HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Locked
samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sat Mar 02, 2019 8:55 pm

dave_k wrote:
Sat Mar 02, 2019 6:10 pm
I get that real return distributions are not the same as a normal distribution, and that when they looked at INDU over the long period they were considering, they found that the positive effects of compounding outweighed the negative effects of volatility. However, in the simulated data we have for UPRO over the last 30 years (as real as any data they used apparently), relative to the S&P 500 returns, that unfortunately does not hold true.

I wish we could link to PV charts using the simulated data, but it's easy to load for anyone that's interested. Maybe I'll post some screenshots later. To illustrate this point, in every case where the S&P 500 returned to a balance it had at a previous time, whether it was up or down in between, the UPRO simulation was lower. If there was no volatility drag, UPRO would be the same. There were definitely periods of sustained growth where 3x was exceeded, but there were also periods (long ones) where it was not only worse than 3x, but worse than 1x, or even went down while the S&P 500 went up overall. Again, UPRO was actually a bit under 1x the S&P 500 over the whole period. Even if you cherry-pick Dec 87 as a starting point (giving it the best chance by buying at a big drop), it's still under 3x overall, and even though it does over 3x from there to 2001, it's back down to just over 1x in 2002, and down to 1/3x in 2009. I think the very long term INDU (and other) returns the paper considered may paint a rosier picture than we can expect of UPRO in the coming few decades, but maybe I'm wrong.

That said, rebalancing with TMF saves the day in this strategy (over the simulation period anyway), apparently taking advantage of a lot of the big gains in UPRO (and some in TMF) while significantly mitigating the losses either would have alone (especially UPRO). We're banking heavily on the rebalancing continuing to work like this, and not so much on 3x leveraged funds having more positive compounding than volatility drag (which they have not overall in the past few decades).
I don't think anyone disagrees with you. Here's ULPIX (2x S&P 500) vs. VFINX (Vanguard's 500 fund) since 1998:

https://www.portfoliovisualizer.com/bac ... ion2_2=100

I guess I was just trying to point out that abstract math doesn't apparently work here vs. the real historical returns, due to non-normal distributions in the real world, at least according to the authors of this paper. And, no, I wouldn't be surprised to find that in some market conditions, even the 40/60 UPRO/TMF will underperform the 1x S&P 500 for awhile, perhaps decades. That's why I said upthread that this might be my gift to my daughters vs. a bonus for my wife and I in 20 years.

That said, and taking your lead, let's look at the data set we have publicly available for now, the HEDGEFUNDIE revised data set currently in the OP. Let's take the case of Bob, the laughably inept world's worst market timer, who only invested at the market peaks, except this time, he's invested in this project.

Bob lump-summed his money into the market at the then-peak at the end of August, 1987. Little did he or anyone else know that a little over six weeks later, on Monday, October 19th, the Dow Jones Industrial Average would experience the single biggest one-day decline in history, a record that stands to this day (-22.61%). Let's look at the Black Monday period from the beginning of September, 1987 (the peak) to December, 1990 (so we can see the detail of the market movements) comparing the 40/60 3X portfolio vs the Vanguard 500:

Image

A bumpy ride to be sure. Indeed:
Following the stock market crash, a group of 33 eminent economists from various nations met in Washington, D.C. in December 1987, and collectively predicted that "the next few years could be the most troubled since the 1930s".

https://en.wikipedia.org/wiki/Black_Monday_(1987)

Notice that the lowest of the low was hardly lower than the Vanguard 500 investor. Let's say Bob stayed the course, despite the eminent economists' prognostications, until the next market peak before a large draw down (August, 2000). How did Bob fare?

Image

Bob, by now feeling like the smartest guy in the room, invested a new tranche of money into the market at the end of Aug, 2000, the then market peak. He'd be rewarded for his prescience by a market that wouldn't stop basically going straight down for two years:

Image

His 40/60 3X portfolio would return his money sometime early January, 2004. The Vanguard 500 investor wouldn't see their money returned to them before the next downturn. The closest they'd get would be Halloween, 2007, when their $10k would be, inflation adjusted, $9446.

Image

On Halloween 2007, Bob dressed up as a zillionaire. Feeling great--and looking even better in his outfit--he again invested another fistful of dollars the next trading day. For the next 14 months or so, the market tried to shake Bob's confidence as it wouldn't stop dropping. This time period would become known as the Great Recession, of course. How'd Bobby do?

Image

Bob let it ride till the end of last December, when he finally retired. How'd that last tranche do?

Image

Keep in mind of course that this is based on the best simulated data that's been made available to us. It is still being worked on in that other thread. But if the data is even somewhat close, it shows me that Bob made it through the last three biggest downturns in the last 31 years and was no worse for wear than the Vanguard 500 investor. What will the next downturn do to this portfolio vs. the 500? I have no idea, but I am somewhat, maybe a little reassured, that even if I unwittingly bought at the market peak a few weeks ago, I might still have a chance of eventually outperforming the 500 in the next few years.
Last edited by samsdad on Sat Mar 02, 2019 9:12 pm, edited 1 time in total.

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sat Mar 02, 2019 9:07 pm

And the doldrums of the early 1990s.

Image

A disappointment for sure. But no worse for wear than the 500 (save for the heartburn). The 40/60 3X would then take off as shown in the previous post.

Topic Author
HEDGEFUNDIE
Posts: 3649
Joined: Sun Oct 22, 2017 2:06 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by HEDGEFUNDIE » Sat Mar 02, 2019 9:11 pm

Nice work, samsdad. I'll link to this in the OP.

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sat Mar 02, 2019 9:14 pm

HEDGEFUNDIE wrote:
Sat Mar 02, 2019 9:11 pm
Nice work, samsdad. I'll link to this in the OP.
Interesting indeed. I was hoping to know how much Bob retired with though!

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sat Mar 02, 2019 9:17 pm

MotoTrojan wrote:
Sat Mar 02, 2019 9:14 pm
HEDGEFUNDIE wrote:
Sat Mar 02, 2019 9:11 pm
Nice work, samsdad. I'll link to this in the OP.
Interesting indeed. I was hoping to know how much Bob retired with though!
Zillions, of course.

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sat Mar 02, 2019 9:43 pm

HEDGEFUNDIE wrote:
Sat Mar 02, 2019 9:11 pm
Nice work, samsdad. I'll link to this in the OP.
Hey I made the front page! :D Thanks! I’ll update the post if we publicly get what is considered better data by those working on it. Just to reiterate: these figures probably will change, perhaps in either direction depending on the period in question oddly enough.

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sat Mar 02, 2019 9:47 pm

Interesting thought from another thread on leverage; daily rebalancing ETFs works far better for equities than treasuries (and especially commodities) since the higher expected equity returns better combat the volatility drag. Seem to recall someone quoting that UPRO since inception has destroyed it's naive return while TMF is well below it, so that makes sense. Perhaps gives credence to a strategy that combines a 3x equity fund with an unleveraged uncorrelated holding (EDV for example which I learned about in this thread)?

EDIT: While replacing TMF with EDV backtests well (better risk-adjusted return and still crushes S&P500), using a similar exposure via VUSTX and a deeper backtest shows far worse performance than TMFSIM.

finite_difference
Posts: 1473
Joined: Thu Jul 09, 2015 7:00 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by finite_difference » Sat Mar 02, 2019 10:29 pm

HEDGEFUNDIE wrote:
Wed Feb 27, 2019 7:27 pm
sarabayo wrote:
Wed Feb 27, 2019 7:18 pm
My apologies for not reading all 17 pages of the thread, but I have a quick question about the following point from the OP:
HEDGEFUNDIE wrote:
Tue Feb 05, 2019 2:56 pm
What if the funds shut down?
If either long Treasuries or the S&P 500 were to drop -33.4% in a single day, the corresponding 3x fund would be wiped out. But realize that would be the fund working as intended. The strategy I am proposing here rests on the idea that if one asset were to drop so precipitously, the other asset would rise sharply to save the portfolio.
OK, but if this happens, UPRO no longer exists at the end of the trading day. So the next day, your portfolio is now 100% TMF, which is not what you want... right? No matter what you do at this point, since UPRO and other assets equivalent to UPRO have now become worthless, you're forced to change the asset allocation in your portfolio, which, if I'm thinking correctly, is akin to "selling at the bottom", i.e. not a great idea. So how do you recover from such a situation?
It's pretty unlikely that the -33.4% in one day would actually happen. Stock exchange circuit breakers kick in at -20%.

