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

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

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

Post by skierincolorado »

comeinvest wrote: Thu Oct 21, 2021 1:36 am
skierincolorado wrote: Wed Oct 20, 2021 11:34 pm
comeinvest wrote: Wed Oct 20, 2021 11:23 pm I know skier is fond of the ZF; but if I wanted to hedge my bets by diversifying across ZF/ZN/TN/ZB (omitting UB for the moment as it seems like it's considered a sin in this thread) - I cannot decide should I use roughly equal dollar amounts (like NTSX), or roughly equal duration risk in each bucket for diversification. Thoughts?
You know what my answer will be haha. Would be a very solid plan IMO
You would favor equal duration risk per bucket. But then total returns would be dominated by the returns of the short end. Trying to look at it purely from a diversification point of view, considering the future might be different from the past.
Isn't that just an indirect way of saying the short end returns are better? :D From a diversification perspective too I think diversifying the risk is more important than diversifying the return.

Also, if the future is going to be different, you'd make a bunch of money in the interim. For the future to be different, the term premium on shorter durations would have to decrease relative to the term premium on longer durations. In other words, the spread between the 30y and 5y rate would grow. We'd make more money on the 5y / lose less money than we would on the 30y in the interim. So we'd be glad to own the 5y, then we'd end up in this weird unprecedented situation where the yield on the 5y was like .25% while the yield on the 30y was 2%+ (with 0% short term rates - not negative). At that point I think it would be pretty obvious to abandon the 5y - after we had reaped all the benefit.

We can already *guarantee* that over some reasonable time horizon ZF will have better returns than UB (on a per duration basis). This is because the current interest rate per duration is already much much higher. For ZF to do worse, the rate on ZF would have to keep rising faster than the rate on UB. Which just makes it a better and better buy and there is a limit to how much faster it can rise than UB because eventually the curve would go inverted. Like do we really see a situaiton 5 years from now where ZF has 10% interest rate and UB is 5%? Even if that happened, I'd be fine with it because I'd still just hold knowing it can't get more inverted forever.
Last edited by skierincolorado on Thu Oct 21, 2021 9:21 am, edited 1 time in total.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

cometqq wrote: Thu Oct 21, 2021 12:06 am
skierincolorado wrote: Wed Oct 20, 2021 8:24 pm
millennialmillions wrote: Wed Oct 20, 2021 7:07 pm
skierincolorado wrote: Wed Oct 20, 2021 5:45 pm
OK I think I've figured it out (again). While I was wrong about the CF changing, there has to be an explation of CMEs statement that 10M market value/face value of bonds is hedged with only 81 contracts. Or that profit and loss is just change in futures price - not change in invoice price. Hopefully we can all agree and consider the issue resolved. This statement is clear as day and it is repeated across several sources I've looked at. 81 contracts = 10M in bonds when CF is .81. Thats ~125k in bonds per contract.

I think I was thinking about the issue wrong. The delivery process doesn't matter until the end. WHen you enter into the contract what matters is the futures price. The futures price is marked to market every day until expiry. So if the futures price goes up $1000, the exchange literally takes $1000 from you, and gives it to the short. Thus up until expriy, the futures price is what matters and what determines your gains and losses. This is part of the contract and the process. The delivery process is what ties the futures price to real securities, so that the futures prices are actually based upon something real. But your gains and losses are based on changes in the futures price. If you buy a ZF when ZF is 124,000, and it falls to 123,000 the next day, you just lost $1000. Which is like obvious when you look at your account, but maybe not so obvious if you are thinking about the contract simply in terms of the delivery process. The delivery process is what ties it all together at the end and ties the futures contract to real bonds, but up until that expiration, what is being traded and marked to market is a 6% bond with 100k face value.

At first I was confused because I was like, well if my ZF goes down $1000 in futures price, why don't I just hold until expiry, and I'll only lose ~$800? But I can't because every day CME is marking it to market and I already lost $1000. They literally took it from me and gave it to the short. But then I started wondering, well why does the ZF go down by $1000, when the CTD bond only lost $800? Well because if it didn't, then the short position would be screwed at expiration, so the shorts sell until the ZF futures price is down the full $1,000. Then at the end of the day this $1000 lower settlement price is used to mark the shorts and longs positions to market.

If expiration was tomorrow, and if the CTD went down $800, and the futures price only went down $800, the longs would all sell. They'd be like great I'll take my $800 loss and walk because when it expires, I'm going to lose another $200. They would be forced to pay an invoice price of 101.2k for just 101k of bonds (hypothetical #s). No fair! So they sell. The futures price drops another $200. The invoice price drops to 101k. This selling action drives the futures price down the full $1,000. At that point the longs may reenter the market because there is no gain or loss at expiration anymore (ignoring financing), because the invoice price is actually equal to what is being delivered.

This explains the statement by CME and others that 10M in bonds is hedged by just 81 contracts when the CF is .81. Or taking the inverse, 81 contracts has equal exposure to 10M in bonds. Not 8.1M.

https://us.etrade.com/knowledge/library ... -to-market
I agree with the conclusion but think there is a simpler explanation. I edited my above post, restating here:

The futures price of ~$121,000 represents the price of a theoretical 5-year treasury with a $100,000 face value and 6% yield. Such a bond does not exist in practice. So for a given real treasury bond, the CF represents how much it is worth relative to this theoretical bond.

The exposure generated from the futures contract is not the same as the exposure generated by holding the CTD underlying security with the same face value. Instead, it is based on the theoretical asset mentioned above. Its price movement is equivalent to buying $121,000 worth of ITTs ("worth" meaning market value). This is verified by looking at constructor's graph and by the example from the CME paper skier provided.

So your exposure from holding a futures contract with $100,000 face value is greater than holding one underlying security with $100,000. At time of delivery, you couldn't just deliver a $100,000 face value treasury; you'd have to deliver more like $121,000 face value (in this case face value and market value are pretty close).
I agree with your first 3 paragraphs for sure. But I'm not sure about the last paragraph. What is actually delivered is definitely a 100k face value bond with market value also very close to 100k. And what is paid is also very close to 100k.

I think your first three paragraphs basically explain it though. What is being traded is the hypthetical 100k face with 6% yield. I think my post explains how this relates to the whole delivery process though and how the two are tied together. And why even though what is delivered and paid at expiration have market value very near 100k, the holder of the future gets the returns of ~121k in bonds for ZF currently.

I also confirmed in my IB account. I bought and sold at a particular futures price and my return was the difference. The invoice price didn't enter into the equation at all. Pretty simple and obvious in the end.
skier, you said that 10M in bonds is hedged by just 81 contracts when the CF is .81. So one contract hedges 100M/CF worth of bonds. Because CF is fixed for a contract, does this mean your exposure to bond is fixed (i.e. 100M/CF) regardless of when you enter into the futures contract.

But you also said that the exposure to bond is the market value in IBKR, which changes over time. My exposure to Bond will be different if I enter into a zf contract today versus tomorrow.

I feel this becomes even more confusing. Really need some help here.
Sorry I was being sloppy I was just trying to indicate the higher number. But yes it is the futures price that is the exposure which is some hypothetical theoretical 100k face value bond with 6% coupon. Since the CTD bond usually has a coupon near the market interest rate (much lower than 6%) 100k/CF for the CTD will usually get you quite close to the market value of a 6% coupon bond. So I was sloppily using them interchangeably.
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 »

hdas wrote: Thu Oct 21, 2021 8:59 am
Kbg wrote: Thu Oct 21, 2021 8:34 am
comeinvest wrote: Wed Oct 20, 2021 11:13 pm
Kbg wrote: Wed Oct 20, 2021 8:07 am Relevant article to the discussion

https://www.simplify.us/blog/efficient- ... e=hs_email
Every decade in those charts had overall falling interest rates, except the 70ies. The chart for the 70ies is the only one that shows LTT better than STT and part of ITT. Any concern that the results will reverse if the interest rate trend reverts to rising?

