTA. Longer backtests can be run via mutual funds like QMHIX (AQR rules-based), PQTIX (more active), ASFYX and others. To get data past 2010 you have to reach out to the companies themselves, but they've done their "job" since 2000.

^This is the most interesting finding for me, since adding time-series-momentum to a mHFEA portfolio as a pure overlay increases CAGR multiplicatively while protecting against the 2022 drawdown. Past performance is not indicative of future returns; it could be an overcrowded trade now or any other number of things. There's a great extended discussion about this on the RationalReminder forum.

The "Re-insurance" premium is uncorrelated to stocks/bonds because natural disasters don't obey the stock market. Unfortunately, it's yet to be democratized and can be difficult to get into the SHRIX/SRRIX funds, especially on IBKR ($500K minimum).

Long-volatility can be used to mitigate tail-events. It bleeds a little money most of the time but spikes tremendously in the most volatile times (Covid March 2020, GFC). More or less helpful depending on your leverage ratio and de-leveraging algorithm, since it could help prevent some of the worst de-leveraging events, but can increase tiny drawdowns not caused by extreme volatility. The only currently recommended fund is AVOLX - the CEO of AlphaArchitect uses it.
^EDIT: AVOLX seems to have interest-rate risk (2018 and 2022), perhaps because it uses leverage to go long on volatility contracts, so it may not be a good hedge during inflation. Still, it could have a use as a capital-efficient risk-parity asset. In other words, with 12% SD since inception, perhaps it could "replace" a treasury allocation of the same volatility. Since it's a black box, I wouldn't trust it to replace more than 1/4th of my treasury position.

Lastly, there are quirky long-short strategies like the recently democratized FLSP ETF. No one really recommends these since they're vague black boxes with unproven risk limits.

All the above strategies are uncorrelated to stocks, treasuries, and each other.

Of course this all adds complexity, but their past performance has increased the sortino ratio of mHFEA. Your mileage may vary, but I found it very interesting.
There are plenty of critical comments in this forum about the reinsurance, the managed futures, and almost all other things that you mentioned. Some of the problems that most of them have in common are relatively short history, perceived safety (but no effective safety) due to no mark-to-market quotations, and generally the risk of parameter fitting once you increase the number of free variables of your opportunity set. You just increased the number of variables (assets) from 2 to at least 5, which might have been selected via cognitive "survivorship bias" from a larger opportunity set of maybe 20 out of which the 3 additional fitted the narrative best, probably all with a fraction of the backtested history of stocks and bonds, and most of them with little to no theoretical foundation why they would improve the risk/return of a stock/bond portfolio.
Post Reply