But even if it did happen, and UPRO "no longer exists", you should do what any Boglehead is trained to do: stay the course, and when the next quarterly rebalance comes around, rebalance some of your TMF into the nearest equivalent of UPRO that is still around, or even rebalance into the regular S&P 500 itself.
Yeah I think the circuit breakers do help make this strategy a lot safer. So if the stock market crashes 50% over a week. And your next rebalancing is 3 months away, do you still wait?

What if the quarterly rebalancing “got lucky” in the past for the 4 dates that were used in the backtest? Or was it tested that any 4 equally-spaced arbitrary dates work the same? I am curious about how everyone is implementing this strategy when it seems like there isn’t a very clear idea of how to rebalance it. Because what you probably don’t want to do is start changing things as you go along..
The most precious gift we can offer anyone is our attention. - Thich Nhat Hanh

finite_difference
Posts: 1473
Joined: Thu Jul 09, 2015 7:00 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by finite_difference » Sat Mar 02, 2019 10:35 pm

Also, what would happen if Treasury yields went negative like they did in Europe?

Seems like the Fed is responding to pressure to keep rates low despite the upswing, and then there is less room to work with if the economy becomes overheated.
The most precious gift we can offer anyone is our attention. - Thich Nhat Hanh

PluckyDucky
Posts: 156
Joined: Tue Jan 15, 2019 8:29 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by PluckyDucky » Sat Mar 02, 2019 10:42 pm

Does earnings happening quarterly affect quarterly rebalancing?

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sat Mar 02, 2019 10:43 pm

finite_difference wrote:
Sat Mar 02, 2019 10:29 pm
HEDGEFUNDIE wrote:
Wed Feb 27, 2019 7:27 pm
sarabayo wrote:
Wed Feb 27, 2019 7:18 pm
My apologies for not reading all 17 pages of the thread, but I have a quick question about the following point from the OP:
HEDGEFUNDIE wrote:
Tue Feb 05, 2019 2:56 pm
What if the funds shut down?
If either long Treasuries or the S&P 500 were to drop -33.4% in a single day, the corresponding 3x fund would be wiped out. But realize that would be the fund working as intended. The strategy I am proposing here rests on the idea that if one asset were to drop so precipitously, the other asset would rise sharply to save the portfolio.
OK, but if this happens, UPRO no longer exists at the end of the trading day. So the next day, your portfolio is now 100% TMF, which is not what you want... right? No matter what you do at this point, since UPRO and other assets equivalent to UPRO have now become worthless, you're forced to change the asset allocation in your portfolio, which, if I'm thinking correctly, is akin to "selling at the bottom", i.e. not a great idea. So how do you recover from such a situation?
It's pretty unlikely that the -33.4% in one day would actually happen. Stock exchange circuit breakers kick in at -20%.

But even if it did happen, and UPRO "no longer exists", you should do what any Boglehead is trained to do: stay the course, and when the next quarterly rebalance comes around, rebalance some of your TMF into the nearest equivalent of UPRO that is still around, or even rebalance into the regular S&P 500 itself.
Yeah I think the circuit breakers do help make this strategy a lot safer. So if the stock market crashes 50% over a week. And your next rebalancing is 3 months away, do you still wait?

What if the quarterly rebalancing “got lucky” in the past for the 4 dates that were used in the backtest? Or was it tested that any 4 equally-spaced arbitrary dates work the same? I am curious about how everyone is implementing this strategy when it seems like there isn’t a very clear idea of how to rebalance it. Because what you probably don’t want to do is start changing things as you go along..
Rebalancing is certainly the thing about backtesting that I’m least comfortable with but sticking with quarterly for now. I’ll likely do monthly contributions in the near future which will go to the underweight asset but that’ll eventually stop.

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sat Mar 02, 2019 10:44 pm

PluckyDucky wrote:
Sat Mar 02, 2019 10:42 pm
Does earnings happening quarterly affect quarterly rebalancing?
If it did it would be arbitraged out of the market just as dividends are, no?

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sat Mar 02, 2019 10:54 pm

finite_difference wrote:
Sat Mar 02, 2019 10:29 pm
HEDGEFUNDIE wrote:
Wed Feb 27, 2019 7:27 pm
sarabayo wrote:
Wed Feb 27, 2019 7:18 pm
My apologies for not reading all 17 pages of the thread, but I have a quick question about the following point from the OP:
HEDGEFUNDIE wrote:
Tue Feb 05, 2019 2:56 pm
What if the funds shut down?
If either long Treasuries or the S&P 500 were to drop -33.4% in a single day, the corresponding 3x fund would be wiped out. But realize that would be the fund working as intended. The strategy I am proposing here rests on the idea that if one asset were to drop so precipitously, the other asset would rise sharply to save the portfolio.
OK, but if this happens, UPRO no longer exists at the end of the trading day. So the next day, your portfolio is now 100% TMF, which is not what you want... right? No matter what you do at this point, since UPRO and other assets equivalent to UPRO have now become worthless, you're forced to change the asset allocation in your portfolio, which, if I'm thinking correctly, is akin to "selling at the bottom", i.e. not a great idea. So how do you recover from such a situation?
It's pretty unlikely that the -33.4% in one day would actually happen. Stock exchange circuit breakers kick in at -20%.

But even if it did happen, and UPRO "no longer exists", you should do what any Boglehead is trained to do: stay the course, and when the next quarterly rebalance comes around, rebalance some of your TMF into the nearest equivalent of UPRO that is still around, or even rebalance into the regular S&P 500 itself.
Yeah I think the circuit breakers do help make this strategy a lot safer. So if the stock market crashes 50% over a week. And your next rebalancing is 3 months away, do you still wait?

What if the quarterly rebalancing “got lucky” in the past for the 4 dates that were used in the backtest? Or was it tested that any 4 equally-spaced arbitrary dates work the same? I am curious about how everyone is implementing this strategy when it seems like there isn’t a very clear idea of how to rebalance it. Because what you probably don’t want to do is start changing things as you go along..
I’ll take a look tomorrow if I get the opportunity regarding your question as to whether the quarterly rebalancing was lucky. You of course can download the data set and upload them into PV and try them out for yourself. I’m going to shoot from the hip here and say that that would be fairly “lucky” circumstances indeed for all three downturns plus the doldrums to come out like that based solely on a lucky draw of rebalancing dates. But maybe.

For me, I’m going to try ignoring the market until it’s time to rebalance, unless new data comes out that shows that strategy to be foolish a majority of the time, etc. Though anything else starts to smell of market timing.

Topic Author
HEDGEFUNDIE
Posts: 3649
Joined: Sun Oct 22, 2017 2:06 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by HEDGEFUNDIE » Sat Mar 02, 2019 10:56 pm

finite_difference wrote:
Sat Mar 02, 2019 10:29 pm
HEDGEFUNDIE wrote:
Wed Feb 27, 2019 7:27 pm
sarabayo wrote:
Wed Feb 27, 2019 7:18 pm
My apologies for not reading all 17 pages of the thread, but I have a quick question about the following point from the OP:
HEDGEFUNDIE wrote:
Tue Feb 05, 2019 2:56 pm
What if the funds shut down?
If either long Treasuries or the S&P 500 were to drop -33.4% in a single day, the corresponding 3x fund would be wiped out. But realize that would be the fund working as intended. The strategy I am proposing here rests on the idea that if one asset were to drop so precipitously, the other asset would rise sharply to save the portfolio.
OK, but if this happens, UPRO no longer exists at the end of the trading day. So the next day, your portfolio is now 100% TMF, which is not what you want... right? No matter what you do at this point, since UPRO and other assets equivalent to UPRO have now become worthless, you're forced to change the asset allocation in your portfolio, which, if I'm thinking correctly, is akin to "selling at the bottom", i.e. not a great idea. So how do you recover from such a situation?
It's pretty unlikely that the -33.4% in one day would actually happen. Stock exchange circuit breakers kick in at -20%.

But even if it did happen, and UPRO "no longer exists", you should do what any Boglehead is trained to do: stay the course, and when the next quarterly rebalance comes around, rebalance some of your TMF into the nearest equivalent of UPRO that is still around, or even rebalance into the regular S&P 500 itself.
Yeah I think the circuit breakers do help make this strategy a lot safer. So if the stock market crashes 50% over a week. And your next rebalancing is 3 months away, do you still wait?