Image
That's a given. What these charts tell me is that ITTs are probably the place to be as they appear to be the most stable for someone who doesn't want to get into trying to time the market.
Quite the opposite, the non timing strategy would be to select the duration that matches your investment horizon. H
The risk-adjusted return of the 5y has historically been far superior to the 30y, I'll take my chances with the 5y. That's not market timing. I'm not "timing" anything - there is no "timing" involved. I'm simply picking the better investment. Am I "bond picking" the way some people "stock pick" - well maybe, but I would argue no since the 5y is much more representative of the total bond market return. If somebody wanted to take a leveraged total bond position that would be very similar to what I'm doing and have a very similar effective duration.
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 »

Bentonkb wrote: Thu Oct 21, 2021 8:02 am
skierincolorado wrote: Wed Oct 20, 2021 5:45 pm This explains the statement by CME and others that 10M in bonds is hedged by just 81 contracts when the CF is .81. Or taking the inverse, 81 contracts has equal exposure to 10M in bonds. Not 8.1M.
I have a copy of the CME paper, but can you provide references to the "others" that you mention?

The example in the CME paper does answer the question unambiguously if it is correct. I've been thinking it was an error because the CME paper doesn't seem to be very well written.

Your explanation based on how the MTM process is based on futures price, which is inflated by the CF, matches my thinking six months ago when I first started looking into it. I've been going back and forth ever since. It is obviously a difficult thing to figure out, given the discussion. I'm fully prepared to accept that I need to switch back to my previous position. We have about a month before the next roll period. Hopefully I won't need all that time to get it figured out to my satisfaction. :?

On the topic of changing my point of view . . .
This thread and some PV models have convinced me to switch from ZB a shorter duration contract and some more leverage. Unfortunately, it looks like this might be pretty bad timing. What are your thoughts on the short term outlook for ZB vs ZF? I've read that the market has priced in two rate increases next year of 25 basis points each. Seems like a reasonable assumption at this point. The question, then, is whether this will result in a curve flattening or just an a 50 basis point rise across the whole curve. Curve flattening would favor sticking with ZB until things settle down. Of course, if the change is fully priced in it won't matter.
Equation 5 on page 11 here for example: http://www.yieldcurve.com/mktresearch/l ... utures.pdf. Nominal value is same as face value.

I read several others haha - they all said the same thing.

The market has priced in the rate increases. I'll use the 30y and 5y for comparison. The 30y is 2.14% today. The 5y is 1.16% today.

Let's say that 1 year from now, the 30y is still 2.14%, but the 5y rises .3% to 1.46%. What would our return be if we had bought 4 or 5 5y bonds vs a 30y bond today?

A year from now we would own a 4y bond and a 29y bond, with a yields of 1.26% and 2.13% respectively. The price of the 4y bond would have decreased .39%. The price of the 30y would have increased .21%.

Then add in the coupon our total return for the 5y would be 1.16% - .39% = .77%. The total return for the 30 would be 2.14% + .39% = 2.53%.

But if we were buying 5y bonds we would buy ~4-5x more than for 30y bonds. So multiply the .77% * 4 = 3.08%

The 5y rate increased while the 30y rate did not. And yet we still made more money on the 5y. This is a substantial flattening of the curve past 5 years, and a substantial decrease in the spread between the 30y and 5y interest rate. If this continued, the yield curve would soon become inverted with a higher 5y interest rate than the 30y rate.


Another example with more aggressive rate increases. The 30y rises .13% from 2.14% to 2.27%. The 5y rises .5% from 1.16% to 1.66%. Getting pretty flat past 5 years (1.66% vs 2.27%) so this would be a pretty extreme decrease in the spread between the 30y and the 5y. The 5y rate rose a lot more than the 30y rate.

We would then have a 29 y bond with a yield around 2.26% and a 4y bond with a yield around 1.46%.

The 29y bond price would have decreased 2.54%. The 4y bond price would have decreased 1.16%.

Adding in the coupons we get -2.54% + 2.14% = -.4% for the 30y and -1.16% + 1.16% = 0% for the 5 year.


The 5y rate rose much more than the 30y rate (.5% vs .16%) and yet we still made more money on the 5y. This is a substantial flattening of the curve past 5 years, and a substantial decrease in the spread between the 30y and 5y interest rate. If this continued, the yield curve would soon become inverted with a higher 5y interest rate than the 30y rate.


These calculations aren't perfect, because for example we roll once a quarter not once a year, but they illustrate the point. The 5y rate can rise a lot faster than the 30y rate, and we are still better off with the 5y. Mainly because we are rolling down the curve towards the 4y and the curve is quite steep.


I think the market expectation, based on interest rate forwards, is somewhere between these two scenarios. Maybe a .3-.4% increase on the 5 y and a .05-.1% increase on the 30y.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

Bentonkb wrote: Thu Oct 21, 2021 8:02 am Unfortunately, it looks like this might be pretty bad timing. What are your thoughts on the short term outlook for ZB vs ZF? I've read that the market has priced in two rate increases next year of 25 basis points each. Seems like a reasonable assumption at this point. The question, then, is whether this will result in a curve flattening or just an a 50 basis point rise across the whole curve. Curve flattening would favor sticking with ZB until things settle down. Of course, if the change is fully priced in it won't matter.
skierincolorado wrote: Thu Oct 21, 2021 10:01 am On the topic of changing my point of view . . .
h
Last edited by hdas on Tue Oct 26, 2021 6:40 pm, edited 1 time in total.
....
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

hdas wrote: Thu Oct 21, 2021 10:37 am
Bentonkb wrote: Thu Oct 21, 2021 8:02 am Unfortunately, it looks like this might be pretty bad timing. What are your thoughts on the short term outlook for ZB vs ZF? I've read that the market has priced in two rate increases next year of 25 basis points each. Seems like a reasonable assumption at this point. The question, then, is whether this will result in a curve flattening or just an a 50 basis point rise across the whole curve. Curve flattening would favor sticking with ZB until things settle down. Of course, if the change is fully priced in it won't matter.
skierincolorado wrote: Thu Oct 21, 2021 10:01 am On the topic of changing my point of view . . .
I rather go to the empirical and see what actually happens when flattening, this is the most recent episode (as of last friday), 7 months going, let us see how this evolves.

Image

ZN, ZF leveraged using DV01 to match UB, which in turns matches VUSTX almost perfectly. 90% funded with VFISX, which is very generous to the leveraged futures position.
Yes, longer durations can do better during rapid flattening. But shorter durations do better during modest flattening, slight flattening, and un-flattening (ie 90% of the time).

For example, from January 31 2013 to October 31 2018 the 5y rose from .88 to 2.98 while the 30y rose from 3.17 to 3.39. This represented a extreme flattening of the spread from 2.29 to .41. We went from quite a large spread (2.29) to a very small spread (.41). The curve went from very steep to nearly flat.

So the 30y returned better in this period right?