What if the quarterly rebalancing “got lucky” in the past for the 4 dates that were used in the backtest? Or was it tested that any 4 equally-spaced arbitrary dates work the same? I am curious about how everyone is implementing this strategy when it seems like there isn’t a very clear idea of how to rebalance it. Because what you probably don’t want to do is start changing things as you go along..
In the 32 year backtest, monthly, quarterly, semi-annual, annual rebalancing all delivered results between 14-16% CAGR. This is also true if you rebalanced using calendar quarters or just every three months.

To me, this shows that the strategy is robust to the rebalancing frequency, and that as long as you rebalance at some interval, it should work fine.

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sat Mar 02, 2019 11:24 pm

finite_difference wrote:
Sat Mar 02, 2019 10:29 pm
HEDGEFUNDIE wrote:
Wed Feb 27, 2019 7:27 pm
sarabayo wrote:
Wed Feb 27, 2019 7:18 pm
My apologies for not reading all 17 pages of the thread, but I have a quick question about the following point from the OP:
HEDGEFUNDIE wrote:
Tue Feb 05, 2019 2:56 pm
What if the funds shut down?
If either long Treasuries or the S&P 500 were to drop -33.4% in a single day, the corresponding 3x fund would be wiped out. But realize that would be the fund working as intended. The strategy I am proposing here rests on the idea that if one asset were to drop so precipitously, the other asset would rise sharply to save the portfolio.
OK, but if this happens, UPRO no longer exists at the end of the trading day. So the next day, your portfolio is now 100% TMF, which is not what you want... right? No matter what you do at this point, since UPRO and other assets equivalent to UPRO have now become worthless, you're forced to change the asset allocation in your portfolio, which, if I'm thinking correctly, is akin to "selling at the bottom", i.e. not a great idea. So how do you recover from such a situation?
It's pretty unlikely that the -33.4% in one day would actually happen. Stock exchange circuit breakers kick in at -20%.

But even if it did happen, and UPRO "no longer exists", you should do what any Boglehead is trained to do: stay the course, and when the next quarterly rebalance comes around, rebalance some of your TMF into the nearest equivalent of UPRO that is still around, or even rebalance into the regular S&P 500 itself.
Yeah I think the circuit breakers do help make this strategy a lot safer. So if the stock market crashes 50% over a week. And your next rebalancing is 3 months away, do you still wait?

What if the quarterly rebalancing “got lucky” in the past for the 4 dates that were used in the backtest? Or was it tested that any 4 equally-spaced arbitrary dates work the same? I am curious about how everyone is implementing this strategy when it seems like there isn’t a very clear idea of how to rebalance it. Because what you probably don’t want to do is start changing things as you go along..
Well you got me curious so I just did a real quick look, and I turned off “calendar aligned” option in PV and the CAGR for the Black Monday period went up to 3.63% from 2.63%;
extended from that date to the August, 2000 peak, the CAGR changed to 21.19% from 21.36%; the dot-com period from the beignning of September 2000 till December 2003 became -1.73% versus 1.33%. Then the Great Recession period Nov. 2007 to Dec. 2010 changed to -1.07% from 3.71%. Zoomed out to December 2018 from Nov. 2007, the CAGR changed to 13.89% from 16.82%.

So yes, there were some differences, but nothing where I would say it was such a big difference that we all need to get our rebalancing dates back-tested for every year. Heck, if I recall correctly, there were some times that yearly rebalancing beat out quarterly. I don’t know if there’s any way to know what it’d be going forward.

I’m rebalancing quarterly shortly after the dividends are paid. At least that’s the current plan.

dave_k
Posts: 342
Joined: Sat Dec 26, 2015 8:25 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by dave_k » Sat Mar 02, 2019 11:27 pm

samsdad wrote:
Sat Mar 02, 2019 8:55 pm
I guess I was just trying to point out that abstract math doesn't apparently work here vs. the real historical returns, due to non-normal distributions in the real world, at least according to the authors of this paper. And, no, I wouldn't be surprised to find that in some market conditions, even the 40/60 UPRO/TMF will underperform the 1x S&P 500 for awhile, perhaps decades. That's why I said upthread that this might be my gift to my daughters vs. a bonus for my wife and I in 20 years.

That said, and taking your lead, let's look at the data set we have publicly available for now, the HEDGEFUNDIE revised data set currently in the OP. Let's take the case of Bob, the laughably inept world's worst market timer, who only invested at the market peaks, except this time, he's invested in this project.

...
We're on the same page. When I was referring to the leveraged fund performance in the paper, I was comparing 100% simulated UPRO (not the balanced portfolio) to the S&P 500, to see how it held up to the paper's indication of daily leveraged fund performance matching or exceeding the naive multiple performance over long timescales. My first take on the paper was that it was reassuring about holding leveraged ETFs for decades, but the posts I responded to got me thinking and taking a closer look at the data, which shows that UPRO did not perform as the paper described over the 32.5 year period of the simulated UPRO data. We can't be too reassured by that paper, because it may not apply to UPRO over the timescales we're working with (at least people like me who hope to get something out of it after just 2-3 decades). Volatility decay is very significant over the simulation time period, and isn't affected by sequences like UDUD vs. UUDD as some had been thinking (including me at first). However, that doesn't take rebalancing with TMF into account, which makes all the difference.

I have been looking at the data HEDGEFUNDIE provided, and the 60/40 TMF/UPRO portfolio, comparing it to the S&P 500, and noted many of the same things you pointed out. In fact, before I saw the post the I responded to this morning and got sidetracked, I was planning to write a post indicating many of the same things you did. Now I don't need to. :)

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sat Mar 02, 2019 11:47 pm

Anybody know why DZK has such irregular dividends? Seems like recently it has only had one Q3 and Q4, so not even semi-annually. First few active years it had a regular quarterly div.

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sat Mar 02, 2019 11:50 pm

samsdad wrote:
Sat Mar 02, 2019 11:24 pm
finite_difference wrote:
Sat Mar 02, 2019 10:29 pm
HEDGEFUNDIE wrote:
Wed Feb 27, 2019 7:27 pm
sarabayo wrote:
Wed Feb 27, 2019 7:18 pm
My apologies for not reading all 17 pages of the thread, but I have a quick question about the following point from the OP:
HEDGEFUNDIE wrote:
Tue Feb 05, 2019 2:56 pm
What if the funds shut down?
If either long Treasuries or the S&P 500 were to drop -33.4% in a single day, the corresponding 3x fund would be wiped out. But realize that would be the fund working as intended. The strategy I am proposing here rests on the idea that if one asset were to drop so precipitously, the other asset would rise sharply to save the portfolio.
OK, but if this happens, UPRO no longer exists at the end of the trading day. So the next day, your portfolio is now 100% TMF, which is not what you want... right? No matter what you do at this point, since UPRO and other assets equivalent to UPRO have now become worthless, you're forced to change the asset allocation in your portfolio, which, if I'm thinking correctly, is akin to "selling at the bottom", i.e. not a great idea. So how do you recover from such a situation?
It's pretty unlikely that the -33.4% in one day would actually happen. Stock exchange circuit breakers kick in at -20%.

But even if it did happen, and UPRO "no longer exists", you should do what any Boglehead is trained to do: stay the course, and when the next quarterly rebalance comes around, rebalance some of your TMF into the nearest equivalent of UPRO that is still around, or even rebalance into the regular S&P 500 itself.
Yeah I think the circuit breakers do help make this strategy a lot safer. So if the stock market crashes 50% over a week. And your next rebalancing is 3 months away, do you still wait?

What if the quarterly rebalancing “got lucky” in the past for the 4 dates that were used in the backtest? Or was it tested that any 4 equally-spaced arbitrary dates work the same? I am curious about how everyone is implementing this strategy when it seems like there isn’t a very clear idea of how to rebalance it. Because what you probably don’t want to do is start changing things as you go along..
Well you got me curious so I just did a real quick look, and I turned off “calendar aligned” option in PV and the CAGR for the Black Monday period went up to 3.63% from 2.63%;
extended from that date to the August, 2000 peak, the CAGR changed to 21.19% from 21.36%; the dot-com period from the beignning of September 2000 till December 2003 became -1.73% versus 1.33%. Then the Great Recession period Nov. 2007 to Dec. 2010 changed to -1.07% from 3.71%. Zoomed out to December 2018 from Nov. 2007, the CAGR changed to 13.89% from 16.82%.