Wrong. EDV returned 18.7% while a 5x leveraged position in VFITX (borrowing at the T-Bill rate + .25%) returned 19.3% with a lower max-drawdown. Yes we could tweak the borrowing cost a bit higher, like T-Bill + .2%, and VGIT would do a bit worse. But this is to illustrate a point that even during an extreme flattening event, VGIT can do very similar to EDV.

https://www.portfoliovisualizer.com/bac ... on3_2=-200


What about a large, but less extreme, period of flattening? Like Jan 31 2015 to October 31 2018. The 5y rose from 1.33 to 2.98. The 30y rose from 2.25 to 3.39. The spread durng this period decreased by .51 - a substantial flattening in 3.5 years.

Again we see the 5y with higher return and lower max-draw:

https://www.portfoliovisualizer.com/bac ... on3_2=-400


These are perios of flattening. The curve can't flatten forever. During periods of less flattening, no-flattening, and unflattening, the outperformance of the 5y is even greater.
Last edited by skierincolorado on Thu Oct 21, 2021 4:36 pm, edited 3 times in total.
Kbg
Posts: 355
Joined: Thu Mar 23, 2017 11:33 am

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

Post by Kbg »

hdas wrote: Thu Oct 21, 2021 8:59 am
Kbg wrote: Thu Oct 21, 2021 8:34 am
comeinvest wrote: Wed Oct 20, 2021 11:13 pm
Kbg wrote: Wed Oct 20, 2021 8:07 am Relevant article to the discussion

https://www.simplify.us/blog/efficient- ... e=hs_email
Every decade in those charts had overall falling interest rates, except the 70ies. The chart for the 70ies is the only one that shows LTT better than STT and part of ITT. Any concern that the results will reverse if the interest rate trend reverts to rising?

Image
That's a given. What these charts tell me is that ITTs are probably the place to be as they appear to be the most stable for someone who doesn't want to get into trying to time the market.
Quite the opposite, the non timing strategy would be to select the duration that matches your investment horizon. H
Fair point, but that's like trying to optimize when to take social security. Somehow, I just don't think I want to know when I'm going to die right now. :-)
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 »

November 30 2002 to October 31 2004:

5y rose from 3.28 to 3.30. 30y fell from 5.23 to 4.87. A tightening of .37% in 2 years.


https://www.portfoliovisualizer.com/bac ... on3_2=-130
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

hhhhh
Last edited by hdas on Tue Oct 26, 2021 6:41 pm, edited 2 times in total.
....
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

skierincolorado wrote: Thu Oct 21, 2021 11:29 am November 30 2002 to October 31 2004:

5y rose from 3.28 to 3.30. 30y fell from 5.23 to 4.87. A tightening of .37% in 2 years.


https://www.portfoliovisualizer.com/bac ... on3_2=-130
h
Last edited by hdas on Tue Oct 26, 2021 6:40 pm, edited 1 time in total.
....
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

Kbg wrote: Thu Oct 21, 2021 11:13 am Fair point, but that's like trying to optimize when to take social security. Somehow, I just don't think I want to know when I'm going to die right now. :-)
h
Last edited by hdas on Tue Oct 26, 2021 6:41 pm, edited 1 time in total.
....
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

hdas wrote: Thu Oct 21, 2021 11:32 am
skierincolorado wrote: Thu Oct 21, 2021 11:04 am Yes, longer durations can do better during rapid flattening. But shorter durations do better during modest flattening, slight flattening, and un-flattening (ie 90% of the time).

For example, from January 31 2013 to October 31 2018 the 5y rose from .88 to 2.98 while the 30y rose from 3.17 to 3.39. This represented a extreme flattening of the spread from 2.29 to .41. We went from quite a large spread (2.29) to a very small spread (.41). The curve went from very steep to nearly flat.

So the 30y returned better in this period right?

Wrong. EDV returned 11.87% while a 5x leveraged position in VGIT (borrowing at the T-Bill rate + .05%) returned 11.93% with a lower max-drawdown. Yes we could tweak the borrowing cost a bit higher, like T-Bill + .2%, and VGIT would do a bit worse. But this is to illustrate a point that even during an extreme flattening event, VGIT can do very similar to EDV.

https://www.portfoliovisualizer.com/bac ... on3_2=-200


What about a large, but less extreme, period of flattening? Like Jan 31 2015 to October 31 2018. The 5y rose from 1.33 to 2.98. The 30y rose from 2.25 to 3.39. The spread durng this period decreased by .51 - a substantial flattening in 3.5 years.

Again we see the 5y with higher return and lower max-draw:

https://www.portfoliovisualizer.com/bac ... on3_2=-400


These are perios of flattening. The curve can't flatten forever. During periods of less flattening, no-flattening, and unflattening, the outperformance of the 5y is even greater.
Mr. Skier, the overarching point I'm making is that the main component of this strategy is the play in the yield curve movement. If you look at the peak to trough of the last flattening cycle, you'll see that all the risk adjusted advantages disappear.

Image

NOTE: While in the image of the chart the name of the series says "EDV", the actual time series is VUSTX, as is appropriate.
You can press link in PV and then provide the link so others can verify the results. Even using VUSTX I cannot replicate your results in PV: VFITX already includes a 0.2% fee. We are leveraging this fee 3.75x. This is the same as leveraging at T-Bill + .2%. Actually it's a bit over .2% because the .2% is on the entire 375%, not just the 275% that is leveraged. In reality, that's like leveraging at T-Bill + .28%.

So leveraging 3.75x with leverage costs of T-Bill + .28%, I still find VFITX to perform better than VUSTX, during a major period of tightening where the curve got much flatter.

And VUSTX has much shorter duration than EDV... I don't object as strongly to VUSTX. And yet still, during a period of major tightening, VFITX still does better. Let's be clear: the 30y rate fell 0.3% during this period, while the 5y rate rose 1.71%. That's an extreme flattening of 2.01%.

I don't think I can predict exactly when and how much the curve will flatten. And even if I could, the 5y did just as well as VUSTX, even during an extreme flattening event. And infinitely better during periods of no-flattening or un-flattening. The 5y DESTROYED the 30y in 2020. What if COVID had happened in 2017? Someone in the VUSTX would have missed all the beautiful unflattening.

I don't want to try to time the market like that. If I knew when tightening would occur, and by how much, *maybe* it would be better to be in VUSTX in those periods (although from 2011-2018 it was a tie despite the extreme tightening). I don't think I can predict when tightening will happen, or when COVID will happen, or when other economic shocks will happen - so I stay in the 5y.


https://www.portfoliovisualizer.com/bac ... on3_2=-275
Last edited by skierincolorado on Thu Oct 21, 2021 11:53 am, edited 1 time in total.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

hhhhh
Last edited by hdas on Tue Oct 26, 2021 6:42 pm, edited 1 time in total.
....
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

hdas wrote: Thu Oct 21, 2021 11:53 am
skierincolorado wrote: Thu Oct 21, 2021 11:45 am
hdas wrote: Thu Oct 21, 2021 11:32 am
skierincolorado wrote: Thu Oct 21, 2021 11:04 am Yes, longer durations can do better during rapid flattening. But shorter durations do better during modest flattening, slight flattening, and un-flattening (ie 90% of the time).

For example, from January 31 2013 to October 31 2018 the 5y rose from .88 to 2.98 while the 30y rose from 3.17 to 3.39. This represented a extreme flattening of the spread from 2.29 to .41. We went from quite a large spread (2.29) to a very small spread (.41). The curve went from very steep to nearly flat.

So the 30y returned better in this period right?

Wrong. EDV returned 11.87% while a 5x leveraged position in VGIT (borrowing at the T-Bill rate + .05%) returned 11.93% with a lower max-drawdown. Yes we could tweak the borrowing cost a bit higher, like T-Bill + .2%, and VGIT would do a bit worse. But this is to illustrate a point that even during an extreme flattening event, VGIT can do very similar to EDV.

https://www.portfoliovisualizer.com/bac ... on3_2=-200


What about a large, but less extreme, period of flattening? Like Jan 31 2015 to October 31 2018. The 5y rose from 1.33 to 2.98. The 30y rose from 2.25 to 3.39. The spread durng this period decreased by .51 - a substantial flattening in 3.5 years.

Again we see the 5y with higher return and lower max-draw:

https://www.portfoliovisualizer.com/bac ... on3_2=-400


These are perios of flattening. The curve can't flatten forever. During periods of less flattening, no-flattening, and unflattening, the outperformance of the 5y is even greater.
Mr. Skier, the overarching point I'm making is that the main component of this strategy is the play in the yield curve movement. If you look at the peak to trough of the last flattening cycle, you'll see that all the risk adjusted advantages disappear.

Image

NOTE: While in the image of the chart the name of the series says "EDV", the actual time series is VUSTX, as is appropriate.
You can press link in PV and then provide the link so others can verify the results. Even using VUSTX I cannot replicate your results in PV: VFITX already includes a 0.2% fee. We are leveraging this fee 3.75x. This is the same as leveraging at T-Bill + .2%. Actually it's a bit over .2% because the .2% is on the entire 375%, not just the 275% that is leveraged. In reality, that's like leveraging at T-Bill + .28%.

So leveraging 3.75x with leverage costs of T-Bill + .28%, I still find VFITX to perform better than VUSTX, during a major period of tightening where the curve got much flatter.

And VUSTX has much shorter duration than EDV... I don't object as strongly to VUSTX. And yet still, during a period of major tightening, VFITX still does better. Let's be clear: the 30y rate fell 0.3% during this period, while the 5y rate rose 1.71%. That's an extreme flattening of 2.01%.

I don't think I can predict exactly when and how much the curve will flatten. And even if I could, the 5y did just as well as VUSTX, even during an extreme flattening event. And infinitely better during periods of no-flattening or un-flattening. The 5y DESTROYED the 30y in 2020. What if COVID had happened in 2017? Someone in the VUSTX would have missed all the beautiful unflattening.

https://www.portfoliovisualizer.com/bac ... on3_2=-275
Here's the link https://www.portfoliovisualizer.com/bac ... ion3_2=100
OK looks like our start and end dates were just a couple months different. Thanks.

Nevertheless, this is a period of extreme flattening and VUSTX and leveraged VFITX still tied (with financing cost of T-Bill + .28%). What if COVID had happened in 2017 and you missed all the sweet unflattening? If you think you can predict when the next COVID will happen, go for it!

Just continuing your timeseries to present, we see leverage VFITX (financing cost T-Bill + .28%) up 51.4% and VUSTX up 26.9%. Ouch!

https://www.portfoliovisualizer.com/bac ... on3_2=-275
Last edited by skierincolorado on Thu Oct 21, 2021 11:59 am, edited 1 time in total.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

hhhhh
Last edited by hdas on Tue Oct 26, 2021 6:42 pm, edited 1 time in total.
....
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

skierincolorado wrote: Thu Oct 21, 2021 11:56 am OK looks like our start and end dates were just a couple months different. Thanks.

Nevertheless, this is a period of extreme flattening and VUSTX and leveraged VFITX still tied (with financing cost of T-Bill + .28%). What if COVID had happened in 2017 and you missed all the sweet unflattening? If you think you can predict when the next COVID will happen, go for it!

Just continuing your timeseries to present, we see leverage VFITX (financing cost T-Bill + .28%) up 51.4% and VUSTX up 26.9%. Ouch!

https://www.portfoliovisualizer.com/bac ... on3_2=-275
h
Last edited by hdas on Tue Oct 26, 2021 6:42 pm, edited 1 time in total.
....
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

hdas wrote: Thu Oct 21, 2021 11:59 am Furthermore, I believe this indeed was a very valid strategy until operation twist in 2011 and the race to 0 interest rates changed the game. I derive this conclusion from the following exercise:

- Look at the index of long VFITX, short VUSTX
- Leverage VFITX to match the volatility of VUSTX, using a 3 month rolling window.
- Rebalance daily

Here's the result:

Image
Yes the relative outperformance of VFITX since 2011, during a period of rapid flattening, is much less. It's still done better since 2011. However, I don't base my investments on 8 or 10 year periods. The curve was quite steep in 2011. It is quite flat beyond 5 or 10 years today. This is not the time to be trying to market time and predict that this will be one of the *only* periods in history that LTT will do better than ITT.
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 »

hdas wrote: Thu Oct 21, 2021 12:04 pm
skierincolorado wrote: Thu Oct 21, 2021 11:56 am OK looks like our start and end dates were just a couple months different. Thanks.

Nevertheless, this is a period of extreme flattening and VUSTX and leveraged VFITX still tied (with financing cost of T-Bill + .28%). What if COVID had happened in 2017 and you missed all the sweet unflattening? If you think you can predict when the next COVID will happen, go for it!

Just continuing your timeseries to present, we see leverage VFITX (financing cost T-Bill + .28%) up 51.4% and VUSTX up 26.9%. Ouch!

https://www.portfoliovisualizer.com/bac ... on3_2=-275
I don't know what you were doing in the summer of 2018, but the term premium evaporated and all of the sudden after so many hikes we were all making some money in the short end. So THAT'S when you leverage ZF, and ride it until the fed and the term premium tell you what to do.

This has nothing to do with predictions, but everything to do with the spread.
Yes you might have switched to the short end in summer 2018 before COVID. What if COVID happened in 2017? The real un-flattening happened in 2020 due to massive QE driving down short rates and eventually pushing up inflation expectations and long rates. What if inflation had taken of 2011-2018 and long rates had crept up? Aruably that is what the Fed was aiming for but it never materialized.

You're describing a very complicated strategy. I've heard you articulate several different rules for when to switch between short and long. I suggest you backtest them thoroughly and implement it.

I acknowledge 2011-2018 was one of the only (probably *the* only) period where VUSTX performed similarly to leveraged VFITX. I concede this but don't care. This was a period of extreme flattening of the curve and I don't care to try to time the market like that. I don't believe that this flattening was entirely predictable. The flattening could have been slightly less extreme. COVID could have happened in 2017 and the unflattening we saw in 2020, and large outperformance of ITT vs LTT, could have happened in 2017.

I've read several papers describing dynamic allocation rules based strategies for where to allocate on the yield curve. None of them go past 10 years of duration. Switching allocaitons between 2y 5y and 10y (never longer than 10) has yielded some *slight* historical outperformance based on a dynamic rules based strategy. These are Hedge Fund strategies with infinite resources and expertise. If you think you can do it, backtest it and show us.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

skierincolorado wrote: Thu Oct 21, 2021 12:06 pm I acknowledge 2011-2018 was one of the only (probably *the* only) period where VUSTX performed similarly to leveraged VFITX.
h
Last edited by hdas on Tue Oct 26, 2021 6:43 pm, edited 1 time in total.
....
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

hdas wrote: Thu Oct 21, 2021 1:39 pm
skierincolorado wrote: Thu Oct 21, 2021 12:06 pm I acknowledge 2011-2018 was one of the only (probably *the* only) period where VUSTX performed similarly to leveraged VFITX.
I think we can say that since Nov 2010 the strategy hasn't worked as intended, and 11 years is a long time. I just noticed that the DV01 hedge ratios are actually different at the beginning of the UltraBond existance (see table). What this means is that the performance is vastly overstated with 3.75x. This is another reason, (besides historical financing costs) to really distrust your "backtests" from PV, if one wants to be rigorous about comparing apples to apples, the hedge ratio has to be dynamically adjusted at the moment, you can go as granular as you want, this is just quarterly.