So yes, there were some differences, but nothing where I would say it was such a big difference that we all need to get our rebalancing dates back-tested for every year. Heck, if I recall correctly, there were some times that yearly rebalancing beat out quarterly. I don’t know if there’s any way to know what it’d be going forward.

I’m rebalancing quarterly shortly after the dividends are paid. At least that’s the current plan.
Interesting. Some of those periods are long and different enough (namely Nov 2007 to Dec 2018) to be worth a deeper dive into which event gave a 3% CAGR improvement; could potentially help inform some better rules-based rebalancing.

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sun Mar 03, 2019 12:02 am

Fascinating how sensitive to rebalancing this is; makes me curious to play around with this using a standard 60/40 3-fund to see how sensitive it is.

Full available simulated timeframe:

17.5% absolute: 15.83% CAGR

20.0% absolute: 17.22% (beats quarterly)

22.5% absolute: 15.85%

danw70
Posts: 3
Joined: Sun Apr 27, 2014 9:55 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by danw70 » Sun Mar 03, 2019 2:47 am

In the back-test you show a CAGR of 16.71%, but elsewhere you mentioned you expected a CAGR in the mid-20s. Can you explain the discrepancy? Thanks, Dan

interestediniras
Posts: 82
Joined: Thu Jan 10, 2019 2:06 am

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by interestediniras » Sun Mar 03, 2019 3:12 am

MotoTrojan wrote:
Sat Mar 02, 2019 9:47 pm
Interesting thought from another thread on leverage; daily rebalancing ETFs works far better for equities than treasuries (and especially commodities) since the higher expected equity returns better combat the volatility drag. Seem to recall someone quoting that UPRO since inception has destroyed it's naive return while TMF is well below it, so that makes sense. Perhaps gives credence to a strategy that combines a 3x equity fund with an unleveraged uncorrelated holding (EDV for example which I learned about in this thread)?

EDIT: While replacing TMF with EDV backtests well (better risk-adjusted return and still crushes S&P500), using a similar exposure via VUSTX and a deeper backtest shows far worse performance than TMFSIM.
Yes, I think ultimately we need to figure out how to use futures to leverage the bond part.

Previously, we discussed the paper from Loviscek, Tang, and Xu (2014) which performed simulations suggesting that a 3x leveraged market index can perform well, perhaps even in excess of the nominal leverage multiplier. Well, especially if we accept this as evidence, we should also recognize that the opposite is likely for TMF: see, for example, the following article from Tang and Xu (2014) describing substantial detrimental tracking errors with leveraged ETFs tracking fixed income indices:
https://search.proquest.com/openview/c9 ... r&cbl=3598

Unfortunately, the full text is not readily available to me at the moment, although I should be able to access it in a day or two.

SVT
Posts: 265
Joined: Mon Oct 13, 2008 8:56 am

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by SVT » Sun Mar 03, 2019 3:14 am

danw70 wrote:
Sun Mar 03, 2019 2:47 am
In the back-test you show a CAGR of 16.71%, but elsewhere you mentioned you expected a CAGR in the mid-20s. Can you explain the discrepancy? Thanks, Dan
Dan, the first sentence in the OP mentions the update:
UPDATE: With the progress that has been made by EfficientInvestor, siamond, JackoC and others, we now have decent simulated daily data for 3x leveraged funds going back to the mid 80s. This post has been updated with the latest backtest, which shows a 32 year CAGR of 16.7%.
The OP has been updated dozens of times with additional info/graphs and in some cases, changes, like the CAGR.

The reason for the discrepancy, is taking account expenses and other things that I can't recall at the moment.

typical.investor
Posts: 1258
Joined: Mon Jun 11, 2018 3:17 am

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by typical.investor » Sun Mar 03, 2019 4:00 am

interestediniras wrote:
Sun Mar 03, 2019 3:12 am

Yes, I think ultimately we need to figure out how to use futures to leverage the bond part.

Previously, we discussed the paper from Loviscek, Tang, and Xu (2014) which performed simulations suggesting that a 3x leveraged market index can perform well, perhaps even in excess of the nominal leverage multiplier. Well, especially if we accept this as evidence, we should also recognize that the opposite is likely for TMF: see, for example, the following article from Tang and Xu (2014) describing substantial detrimental tracking errors with leveraged ETFs tracking fixed income indices:
https://search.proquest.com/openview/c9 ... r&cbl=3598

Unfortunately, the full text is not readily available to me at the moment, although I should be able to access it in a day or two.
Abstract:
This paper examines the tracking performance of the Leveraged and Regular Fixed Income Exchange-traded Funds (FIETFs) on four major indices: medium-term Treasury, long-term Treasury, investment grade corporate, and high yield corporate bond indices. All sample FIETFs display significant tracking errors, and these tracking errors are much larger for funds on the longer maturity bond index. In addition, funds tracking corporate bond indices show greater tracking errors than those on Treasury bond indices. Finally, tracking errors are larger for leveraged FIETFs than for regular FIETFs and increase as the magnitude of target leverage increases for bull/bear funds. Over multiple trading days such as a week, the return deviation of a leveraged FIETF can be driven by both the NAV deviation due to fund management tracking error and compounding effect due to the daily rebalancing nature of LETFs. While both return deviation components are significant on a weekly basis, the size of the NAV deviation dominates that of the compounding deviation, reflecting the difficulty for fund managers to track fixed income indices.
It seems the same story as found for equities except larger. NAV deviation due to tracking error is more relevant than compounding error.

In another paper it shows interestingly that if instead of everyone rebalancing quarterly just after dividends are paid, everyone chose a different date, that the need for the fund to rebalance at all could theoretically be eliminated. Practically though friction (such as the constraint of creation unit size) present a limit to the benefit.
Empirically, we find that capital flows substantially reduce the need for ETFs to rebalance when returns are large in magnitude and, therefore, mitigate the potential for these products to amplify volatility. We also show theoretically that flows can completely eliminate ETF rebalancing in the limit.
https://www.federalreserve.gov/econresd ... 106pap.pdf

Topic Author
HEDGEFUNDIE
Posts: 3649
Joined: Sun Oct 22, 2017 2:06 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by HEDGEFUNDIE » Sun Mar 03, 2019 4:19 am

interestediniras wrote:
Sun Mar 03, 2019 3:12 am
MotoTrojan wrote:
Sat Mar 02, 2019 9:47 pm
Interesting thought from another thread on leverage; daily rebalancing ETFs works far better for equities than treasuries (and especially commodities) since the higher expected equity returns better combat the volatility drag. Seem to recall someone quoting that UPRO since inception has destroyed it's naive return while TMF is well below it, so that makes sense. Perhaps gives credence to a strategy that combines a 3x equity fund with an unleveraged uncorrelated holding (EDV for example which I learned about in this thread)?

EDIT: While replacing TMF with EDV backtests well (better risk-adjusted return and still crushes S&P500), using a similar exposure via VUSTX and a deeper backtest shows far worse performance than TMFSIM.
Yes, I think ultimately we need to figure out how to use futures to leverage the bond part.

Previously, we discussed the paper from Loviscek, Tang, and Xu (2014) which performed simulations suggesting that a 3x leveraged market index can perform well, perhaps even in excess of the nominal leverage multiplier. Well, especially if we accept this as evidence, we should also recognize that the opposite is likely for TMF: see, for example, the following article from Tang and Xu (2014) describing substantial detrimental tracking errors with leveraged ETFs tracking fixed income indices:
https://search.proquest.com/openview/c9 ... r&cbl=3598

Unfortunately, the full text is not readily available to me at the moment, although I should be able to access it in a day or two.
Interestediniras, can I give you some friendly advice?

You’re going about this all wrong.

Let’s compare TMF (3x LTT) with TYD (3x ITT) since inception:

https://www.portfoliovisualizer.com/fun ... F02%2F2019

https://www.portfoliovisualizer.com/fun ... F02%2F2019

https://www.portfoliovisualizer.com/fun ... F02%2F2019

It is clear that TMF exhibits more volatility decay and even returns less at the end of the day than TYD (which happens to have delivered almost exactly 3x its index). So if I were to ask you which one you should pair with UPRO, you would say TYD, right?

Wrong:

https://www.portfoliovisualizer.com/bac ... tion3_2=60

It turns out that UPRO/TMF handily beats UPRO/TYD. Why? Because TMF is more uncorrelated with UPRO and is more volatile, and therefore offers a greater rebalancing bonus.