Code: Select all

Mar 2010	4:1
June 2010	4:1
Sept 2010	4:1
Dec 2010	4:1
Mar 2011	4:1
June 2011	4:1
Sept 2011	4:1
Dec 2011	4:1
Mar 2012	4:1
June 2012	5:1
June 2012	5:1
Sept 2012	5:1
Dec 2012	5:1
Mar 2013	5:1
June 2013	5:1
Sept 2013	5:1
Dec 2013	4:1
Mar 2014	4:1
Jun 2014	4:1
Sep 2014	5:1
Dec 2014	5:1
Mar 2015	5:1
Jun 2015	6:1
Sep 2015	5:1
Dec 2015	5:1
Mar 2016	5:1
Jun 2016	6:1
Sep 2016	6:1
Dec 2016	7:1
Mar 2017	6:1
Jun 2017	6:1
Sep 2017	6:1
Dec 2017	6:1
Mar 2018	6:1
Jun 2018	6:1
Sep 2018	6:1
Dec 2018	6:1
Mar 2019	6:1
Jun 2019	6:1
Sep 2019	6:1
Dec 2019	6:1
Mar 2020	6:1
Jun 2020	6:1
Sep 2020	6:1
Dec 2020	6:1
Mar 2021	6:1
Jun 2021	6:1
Sep 2021	6:1
Dec 2021	6:1

Since I plan to do a fixed ratio going forward, that is what I should test historically as well. If a dynamic ratio was better historically, maybe I would consider a dynamic ratio going forward. The results are what they are - if fixed ratios worked better historically then I see no reason that would not continue. And vice versa.

I wouldn't say that it has not worked as intended since 2010. I would simply say that the outperformance has been minimal since 2010. Would you say that stocks were a bad investment because they had negative returns from 2000 to 2009? No. And it's not like ITT did worse than LTT from 2010-present. ITT were solidly better than LTT 2010-present. Only during the worst period of 2011-2018 did LTT even "tie" ITT. \This is because of the tightening and flattening - which can't continue forever as we saw from 2018-present.

Even duing this "bad" period for ITT... they still solidly outperformed LTT:

https://www.portfoliovisualizer.com/bac ... on3_1=-250


We really have to cherry pick the worst period for ITT to even find a period where LTT were tied. 2011-2018 is a "tie" and that's as bad a period as you can get for ITT. 2010-present is still a solid win for ITT. And you'd have to predict all of this to take advantage - and even then you're not beating ITT you're just getting equal. And then if COVID happened in 2017 you'd miss all the great returns for ITT.

https://www.portfoliovisualizer.com/bac ... on3_1=-250
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

skierincolorado wrote: Thu Oct 21, 2021 1:49 pm Since I plan to do a fixed ratio going forward, that is what I should test historically as well. If a dynamic ratio was better historically, maybe I would consider a dynamic ratio going forward. The results are what they are - if fixed ratios worked better historically then I see no reason that would not continue. And vice versa.
h
Last edited by hdas on Tue Oct 26, 2021 6:43 pm, edited 1 time in total.
....
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

hdas wrote: Thu Oct 21, 2021 2:07 pm
skierincolorado wrote: Thu Oct 21, 2021 1:49 pm Since I plan to do a fixed ratio going forward, that is what I should test historically as well. If a dynamic ratio was better historically, maybe I would consider a dynamic ratio going forward. The results are what they are - if fixed ratios worked better historically then I see no reason that would not continue. And vice versa.
I don't think you seem to understand. In order to compare the two series, you need to take into account their interest rate sensitivity, you capture this by calculating the DV01 of each and then the ratio. Otherwise is just a hack job of naïve backtest overfitting.
It's not overfitting anything. It's not like I can just pick a bigger ratio and improve the return without also increase the max-draw and stdev. When leveraged VFITX has better CAGR, and lower stdev and max-draw, there's nothing "overfit" about that. It's simply better.

I could have decided in 2011 I was going to replace my LTT at a fixed ratio with ITT. Done. This is the actual improved return I would have gotten.

Likewise, I am deciding today that I will use a fixed ratio. I will not adjust for DV01 perfectly. This is therefore what I can, should, and have backtested across various historical periods including back to 1955.

You can use a dynamic ratio if you would like. I will use a fixed ratio because it is simpler and because I very easily could have used a fixed ratio in real life and will be using a fixed ratio going forward.

If fixed ratios are better, and somehow are giving me better results, that's an argument for using a fixed ratio in real life going forward .. not changing how we backtest. I can actually invest in real life using a fixed ratio. It's therefore a very realistic assumption and since it is what I intend to actually do, it's exactly how I *should* be backtesting. If using a dynamic ratio gives better results, which it probably does, then maybe I would consider using a dynamic ratio in my backtests and my actual investing. Although I probably wouldn't bother.

I very much doubt that it makes a significant difference to the sharpe ratio anyways.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

skierincolorado wrote: Thu Oct 21, 2021 9:13 am
comeinvest wrote: Thu Oct 21, 2021 1:36 am
skierincolorado wrote: Wed Oct 20, 2021 11:34 pm
comeinvest wrote: Wed Oct 20, 2021 11:23 pm I know skier is fond of the ZF; but if I wanted to hedge my bets by diversifying across ZF/ZN/TN/ZB (omitting UB for the moment as it seems like it's considered a sin in this thread) - I cannot decide should I use roughly equal dollar amounts (like NTSX), or roughly equal duration risk in each bucket for diversification. Thoughts?
You know what my answer will be haha. Would be a very solid plan IMO
You would favor equal duration risk per bucket. But then total returns would be dominated by the returns of the short end. Trying to look at it purely from a diversification point of view, considering the future might be different from the past.
Isn't that just an indirect way of saying the short end returns are better? :D From a diversification perspective too I think diversifying the risk is more important than diversifying the return.

Also, if the future is going to be different, you'd make a bunch of money in the interim. For the future to be different, the term premium on shorter durations would have to decrease relative to the term premium on longer durations. In other words, the spread between the 30y and 5y rate would grow. We'd make more money on the 5y / lose less money than we would on the 30y in the interim. So we'd be glad to own the 5y, then we'd end up in this weird unprecedented situation where the yield on the 5y was like .25% while the yield on the 30y was 2%+ (with 0% short term rates - not negative). At that point I think it would be pretty obvious to abandon the 5y - after we had reaped all the benefit.