You can’t just look at the funds in isolation and say “that one has worse volatility decay, it’s useless”. You have to consider what it contributes to the overall portfolio.

interestediniras
Posts: 82
Joined: Thu Jan 10, 2019 2:06 am

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by interestediniras » Sun Mar 03, 2019 4:27 am

HEDGEFUNDIE wrote:
Sun Mar 03, 2019 4:19 am
Interestediniras, can I give you some friendly advice?

You’re going about this all wrong.

Let’s compare TMF (3x LTT) with TYD (3x ITT) since inception:

https://www.portfoliovisualizer.com/fun ... F02%2F2019

It is clear that TMF exhibits more volatility decay and even returns less at the end of the day than TYD. So if I were to ask you which one you should pair with UPRO, you would say TYD, right?

Wrong:

https://www.portfoliovisualizer.com/bac ... tion3_2=60

It turns out that UPRO/TMF handily beats UPRO/TYD. Why? Because the bond fund that is more volatile and more uncorrelated with UPRO offers a greater rebalancing bonus.

You can’t just look at the funds in isolation and say “that one has worse volatility decay, it’s useless”. You have to see what it contributes to the overall portfolio.
Well, surely that might just be due to the structural differences between ITTs and LTTs. What I'm wondering is not whether I should use TYD or TMF, but rather if I should include leveraged LTTs through a leveraged ETF or through futures (or some other mechanism).

ideally, I would like to first make observations about portfolio construction based on unleveraged funds, for example noting that LTTs plus market index performs well at 40/60, and then apply leverage equally in a way which changes the relationship between LTTs and the index as little as possible. The problems inherent to leveraged ETFs distort that relationship. They may distort that relationship in a way which is either beneficial or detrimental, but the overall result is that the theoretical justification for the portfolio is less principled, and our incomplete understanding of the functionality of LETFs adds undesirable uncertainty, IMHO.

Anyway, maybe saying that we "have to" figure out how to use futures was too strong of an assertion on my end, but it appeals to my perfectionism. To be sure, I agree that TMF is not "useless." I'm not trying to figure out whether or not I should include leveraged LTTs, but rather what form of leverage is most preferable.

jaj2276
Posts: 472
Joined: Sat Apr 16, 2011 5:13 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by jaj2276 » Sun Mar 03, 2019 8:31 am

MotoTrojan wrote:
Sat Mar 02, 2019 11:40 am
jaj2276 wrote:
Sat Mar 02, 2019 9:47 am
MotoTrojan wrote:
Fri Mar 01, 2019 8:34 pm
...

Understood. I would be uncomfortable if for example we didn't have good data for decades on LTT, but dev-market is similar enough in behavior to US equities (volatility etc...) that I am feeling fine. It also scratches my itch for more moving pieces in a more reasonable manner than TQQQ or something along those lines would've.

Been enjoying watching how it behaves; already have XIRR setup on my spreadsheet but it'll be a few years coming before that means much.
This is exactly why I'm not using it. It's too similar to US equities. I want to add components to the trade that will maintain (or increase even) when US equities goes in the tank. I looked at UGLD but the amount I would need to save my portfolio would nuke the returns if it never needed saving.
To be clear they’re similar enough that I don’t need a backtest to say they won’t be a problem, but they still return differently enough to add value. They’d have killed a non existent UPRO in the lost 2000-2010 decade.
Really? Since DZK only goes back to 2008 I didn't know how to backtest with that data. Using the available trading vehicles for this strategy, PV starts at July 2009. And those numbers look awful for DZK. A 30/10 UPRO/DZK split has a CAGR 2.7% lower and a 20/20 UPRO/DZK split is 5.5% lower than simply going with 40% UPRO.

As to the "2000-2010 lost decade", I take those statements with a grain of salt. Yes, if someone had bought in 2000 and then did nothing else but sell in 2010, they would have had a lost decade. However if they were able to either a) purchase more or b) rebalance during that decade, then the decade likely wouldn't have been lost. And since this strategy has 4 rebalances per year, I'd be shocked if 40 rebalance opportunities wouldn't have provided some positive return.

Maybe I'm missing something but I went to PV to look at 2000-2010 using EZU and SPY (starting in Aug 2000 since that is when EZU started). A 100% SPY portfolio would have returned 1.42% CAGR while a 100% EZU portfolio would have returned 1.94%. A 50/50 split with annual rebalancing would have returned 1.88% so I'm still not seeing how EU saved US during this decade.

But back to the original point, if you could be kind enough to show me how to simulate developed markets for this strategy, I might add this to my mix. I started the strategy in early Feb and will be rebalancing end of this month. I'm early enough in this strategy that adding international won't change the outcome 20 years down the road.

And btw, I have international in my non-RP portfolio (2/3 US, 1/3 Intl) so I'm not opposed to international. It simply goes back to my original statement that DZK is not uncorrelated enough to US equities for it to help (IMO) in this strategy. And by help, I don't mean returns but rather zig and zag.

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sun Mar 03, 2019 8:49 am

Just a quick PV backtest on my phone shows the 500 investor got a 0.32% CAGR for the Jan. 2000 to Dec. 2010 period whereas the 40/60 3x returned a 6.96% CAGR.

Edit: I previously posted a different number due to my error in looking at the wrong window on my phone just now. No more posting from me till coffee sinks in!

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sun Mar 03, 2019 9:18 am

SVT wrote:
Sun Mar 03, 2019 3:14 am
danw70 wrote:
Sun Mar 03, 2019 2:47 am
In the back-test you show a CAGR of 16.71%, but elsewhere you mentioned you expected a CAGR in the mid-20s. Can you explain the discrepancy? Thanks, Dan
Dan, the first sentence in the OP mentions the update:
UPDATE: With the progress that has been made by EfficientInvestor, siamond, JackoC and others, we now have decent simulated daily data for 3x leveraged funds going back to the mid 80s. This post has been updated with the latest backtest, which shows a 32 year CAGR of 16.7%.
The OP has been updated dozens of times with additional info/graphs and in some cases, changes, like the CAGR.

The reason for the discrepancy, is taking account expenses and other things that I can't recall at the moment.
I believe the change in CAGR is due to the change in the simulated data set from the original OP. Note that it's a continuing work in progress (at least in the other thread). Someone will come by and correct me if I'm wrong.

User avatar
hdas
Posts: 1270
Joined: Thu Jun 11, 2015 8:24 am

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by hdas » Sun Mar 03, 2019 9:25 am

HEDGEFUNDIE wrote:
Sun Mar 03, 2019 4:19 am
You’re going about this all wrong.

It turns out that UPRO/TMF handily beats UPRO/TYD. Why? Because TMF is more uncorrelated with UPRO and is more volatile, and therefore offers a greater rebalancing bonus.

You can’t just look at the funds in isolation and say “that one has worse volatility decay, it’s useless”. You have to consider what it contributes to the overall portfolio.
TMF vs TYD is the wrong comparison. What one would do to this right is leverage the shorter duration leg to match the vol of the longer and then you will readily see why it's better ALWAYS to lever a strategy or a product with better sharpe profile.

Portfolio 1 == TMF, Portfolio 2 == IEF *7.5X, CASHX -6.5X (Both portfolios have ~38% STD per PV)

You can achieve the optimal leverage duration with the right mix of treasury futures.

Image
"whenever there is a randomized way of doing something, then there is a nonrandomized way that delivers better performance but requires more thought" ET Jaynes

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Sun Mar 03, 2019 9:28 am

MotoTrojan wrote:
Sat Mar 02, 2019 11:50 pm
Interesting. Some of those periods are long and different enough (namely Nov 2007 to Dec 2018) to be worth a deeper dive into which event gave a 3% CAGR improvement; could potentially help inform some better rules-based rebalancing.
I suspected that it was just a simple matter of timing and I was right. If you go back one month to October 2007, the CAGR returns to 16.99%.

Obviously, if you know how to time this going forward into the future, that'd be great for you to share. I recall messing around with PV a week ago or so and picking random starting and ending dates and trying to figure out the optimal rebalancing period overall and came to the conclusion that quarterly kept coming up most of the time. I think it was followed by annual. I didn't write anything down or conduct a rigorous approach to this.

I figure that rebalancing quarterly, with the help of dividends that I've set to drop into my core account at Fidelity, will be "good enough" going forward.

Topic Author
HEDGEFUNDIE
Posts: 3649
Joined: Sun Oct 22, 2017 2:06 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by HEDGEFUNDIE » Sun Mar 03, 2019 10:22 am

Deleted
Last edited by HEDGEFUNDIE on Sun Mar 03, 2019 1:30 pm, edited 1 time in total.