We can already *guarantee* that over some reasonable time horizon ZF will have better returns than UB (on a per duration basis). This is because the current interest rate per duration is already much much higher. For ZF to do worse, the rate on ZF would have to keep rising faster than the rate on UB. Which just makes it a better and better buy and there is a limit to how much faster it can rise than UB because eventually the curve would go inverted. Like do we really see a situaiton 5 years from now where ZF has 10% interest rate and UB is 5%? Even if that happened, I'd be fine with it because I'd still just hold knowing it can't get more inverted forever.
To be clear, I don't disagree with you. I'm just playing devil's advocate, and examining the strategy from every angle. While I don't disagree with your findings and your strategy, I think it's not a guarantee. If the curve become quite shallow at the short end, then steepens a bit on the long end, and stays that way for a long time, then I think LTT would be better. I think the German curve has been like this for a long time, and I think it still is, as I just eyeballed the rolldown yields on the German curve. One possibility would be the U.S. curve would look similar soon, except shifted up on the Y axis to reflect the higher inflation expectations. http://www.worldgovernmentbonds.com/country/germany/
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: Thu Oct 21, 2021 4:05 pm
skierincolorado wrote: Thu Oct 21, 2021 9:13 am
comeinvest wrote: Thu Oct 21, 2021 1:36 am
skierincolorado wrote: Wed Oct 20, 2021 11:34 pm
comeinvest wrote: Wed Oct 20, 2021 11:23 pm I know skier is fond of the ZF; but if I wanted to hedge my bets by diversifying across ZF/ZN/TN/ZB (omitting UB for the moment as it seems like it's considered a sin in this thread) - I cannot decide should I use roughly equal dollar amounts (like NTSX), or roughly equal duration risk in each bucket for diversification. Thoughts?
You know what my answer will be haha. Would be a very solid plan IMO
You would favor equal duration risk per bucket. But then total returns would be dominated by the returns of the short end. Trying to look at it purely from a diversification point of view, considering the future might be different from the past.
Isn't that just an indirect way of saying the short end returns are better? :D From a diversification perspective too I think diversifying the risk is more important than diversifying the return.

Also, if the future is going to be different, you'd make a bunch of money in the interim. For the future to be different, the term premium on shorter durations would have to decrease relative to the term premium on longer durations. In other words, the spread between the 30y and 5y rate would grow. We'd make more money on the 5y / lose less money than we would on the 30y in the interim. So we'd be glad to own the 5y, then we'd end up in this weird unprecedented situation where the yield on the 5y was like .25% while the yield on the 30y was 2%+ (with 0% short term rates - not negative). At that point I think it would be pretty obvious to abandon the 5y - after we had reaped all the benefit.

We can already *guarantee* that over some reasonable time horizon ZF will have better returns than UB (on a per duration basis). This is because the current interest rate per duration is already much much higher. For ZF to do worse, the rate on ZF would have to keep rising faster than the rate on UB. Which just makes it a better and better buy and there is a limit to how much faster it can rise than UB because eventually the curve would go inverted. Like do we really see a situaiton 5 years from now where ZF has 10% interest rate and UB is 5%? Even if that happened, I'd be fine with it because I'd still just hold knowing it can't get more inverted forever.
To be clear, I don't disagree with you. I'm just playing devil's advocate, and examining the strategy from every angle. While I don't disagree with your findings and your strategy, I think it's not a guarantee. If the curve become quite shallow at the short end, then steepens a bit on the long end, and stays that way for a long time, then I think LTT would be better. I think the German curve has been like this for a long time, and I think it still is, as I just eyeballed the rolldown yields on the German curve. One possibility would be the U.S. curve would look similar soon, except shifted up on the Y axis to reflect the higher inflation expectations. http://www.worldgovernmentbonds.com/country/germany/
Oh absolutely if the curve got like that it would make me question the strategy. But to even get there we’d make a ton more money on ITT first as the spread between ITT widened. Once we got to the German situation and pocketed all those gains, then I might reconsider.

We were also kind of in this situation from mid 2020 to early 2021 where I might have avoided ITT or bonds altogether
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

skierincolorado wrote: Thu Oct 21, 2021 9:32 am
hdas wrote: Thu Oct 21, 2021 8:59 am
Kbg wrote: Thu Oct 21, 2021 8:34 am
comeinvest wrote: Wed Oct 20, 2021 11:13 pm
Kbg wrote: Wed Oct 20, 2021 8:07 am Relevant article to the discussion

https://www.simplify.us/blog/efficient- ... e=hs_email
Every decade in those charts had overall falling interest rates, except the 70ies. The chart for the 70ies is the only one that shows LTT better than STT and part of ITT. Any concern that the results will reverse if the interest rate trend reverts to rising?

Image
That's a given. What these charts tell me is that ITTs are probably the place to be as they appear to be the most stable for someone who doesn't want to get into trying to time the market.
Quite the opposite, the non timing strategy would be to select the duration that matches your investment horizon. H
The risk-adjusted return of the 5y has historically been far superior to the 30y, I'll take my chances with the 5y. That's not market timing. I'm not "timing" anything - there is no "timing" involved. I'm simply picking the better investment. Am I "bond picking" the way some people "stock pick" - well maybe, but I would argue no since the 5y is much more representative of the total bond market return. If somebody wanted to take a leveraged total bond position that would be very similar to what I'm doing and have a very similar effective duration.
I personally think the definition of "market timing" is subjective, but that's a discussion similar to factor tilting or not. My original question was have you ever tried to run simulations on periods of rising vs. falling rates separately, for purposes of determining the optimal strategic static positioning on the yield curve. I would be very interested, given that the only period with rising rates shows the anomaly in the chart. Sorry if you have already done that as I'm trying to catch up reading the entire thread.
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: Thu Oct 21, 2021 4:12 pm
skierincolorado wrote: Thu Oct 21, 2021 9:32 am
hdas wrote: Thu Oct 21, 2021 8:59 am
Kbg wrote: Thu Oct 21, 2021 8:34 am
comeinvest wrote: Wed Oct 20, 2021 11:13 pm

Every decade in those charts had overall falling interest rates, except the 70ies. The chart for the 70ies is the only one that shows LTT better than STT and part of ITT. Any concern that the results will reverse if the interest rate trend reverts to rising?

Image
That's a given. What these charts tell me is that ITTs are probably the place to be as they appear to be the most stable for someone who doesn't want to get into trying to time the market.
Quite the opposite, the non timing strategy would be to select the duration that matches your investment horizon. H
The risk-adjusted return of the 5y has historically been far superior to the 30y, I'll take my chances with the 5y. That's not market timing. I'm not "timing" anything - there is no "timing" involved. I'm simply picking the better investment. Am I "bond picking" the way some people "stock pick" - well maybe, but I would argue no since the 5y is much more representative of the total bond market return. If somebody wanted to take a leveraged total bond position that would be very similar to what I'm doing and have a very similar effective duration.
I personally think the definition of "market timing" is subjective, but that's a discussion similar to factor tilting or not. My original question was have you ever tried to run simulations on periods of rising vs. falling rates separately, for purposes of determining the optimal strategic static positioning on the yield curve. I would be very interested, given that the only period with rising rates shows the anomaly in the chart. Sorry if you have already done that as I'm trying to catch up reading the entire thread.
That’s kind of what I do a few posts up. Even in the rising and flattening of 2011 to 2018 vgit beats edv and tied vustx. That’s a terrible environment for ITT. The
30 minus 5 spread tightened nearly 3 % and it still beat EDV!
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

skierincolorado wrote: Thu Oct 21, 2021 4:16 pm That’s kind of what I do a few posts up. Even in the rising and flattening of 2011 to 2018 vgit beats edv and tied vustx. That’s a terrible environment for ITT. The
30 minus 5 spread tightened nearly 3 % and it still beat EDV!
You are faster writing insightful analysis, than I can catch up reading the thread :) Thanks for starting this thread. It's one of my favorite, and one of the most useful for my asset allocation now.
Last edited by comeinvest on Thu Oct 21, 2021 5:48 pm, edited 1 time in total.
Topic Author
skierincolorado
Posts: 2377
Joined: Sat Mar 21, 2020 10:56 am

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

Post by skierincolorado »

comeinvest wrote: Thu Oct 21, 2021 4:30 pm
skierincolorado wrote: Thu Oct 21, 2021 4:16 pm That’s kind of what I do a few posts up. Even in the rising and flattening of 2011 to 2018 vgit beats edv and tied vustx. That’s a terrible environment for ITT. The
30 minus 5 spread tightened nearly 3 % and it still beat EDV!
You are faster writing insightful analysis, than I can catch up reading the thread :) Thanks for starting this thread. It's one of my favorite, and one of the most useful to my asset allocation now.
No problem! The flattening wasn't quite nearly 3% as I stated. The spread 30y-5y decreased from 2.29% to .41%. So nearly 2% flattening between the 5y and 30y. VFITX beats EDV and ties VUSTX. If we use the lower fee VGIT, VGIT wins both, but I use VFITX higher fee to account for financing premium over T-bill. Here's the post (some more comparisons in next several posts):

viewtopic.php?p=6286539#p6286539
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 »

An article related to possible spot/futures distortions: https://www.cfainstitute.org/en/researc ... -the-1990s . Unfortunately only the summary is free.

I think the duration exposure of each futures contract will "drift" over time, as the scheduled coupon payment yields of the underlyings change, based on the corresponding historical prevailing interest rate environment of the time when the treasuries were auctioned that are now in the deliverable time frame. To achieve a constant duration risk exposure, we would have to monitor that and adjust.
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 »

Recent charts showing increased equities / treasuries correlation:

Image

There seem to be some nice contrarian movements on a daily to weekly scale, but not on a monthly scale.
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 »

hdas wrote: Thu Oct 21, 2021 10:37 am ZN, ZF leveraged using DV01 to match UB, which in turn matches VUSTX almost perfectly. 90% funded with VFISX, which is very generous to the leveraged futures position.
Unless I'm missing something, 90% funded with VFISX is not necessarily generous to the futures positions, but overlays the negated return of 2-year futures. Also I'm not understanding why you do this instead of using money market rates as funding rate. Also, are your charts performance or price data?
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

skierincolorado wrote: Thu Oct 21, 2021 11:04 am Wrong. EDV returned 18.7% while a 5x leveraged position in VFITX (borrowing at the T-Bill rate + .25%) returned 19.3% with a lower max-drawdown. Yes we could tweak the borrowing cost a bit higher, like T-Bill + .2%, and VGIT would do a bit worse. But this is to illustrate a point that even during an extreme flattening event, VGIT can do very similar to EDV.
Is CASHX in PV always T-Bill + 0.25% by definition? I can't find the definition of CASHX. Got it. But VFITX has 0.2% ER now.
Last edited by comeinvest on Thu Oct 21, 2021 8:53 pm, edited 1 time in total.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

comeinvest wrote: Thu Oct 21, 2021 7:10 pm
hdas wrote: Thu Oct 21, 2021 10:37 am ZN, ZF leveraged using DV01 to match UB, which in turn matches VUSTX almost perfectly. 90% funded with VFISX, which is very generous to the leveraged futures position.
Unless I'm missing something, 90% funded with VFISX is not necessarily generous to the futures positions, but overlays the negated return of 2-year futures. Also I'm not understanding why you do this instead of using money market rates as funding rate. Also, are your charts performance or price data?
h
Last edited by hdas on Tue Oct 26, 2021 6:43 pm, edited 1 time in total.
....
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

comeinvest wrote: Thu Oct 21, 2021 7:15 pm Is CASHX in PV always T-Bill + 0.25% by definition? I can't find the definition of CASHX.
h
Last edited by hdas on Tue Oct 26, 2021 6:46 pm, edited 1 time in total.
....
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

hhhhh
Last edited by hdas on Tue Oct 26, 2021 6:44 pm, edited 1 time in total.
....
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

comeinvest wrote: Thu Oct 21, 2021 7:10 pm why you do this instead of using money market rates as funding rate.
skierincolorado wrote: Thu Oct 21, 2021 11:45 am So leveraging 3.75x with leverage costs of T-Bill + .28%, I still find VFITX to perform better than VUSTX
h
Last edited by hdas on Tue Oct 26, 2021 6:44 pm, edited 1 time in total.
....
User avatar
LTCM
Posts: 412
Joined: Wed Sep 09, 2020 3:58 am

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

Post by LTCM »

Its very easy to make CASHX+0.1% or CASHX+0.2% in the simba sheet if that's an issue. I just don't have it nailed down what it is - and why its so much harder to find than for equity futures (CASHX+0.3%).
55% VUG - 20% VEA - 20% EDV - 5% BNDX
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 »

hdas wrote: Thu Oct 21, 2021 1:39 pm
skierincolorado wrote: Thu Oct 21, 2021 12:06 pm I acknowledge 2011-2018 was one of the only (probably *the* only) period where VUSTX performed similarly to leveraged VFITX.
I think we can say that since Nov 2010 the strategy hasn't worked as intended, and 11 years is a long time. I just noticed that the DV01 hedge ratios are actually different at the beginning of the UltraBond existance (see table). What this means is that the performance is vastly overstated with 3.75x. This is another reason, (besides historical financing costs) to really distrust your "backtests" from PV, if one wants to be rigorous about comparing apples to apples, the hedge ratio has to be dynamically adjusted at the moment, you can go as granular as you want, this is just quarterly.

Code: Select all

Mar 2010	4:1
June 2010	4:1
Sept 2010	4:1
Dec 2010	4:1
Mar 2011	4:1
June 2011	4:1
Sept 2011	4:1
Dec 2011	4:1
Mar 2012	4:1
June 2012	5:1
June 2012	5:1
Sept 2012	5:1
Dec 2012	5:1
Mar 2013	5:1
June 2013	5:1
Sept 2013	5:1
Dec 2013	4:1
Mar 2014	4:1
Jun 2014	4:1
Sep 2014	5:1
Dec 2014	5:1
Mar 2015	5:1
Jun 2015	6:1
Sep 2015	5:1
Dec 2015	5:1
Mar 2016	5:1
Jun 2016	6:1
Sep 2016	6:1
Dec 2016	7:1
Mar 2017	6:1
Jun 2017	6:1
Sep 2017	6:1
Dec 2017	6:1
Mar 2018	6:1
Jun 2018	6:1
Sep 2018	6:1
Dec 2018	6:1
Mar 2019	6:1
Jun 2019	6:1
Sep 2019	6:1
Dec 2019	6:1
Mar 2020	6:1
Jun 2020	6:1
Sep 2020	6:1
Dec 2020	6:1
Mar 2021	6:1
Jun 2021	6:1
Sep 2021	6:1
Dec 2021	6:1
Where did you get those numbers from? I think hedge ratio between 1 ZF and 1 UF is currently almost exactly 1:7. But the contracts represent different spot treasury dollar values. The duration exposure ratio of equal dollar amounts is about 4.5:1 currently if I'm not mistaken.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

comeinvest wrote: Thu Oct 21, 2021 8:02 pm Where did you get those numbers from? I think hedge ratio between 1 ZF and 1 UF is currently almost exactly 1:7. But the contracts represent different spot treasury dollar values. The duration exposure ratio of equal dollar amounts is about 4.5:1 currently if I'm not mistaken.
h
Last edited by hdas on Tue Oct 26, 2021 6:45 pm, edited 1 time in total.
....
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