User avatar
siamond
Posts: 5017
Joined: Mon May 28, 2012 5:50 am

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by siamond » Sun Mar 03, 2019 11:39 am

MotoTrojan wrote:
Sat Mar 02, 2019 10:43 pm
finite_difference wrote:
Sat Mar 02, 2019 10:29 pm
What if the quarterly rebalancing “got lucky” in the past for the 4 dates that were used in the backtest? Or was it tested that any 4 equally-spaced arbitrary dates work the same? I am curious about how everyone is implementing this strategy when it seems like there isn’t a very clear idea of how to rebalance it. Because what you probably don’t want to do is start changing things as you go along..
Rebalancing is certainly the thing about backtesting that I’m least comfortable with but sticking with quarterly for now. I’ll likely do monthly contributions in the near future which will go to the underweight asset but that’ll eventually stop.
I am skeptical too about the quarterly rebalancing thing. A few years ago, I modeled various rebalancing algorithms on a bunch of data series while varying start/end dates, and the conclusion was clear, anything more frequent than one or twice a year wasn't a good idea (not very damaging, mind you, but slightly detrimental nevertheless). Essentially because overly frequent rebalancing runs against the forces of momentum (which are somewhat short-term forces while the opposite return-the-mean forces are more longer-term).

This being said, given the way leveraged funds exacerbates returns, maybe there is indeed a good reason for quarterly rebalancing to truly make sense. Such hypothesis should really be tested based on more extended monthly data than the UPRO/TMF actuals. And again, varying start/stop dates is crucial to properly analyze things. Anyhoo, rebalancing specifics seem like a small detail for now...

perplexed
Posts: 63
Joined: Thu Feb 02, 2012 4:32 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by perplexed » Sun Mar 03, 2019 12:22 pm

HEDGEFUNDIE wrote:
Sun Mar 03, 2019 4:19 am
interestediniras wrote:
Sun Mar 03, 2019 3:12 am
MotoTrojan wrote:
Sat Mar 02, 2019 9:47 pm
Interesting thought from another thread on leverage; daily rebalancing ETFs works far better for equities than treasuries (and especially commodities) since the higher expected equity returns better combat the volatility drag. Seem to recall someone quoting that UPRO since inception has destroyed it's naive return while TMF is well below it, so that makes sense. Perhaps gives credence to a strategy that combines a 3x equity fund with an unleveraged uncorrelated holding (EDV for example which I learned about in this thread)?

EDIT: While replacing TMF with EDV backtests well (better risk-adjusted return and still crushes S&P500), using a similar exposure via VUSTX and a deeper backtest shows far worse performance than TMFSIM.
Yes, I think ultimately we need to figure out how to use futures to leverage the bond part.

Previously, we discussed the paper from Loviscek, Tang, and Xu (2014) which performed simulations suggesting that a 3x leveraged market index can perform well, perhaps even in excess of the nominal leverage multiplier. Well, especially if we accept this as evidence, we should also recognize that the opposite is likely for TMF: see, for example, the following article from Tang and Xu (2014) describing substantial detrimental tracking errors with leveraged ETFs tracking fixed income indices:
https://search.proquest.com/openview/c9 ... r&cbl=3598

Unfortunately, the full text is not readily available to me at the moment, although I should be able to access it in a day or two.
Interestediniras, can I give you some friendly advice?

You’re going about this all wrong.

Let’s compare TMF (3x LTT) with TYD (3x ITT) since inception:

https://www.portfoliovisualizer.com/fun ... F02%2F2019

https://www.portfoliovisualizer.com/fun ... F02%2F2019

https://www.portfoliovisualizer.com/fun ... F02%2F2019

It is clear that TMF exhibits more volatility decay and even returns less at the end of the day than TYD (which happens to have delivered almost exactly 3x its index). So if I were to ask you which one you should pair with UPRO, you would say TYD, right?

Wrong:

https://www.portfoliovisualizer.com/bac ... tion3_2=60

It turns out that UPRO/TMF handily beats UPRO/TYD. Why? Because TMF is more uncorrelated with UPRO and is more volatile, and therefore offers a greater rebalancing bonus.

You can’t just look at the funds in isolation and say “that one has worse volatility decay, it’s useless”. You have to consider what it contributes to the overall portfolio.
Is there long term data pairing 3X NASDAQ instead of S&P500?

Topic Author
HEDGEFUNDIE
Posts: 3649
Joined: Sun Oct 22, 2017 2:06 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by HEDGEFUNDIE » Sun Mar 03, 2019 12:53 pm

perplexed wrote:
Sun Mar 03, 2019 12:22 pm
Is there long term data pairing 3X NASDAQ instead of S&P500?
Search for 'TQQQ' in this thread, plenty has been said about it.

Topic Author
HEDGEFUNDIE
Posts: 3649
Joined: Sun Oct 22, 2017 2:06 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by HEDGEFUNDIE » Sun Mar 03, 2019 1:27 pm

interestediniras wrote:
Sun Mar 03, 2019 4:27 am
ideally, I would like to first make observations about portfolio construction based on unleveraged funds, for example noting that LTTs plus market index performs well at 40/60, and then apply leverage equally in a way which changes the relationship between LTTs and the index as little as possible. The problems inherent to leveraged ETFs distort that relationship. They may distort that relationship in a way which is either beneficial or detrimental, but the overall result is that the theoretical justification for the portfolio is less principled, and our incomplete understanding of the functionality of LETFs adds undesirable uncertainty, IMHO.

Anyway, maybe saying that we "have to" figure out how to use futures was too strong of an assertion on my end, but it appeals to my perfectionism. To be sure, I agree that TMF is not "useless." I'm not trying to figure out whether or not I should include leveraged LTTs, but rather what form of leverage is most preferable.
Leveraged ETFs may introduce "distortions" into the leveraged relationship, but practical implementation of futures have their own distortions:

1. Gains/losses on futures are cash-settled on a daily basis, so you do not benefit from those gains compounding with the underlying, as you would with a LETF.

2. Futures contracts come in standard sizes, so if you don't want exposure in multiples of $100k, you are out of luck.

3. Unless you are adjusting your collateral on a daily basis to match your futures exposure, your actual leverage ratio will fluctuate up and down with the movements of the underlying.

With LETFs, at least I know that I am getting 3x the daily performance of the underlying (minus fees).

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sun Mar 03, 2019 1:37 pm

jaj2276 wrote:
Sun Mar 03, 2019 8:31 am



But back to the original point, if you could be kind enough to show me how to simulate developed markets for this strategy, I might add this to my mix. I started the strategy in early Feb and will be rebalancing end of this month. I'm early enough in this strategy that adding international won't change the outcome 20 years down the road.

And btw, I have international in my non-RP portfolio (2/3 US, 1/3 Intl) so I'm not opposed to international. It simply goes back to my original statement that DZK is not uncorrelated enough to US equities for it to help (IMO) in this strategy. And by help, I don't mean returns but rather zig and zag.
I guess my argument is that even without seeing a simulation of this, if you feel it is prudent to have International in your non-RP portfolio, wouldn't those same reasons apply here? I'd wager backtesting wouldn't share DZK helps the portfolio, but if International beats US equities in the future it sure should help. Curious to hear counter-points to this.

Why are you rebalancing end of this month? Seems early if you started in February?

sarabayo
Posts: 330
Joined: Fri Jun 29, 2018 6:59 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by sarabayo » Sun Mar 03, 2019 2:09 pm

siamond wrote:
Sun Mar 03, 2019 11:39 am
I am skeptical too about the quarterly rebalancing thing. A few years ago, I modeled various rebalancing algorithms on a bunch of data series while varying start/end dates, and the conclusion was clear, anything more frequent than one or twice a year wasn't a good idea (not very damaging, mind you, but slightly detrimental nevertheless). Essentially because overly frequent rebalancing runs against the forces of momentum (which are somewhat short-term forces while the opposite return-the-mean forces are more longer-term).
This is probably a dumb question, but doesn't "momentum" just mean that between rebalancings your asset allocation drifts towards being heavier on whatever has most recently been overperforming and lighter on whatever has most recently been underperforming? And isn't that basically the same thing as the market timer's tendency to increase their positions in overperformers and decrease their positions in underperformers?

If we're rebalancing at longer intervals in the hope of better return, isn't that equivalent to just wanting a higher allocation of the higher-return components of our portfolio? If so, is it worth exploring tweaking the AA rather than tweaking the rebalancing period?

I thought that the only reason people don't rebalance daily is that it's inconvenient. For example, Vanguard's target retirement funds apparently rebalance daily. (I guess they're mostly able to do that just by directing inflows appropriately, while an individual investor typically isn't investing new dollars every day). If I had free ETF trades and was in a tax-advantaged account I'd probably place a rebalancing order every weekend just on principle (daily is too much hassle). Am I misguided?

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sun Mar 03, 2019 2:39 pm

sarabayo wrote:
Sun Mar 03, 2019 2:09 pm


If we're rebalancing at longer intervals in the hope of better return, isn't that equivalent to just wanting a higher allocation of the higher-return components of our portfolio? If so, is it worth exploring tweaking the AA rather than tweaking the rebalancing period?
In a traditional equity/broad-bond portfolio perhaps, but in something like this you expect to have periods of substantial outperformance with both sides of the portfolio so you can't just generalize and change your AA one way or the other.

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sun Mar 03, 2019 3:23 pm

jaj2276 wrote:
Sun Mar 03, 2019 8:31 am
MotoTrojan wrote:
Sat Mar 02, 2019 11:40 am
jaj2276 wrote:
Sat Mar 02, 2019 9:47 am
MotoTrojan wrote:
Fri Mar 01, 2019 8:34 pm
...

Understood. I would be uncomfortable if for example we didn't have good data for decades on LTT, but dev-market is similar enough in behavior to US equities (volatility etc...) that I am feeling fine. It also scratches my itch for more moving pieces in a more reasonable manner than TQQQ or something along those lines would've.

Been enjoying watching how it behaves; already have XIRR setup on my spreadsheet but it'll be a few years coming before that means much.
This is exactly why I'm not using it. It's too similar to US equities. I want to add components to the trade that will maintain (or increase even) when US equities goes in the tank. I looked at UGLD but the amount I would need to save my portfolio would nuke the returns if it never needed saving.
To be clear they’re similar enough that I don’t need a backtest to say they won’t be a problem, but they still return differently enough to add value. They’d have killed a non existent UPRO in the lost 2000-2010 decade.
Really? Since DZK only goes back to 2008 I didn't know how to backtest with that data. Using the available trading vehicles for this strategy, PV starts at July 2009. And those numbers look awful for DZK. A 30/10 UPRO/DZK split has a CAGR 2.7% lower and a 20/20 UPRO/DZK split is 5.5% lower than simply going with 40% UPRO.

As to the "2000-2010 lost decade", I take those statements with a grain of salt. Yes, if someone had bought in 2000 and then did nothing else but sell in 2010, they would have had a lost decade. However if they were able to either a) purchase more or b) rebalance during that decade, then the decade likely wouldn't have been lost. And since this strategy has 4 rebalances per year, I'd be shocked if 40 rebalance opportunities wouldn't have provided some positive return.

Maybe I'm missing something but I went to PV to look at 2000-2010 using EZU and SPY (starting in Aug 2000 since that is when EZU started). A 100% SPY portfolio would have returned 1.42% CAGR while a 100% EZU portfolio would have returned 1.94%. A 50/50 split with annual rebalancing would have returned 1.88% so I'm still not seeing how EU saved US during this decade.

But back to the original point, if you could be kind enough to show me how to simulate developed markets for this strategy, I might add this to my mix. I started the strategy in early Feb and will be rebalancing end of this month. I'm early enough in this strategy that adding international won't change the outcome 20 years down the road.

And btw, I have international in my non-RP portfolio (2/3 US, 1/3 Intl) so I'm not opposed to international. It simply goes back to my original statement that DZK is not uncorrelated enough to US equities for it to help (IMO) in this strategy. And by help, I don't mean returns but rather zig and zag.
Thanks for EZU, was struggling to find a good fund to go back further than the GFC with International (would love to find something to go further).

https://www.portfoliovisualizer.com/bac ... on4_2=-200

Using the original -200 CASHX it appears the 25% equity tilt to International would've still performed quite nicely; this makes me feel A-okay with this small tilt which could be very valuable if there is a US-centric downturn.

jaj2276
Posts: 472
Joined: Sat Apr 16, 2011 5:13 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by jaj2276 » Sun Mar 03, 2019 4:24 pm

MotoTrojan wrote:
Sun Mar 03, 2019 1:37 pm
jaj2276 wrote:
Sun Mar 03, 2019 8:31 am



But back to the original point, if you could be kind enough to show me how to simulate developed markets for this strategy, I might add this to my mix. I started the strategy in early Feb and will be rebalancing end of this month. I'm early enough in this strategy that adding international won't change the outcome 20 years down the road.

And btw, I have international in my non-RP portfolio (2/3 US, 1/3 Intl) so I'm not opposed to international. It simply goes back to my original statement that DZK is not uncorrelated enough to US equities for it to help (IMO) in this strategy. And by help, I don't mean returns but rather zig and zag.
I guess my argument is that even without seeing a simulation of this, if you feel it is prudent to have International in your non-RP portfolio, wouldn't those same reasons apply here? I'd wager backtesting wouldn't share DZK helps the portfolio, but if International beats US equities in the future it sure should help. Curious to hear counter-points to this.

Why are you rebalancing end of this month? Seems early if you started in February?
I'm rebalancing at quarter end which will be end of March. It's highly unlikely that this particular rebalance will matter given the 2 month instead of 3 month window but bonds have been awful recently so if the trend holds, I'll be selling some equity and buying some bonds.

There's really not much sense in continuing to debate adding international. I don't think it adds much, you do. Hopefully we'll both come out ahead.

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sun Mar 03, 2019 4:38 pm

jaj2276 wrote:
Sun Mar 03, 2019 4:24 pm
MotoTrojan wrote:
Sun Mar 03, 2019 1:37 pm
jaj2276 wrote:
Sun Mar 03, 2019 8:31 am



But back to the original point, if you could be kind enough to show me how to simulate developed markets for this strategy, I might add this to my mix. I started the strategy in early Feb and will be rebalancing end of this month. I'm early enough in this strategy that adding international won't change the outcome 20 years down the road.

And btw, I have international in my non-RP portfolio (2/3 US, 1/3 Intl) so I'm not opposed to international. It simply goes back to my original statement that DZK is not uncorrelated enough to US equities for it to help (IMO) in this strategy. And by help, I don't mean returns but rather zig and zag.
I guess my argument is that even without seeing a simulation of this, if you feel it is prudent to have International in your non-RP portfolio, wouldn't those same reasons apply here? I'd wager backtesting wouldn't share DZK helps the portfolio, but if International beats US equities in the future it sure should help. Curious to hear counter-points to this.

Why are you rebalancing end of this month? Seems early if you started in February?
I'm rebalancing at quarter end which will be end of March. It's highly unlikely that this particular rebalance will matter given the 2 month instead of 3 month window but bonds have been awful recently so if the trend holds, I'll be selling some equity and buying some bonds.

There's really not much sense in continuing to debate adding international. I don't think it adds much, you do. Hopefully we'll both come out ahead.
Fair enough. Just find it interesting that people who hold it in their broad portfolio don't want to here and curious the reasons (expect US to outperform, anchoring to backtest, etc...).

User avatar
mrspock
Posts: 528
Joined: Tue Feb 13, 2018 2:49 am
Location: Vulcan

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by mrspock » Sun Mar 03, 2019 4:53 pm

MotoTrojan wrote:
Sun Mar 03, 2019 4:38 pm
Fair enough. Just find it interesting that people who hold it in their broad portfolio don't want to here and curious the reasons (expect US to outperform, anchoring to backtest, etc...).
FWIW I hold international in neither, I think Buffet/Bogle’s views on international are dead on (most US companies worth their salt have international presence anyways), and bonds are a vastly superior instrument to control volatility.

If you run the math on how well international would need to do to make up for the dismal performance in the last 5-6 or so years... good luck with that.

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Sun Mar 03, 2019 5:39 pm

mrspock wrote:
Sun Mar 03, 2019 4:53 pm
MotoTrojan wrote:
Sun Mar 03, 2019 4:38 pm
Fair enough. Just find it interesting that people who hold it in their broad portfolio don't want to here and curious the reasons (expect US to outperform, anchoring to backtest, etc...).
FWIW I hold international in neither, I think Buffet/Bogle’s views on international are dead on (most US companies worth their salt have international presence anyways), and bonds are a vastly superior instrument to control volatility.