LTCM wrote: Thu Oct 21, 2021 8:02 pm Its very easy to make CASHX+0.1% or CASHX+0.2% in the simba sheet if that's an issue. I just don't have it nailed down what it is - and why its so much harder to find than for equity futures (CASHX+0.3%).
h
Last edited by hdas on Tue Oct 26, 2021 6:45 pm, edited 1 time in total.
....
constructor
Posts: 17
Joined: Fri Sep 03, 2021 6:18 pm

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

Post by constructor »

skierincolorado wrote: Wed Oct 20, 2021 5:45 pm
This explains the statement by CME and others that 10M in bonds is hedged by just 81 contracts when the CF is .81. Or taking the inverse, 81 contracts has equal exposure to 10M in bonds. Not 8.1M.

https://us.etrade.com/knowledge/library ... -to-market
The only puzzle in my whole understanding is: With CF = 0.81, and say I long 1 contract till expiration for actual delivery. This means I need to be delivered a face value of $100,000 / 0.81 = $123,456. But each treasury has face value of $1,000 right? How do I receive 123 and then 456/1000 treasuries? Can one own/deliver a fraction of a $1,000 treasury?
constructor
Posts: 17
Joined: Fri Sep 03, 2021 6:18 pm

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

Post by constructor »

hdas wrote: Thu Oct 21, 2021 7:46 pm
comeinvest wrote: Thu Oct 21, 2021 7:10 pm why you do this instead of using money market rates as funding rate.
skierincolorado wrote: Thu Oct 21, 2021 11:45 am So leveraging 3.75x with leverage costs of T-Bill + .28%, I still find VFITX to perform better than VUSTX
Here's why i don't trust the PV backtest:

Image

In the period of Sep/2019 to now, ZF funded with T-Bills doesn't match or beat VFITX. h
Right now the maturity of ZF is 4.3 years and the "Average effective maturity" of VFITX is 5.3 years. The modified duration would be approximately the same as maturity. This means the duration risk is very different by 19%, i.e. they will look different when put in the same graph.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

hhhhh
Last edited by hdas on Tue Oct 26, 2021 6:45 pm, 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 »

hdas wrote: Thu Oct 21, 2021 8:41 pm While your point it's true, it just doesn't explain away the underperformance of futures. I added ZN funded, which has a duration of 6+ years and that also underperforms VFITX.
Looks similar to the S&P charts that I based my slippage calcs on. skier and zkn thought they explained it with duration and max drawdown discrepancies, and reduced effective estimated slippage to about 0.2% if I'm not mistaken i.e. equal to the expense ratio of VFITX, if you read about 2-3 pages back in this thread. I still have to rationalize all of that and verify the math.
Image

Image
Last edited by comeinvest on Thu Oct 21, 2021 9:26 pm, edited 1 time in total.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

hhhhh
Last edited by hdas on Tue Oct 26, 2021 6:45 pm, 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 »

hdas wrote: Thu Oct 21, 2021 9:26 pm
comeinvest wrote: Thu Oct 21, 2021 9:17 pm
hdas wrote: Thu Oct 21, 2021 8:41 pm While your point it's true, it just doesn't explain away the underperformance of futures. I added ZN funded, which has a duration of 6+ years and that also underperforms VFITX.
skier and zkn thought they explained it with duration discrepancies
Both the higher (ZN) and the lower (ZF) duration futures underperform VFITX, so that is not the factor. It must be the implied financing and the fact that you are not making anything in the funded position 3m T-bills. The difference is very significant, 2-2.5% in 2 years....Imagine the compound error of this on those imaginary 30y backtests of PV, total delusion.
Future / futures 5-y return / bond index 5-y return / CTD MD / S&P bond index MD / return per duration (CTD) / return per duration (bond index)
ZF / 1.63% / 1.87% / 4.29 / 4.81 / 0.38 / 0.39

Seems like the duration adjustment explains the difference in my chart. This is for the 5-year chart. But if you read all the comments of skier and zkn, they demonstrated that standard deviations are more meaningful, as durations might have changed during the time frame.

For your chart ZF vs. VFITX: VFITX currently has average duration of 5.2 years. ZF CTD has 4.29 modified duration. Per your chart, the performance is ca. 3% for ZF vs. 6% for VFITS over about 2 years. Adjusting 3% for the ratio of the different durations gives: 3 / 4.29 * 5.2 = 3.6%. Effective slippage would be 6 - 3.6 = 2.4, annualized ca. 1.2%. That is vs. VFITX that has the expense ratio already subtracted. You are right, something doesn't add up. Not sure if durations of either the future or the Vanguard fund changed dramatically during the last 2 years, I doubt. We may have to wait for skier or zkn.
Last edited by comeinvest on Thu Oct 21, 2021 10:56 pm, edited 10 times in total.
hdas
Posts: 1395
Joined: Thu Jun 11, 2015 8:24 am

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

Post by hdas »

hhhhh
Last edited by hdas on Tue Oct 26, 2021 6:46 pm, 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 »

hdas wrote: Thu Oct 21, 2021 9:52 pm
comeinvest wrote: Thu Oct 21, 2021 9:35 pm
Future / futures 5-y return / bond index 5-y return / CTD MD / S&P bond index MD / return per duration (CTD) / return per duration (bond index)
ZF / 1.63% / 1.87% / 4.29 / 4.81 / 0.38 / 0.39

Seems like the duration adjustment explains the difference in my chart. This for the 5-year chart. But if your read all the comments of skier and zkn, they argued that standard deviations are more meaningful, as durations might have changed during the time frame.
2.5% in 2 years is too much. In any case, the main conclusion is that using VFITX as a backtest tool is very bad. Even worst when you compound with leverage 3x, and then only charge CASHX as the cost of the leverage.
You could also try to chart or calculate the standard deviations or volatilities corresponding to you ZF/VFITX chart in the meantime.
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: Thu Oct 21, 2021 7:15 pm
skierincolorado wrote: Thu Oct 21, 2021 11:04 am Wrong. EDV returned 18.7% while a 5x leveraged position in VFITX (borrowing at the T-Bill rate + .25%) returned 19.3% with a lower max-drawdown. Yes we could tweak the borrowing cost a bit higher, like T-Bill + .2%, and VGIT would do a bit worse. But this is to illustrate a point that even during an extreme flattening event, VGIT can do very similar to EDV.
Is CASHX in PV always T-Bill + 0.25% by definition? I can't find the definition of CASHX. Got it. But VFITX has 0.2% ER now.
The VFITX is 0.2% but it's over the entire amount invested. We want to simulate financing on just the CASHX part. Like if CASHX+.25% existed that would be what we would use. Instead you do 300% VFITX -200% CASHX. The financing is on the -200 CASHX. Subtracting .2% from the 300% VFITX part is like adding .25% to the CASHX part.

Pretty nit picky.
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 »

hdas wrote: Thu Oct 21, 2021 7:46 pm
comeinvest wrote: Thu Oct 21, 2021 7:10 pm why you do this instead of using money market rates as funding rate.
skierincolorado wrote: Thu Oct 21, 2021 11:45 am So leveraging 3.75x with leverage costs of T-Bill + .28%, I still find VFITX to perform better than VUSTX
Here's why i don't trust the PV backtest:

Image

In the period of Sep/2019 to now, ZF funded with T-Bills doesn't match or beat VFITX. h
Data source?
Post Reply