If you run the math on how well international would need to do to make up for the dismal performance in the last 5-6 or so years... good luck with that.
I can respect this.

User avatar
siamond
Posts: 5017
Joined: Mon May 28, 2012 5:50 am

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by siamond » Sun Mar 03, 2019 6:11 pm

sarabayo wrote:
Sun Mar 03, 2019 2:09 pm
siamond wrote:
Sun Mar 03, 2019 11:39 am
I am skeptical too about the quarterly rebalancing thing. A few years ago, I modeled various rebalancing algorithms on a bunch of data series while varying start/end dates, and the conclusion was clear, anything more frequent than one or twice a year wasn't a good idea (not very damaging, mind you, but slightly detrimental nevertheless). Essentially because overly frequent rebalancing runs against the forces of momentum (which are somewhat short-term forces while the opposite return-the-mean forces are more longer-term).
This is probably a dumb question, but doesn't "momentum" just mean that between rebalancings your asset allocation drifts towards being heavier on whatever has most recently been overperforming and lighter on whatever has most recently been underperforming? And isn't that basically the same thing as the market timer's tendency to increase their positions in overperformers and decrease their positions in underperformers?
Yes, some people view it that way (rebalancing as a form of market timing). I don't. What is wrong with market timing is either a blind reliance on technical indicators (e.g. valuations, moving averages, etc) which actually have next to zero predictive power in the short-term, or to let yourself be driven by various forms of emotions and biases (that never works!), or think that you're smarter than the market (you're not). There is no such thing in rebalancing, it's just a simple mechanical discipline to be enforced, no BS math, no emotions, just re-align things back to your plan. Now if you don't do it too often, you may happen to benefit from momentum effects, but this is just a tiny cherry on the cake, not something to aim at or rely on. Many people make a big deal about rebalancing 'bonuses', but when you run the numbers, those are -on average- very small (and definitely not guaranteed) quantities...
sarabayo wrote:
Sun Mar 03, 2019 2:09 pm
I thought that the only reason people don't rebalance daily is that it's inconvenient. For example, Vanguard's target retirement funds apparently rebalance daily. (I guess they're mostly able to do that just by directing inflows appropriately, while an individual investor typically isn't investing new dollars every day). If I had free ETF trades and was in a tax-advantaged account I'd probably place a rebalancing order every weekend just on principle (daily is too much hassle). Am I misguided?
I don't think you would be misguided, as your thinking is fully centered on what's truly important, which is to stick to your plan. But yes, that's a tad inconvenient, it would take time, and what's not to like about a tiny rebalancing bonus coming with spending LESS efforts on the rebalancing thing? ;-)

As a side note, speaking of inflows (as an accumulator) or outflows (as a retiree), this is indeed a great opportunity to rebalance, exactly like TD funds do. And then the annual rebalancing logic (or triggers or whatever system you have) is actually rarely necessary, besides big turmoil (bull or bear) periods of time. Still, it is important to verify every now and then that you're still close to your plan... This is really the primary goal, by far.

sparksfly
Posts: 31
Joined: Mon Oct 16, 2017 9:57 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by sparksfly » Sun Mar 03, 2019 10:41 pm

Is there also simulated data for 2X funds available?

finite_difference
Posts: 1473
Joined: Thu Jul 09, 2015 7:00 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by finite_difference » Sun Mar 03, 2019 11:52 pm

HEDGEFUNDIE wrote:
Sat Mar 02, 2019 10:56 pm
finite_difference wrote:
Sat Mar 02, 2019 10:29 pm
HEDGEFUNDIE wrote:
Wed Feb 27, 2019 7:27 pm
sarabayo wrote:
Wed Feb 27, 2019 7:18 pm
My apologies for not reading all 17 pages of the thread, but I have a quick question about the following point from the OP:
HEDGEFUNDIE wrote:
Tue Feb 05, 2019 2:56 pm
What if the funds shut down?
If either long Treasuries or the S&P 500 were to drop -33.4% in a single day, the corresponding 3x fund would be wiped out. But realize that would be the fund working as intended. The strategy I am proposing here rests on the idea that if one asset were to drop so precipitously, the other asset would rise sharply to save the portfolio.
OK, but if this happens, UPRO no longer exists at the end of the trading day. So the next day, your portfolio is now 100% TMF, which is not what you want... right? No matter what you do at this point, since UPRO and other assets equivalent to UPRO have now become worthless, you're forced to change the asset allocation in your portfolio, which, if I'm thinking correctly, is akin to "selling at the bottom", i.e. not a great idea. So how do you recover from such a situation?
It's pretty unlikely that the -33.4% in one day would actually happen. Stock exchange circuit breakers kick in at -20%.

But even if it did happen, and UPRO "no longer exists", you should do what any Boglehead is trained to do: stay the course, and when the next quarterly rebalance comes around, rebalance some of your TMF into the nearest equivalent of UPRO that is still around, or even rebalance into the regular S&P 500 itself.
Yeah I think the circuit breakers do help make this strategy a lot safer. So if the stock market crashes 50% over a week. And your next rebalancing is 3 months away, do you still wait?

What if the quarterly rebalancing “got lucky” in the past for the 4 dates that were used in the backtest? Or was it tested that any 4 equally-spaced arbitrary dates work the same? I am curious about how everyone is implementing this strategy when it seems like there isn’t a very clear idea of how to rebalance it. Because what you probably don’t want to do is start changing things as you go along..
In the 32 year backtest, monthly, quarterly, semi-annual, annual rebalancing all delivered results between 14-16% CAGR. This is also true if you rebalanced using calendar quarters or just every three months.

To me, this shows that the strategy is robust to the rebalancing frequency, and that as long as you rebalance at some interval, it should work fine.
Thanks to you and samsdad and others for the responses. It is impressive that you are willing to consider every angle that comes up. Looks like one can pick quarterly dates at will and even if one forgets to rebalance for a month or year it wouldn’t be a disaster.
The most precious gift we can offer anyone is our attention. - Thich Nhat Hanh

MotoTrojan
Posts: 6845
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by MotoTrojan » Mon Mar 04, 2019 12:15 pm

I am not advocating allocating based on recent past, let alone a portion of a day, but I quite enjoyed the dance TMF and UPRO went on this morning.

samsdad
Posts: 745
Joined: Sat Jan 02, 2016 6:20 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by samsdad » Mon Mar 04, 2019 12:33 pm

MotoTrojan wrote:
Mon Mar 04, 2019 12:15 pm
I am not advocating allocating based on recent past, let alone a portion of a day, but I quite enjoyed the dance TMF and UPRO went on this morning.
You're not supposed to peek until it's time to rebalance. Are you . . . having doubts?

Walkure
Posts: 136
Joined: Tue Apr 11, 2017 9:59 pm

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by Walkure » Mon Mar 04, 2019 12:51 pm

samsdad wrote:
Mon Mar 04, 2019 12:33 pm
MotoTrojan wrote:
Mon Mar 04, 2019 12:15 pm
I am not advocating allocating based on recent past, let alone a portion of a day, but I quite enjoyed the dance TMF and UPRO went on this morning.
You're not supposed to peek until it's time to rebalance. Are you . . . having doubts?
MotoTrojan wrote:
Mon Mar 04, 2019 12:15 pm
I am not advocating allocating based on recent past, let alone a portion of a day, but I quite enjoyed the dance TMF and UPRO went on this morning.
Just checked and I'm now officially in, just in time for the stock market to fall off a cliff. The 40/60 3x AA is still holding slightly above 100/0 unlevered, so I guess the strategy is doing its noncorrelated thing as intended. Time to tune out until next quarter, and my first rebalance. In honor of Mardi Gras tomorrow, Laissez les bon temps rouler! :sharebeer

shawcroft
Posts: 927
Joined: Sun Feb 24, 2008 4:18 pm
Location: Connecticut

Re: HEDGEFUNDIE's excellent adventure [risk parity strategy using 3x leveraged ETFs]

Post by shawcroft » Mon Mar 04, 2019 1:53 pm

finite_difference wrote:
Sun Mar 03, 2019 11:52 pm
Thanks to you and samsdad and others for the responses. It is impressive that you are willing to consider every angle that comes up. Looks like one can pick quarterly dates at will and even if one forgets to rebalance for a month or year it wouldn’t be a disaster.
Fascinating ( to me) observations in this series of messages.
Shawcroft

Locked