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
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

The original portfolio suggested is a 40% UPRO / 60% TMF portfolio with full 3x leverage. So far we know that less leverage, more intermediate-term bonds, and a higher ratio of beta (stock exposure) to term risk (treasury duration) all helped boost performance pre-1980s. We've also found out that using TMF for 3x exposure to long-term bonds is more efficient than using TYD for 3x exposure to intermediate-term bonds. This isn't for any theoretical reason (in fact a good implementation of 3x intermediate-term bonds could be great) but because of practical issues with TYD, especially bid/ask spread and a history of tracking error.

This suggests exploring the tradeoff of reducing 60% TMF and increasing the allocation to unleveraged intermediate-term treasuries, while keeping the 40% UPRO allocation the same. Doing so reduces the problems during the pre-1980s period in all three ways: it shifts the portfolio towards a higher equity risk to term risk ratio (by reducing the term risk in bonds), it shifts the bond allocation to include more intermediate duration, and it reduces the leverage in the portfolio. The basic idea here is that equities have good real returns in almost any 20-30 year period, but the 60% TMF is a liability that can drag the real return of the whole portfolio down a lot for a long time, even on a 20-30 year time scale.

All comparisons are based on the entire time period 1955-2018, including the 27 years 1955-1981 and the 37 years 1982-2018. This is just the entire data set available in Simba's spreadsheet with the data from siamond (any cherry picking of start/end years is unintentional). Everything is based on year-to-year calculations. Since the funds didn't exist back then, these results are simulated from a model that fits the ETFs in the recent past.

UPRO is a 3x SP500 fund, TMF is a 3x 20+ year treasury fund, and IEF is a 7-10 year treasury fund.

40% UPRO, 60% IEF -> 11.49% CAGR, 20.89% std. dev., 0.42 Sharpe, 2.18 Sortino, 39.09% max drawdown.
40% UPRO, 10% TMF, 50% IEF -> 11.75% CAGR, 21.65% std. dev., 0.42 Sharpe, 2.16 Sortino, 41.78% max drawdown.
40% UPRO, 20% TMF, 40% IEF -> 11.91% CAGR, 22.9% std. dev., 0.41 Sharpe, 2.15 Sortino, 44.41% max drawdown.
40% UPRO, 30% TMF, 30% IEF -> 11.98% CAGR, 24.55% std. dev., 0.40 Sharpe, 2.03 Sortino, 46.98% max drawdown.
40% UPRO, 40% TMF, 20% IEF -> 11.95% CAGR, 26.54% std. dev., 0.38 Sharpe, 1.89 Sortino, 49.49% max drawdown.
40% UPRO, 50% TMF, 10% IEF -> 11.84% CAGR, 28.79% std. dev., 0.37 Sharpe, 1.77 Sortino, 54.13% max drawdown.
40% UPRO, 60% TMF -> 11.65% CAGR, 31.24% std. dev., 0.35 Sharpe, 1.67 Sortino, 63.05% max drawdown.

Of these options, anything with at least 30% IEF (and, at most, 30% TMF) looks reasonable to me.

The last three options seem to have lower expected return along with higher, uncompensated risk.
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

MoneyMarathon wrote: Sat Jul 13, 2019 6:00 pm The original portfolio suggested is a 40% UPRO / 60% TMF portfolio with full 3x leverage. So far we know that less leverage, more intermediate-term bonds, and a higher ratio of beta (stock exposure) to term risk (treasury duration) all helped boost performance pre-1980s. We've also found out that using TMF for 3x exposure to long-term bonds is more efficient than using TYD for 3x exposure to intermediate-term bonds. This isn't for any theoretical reason (in fact a good implementation of 3x intermediate-term bonds could be great) but because of practical issues with TYD, especially bid/ask spread and a history of tracking error.

This suggests exploring the tradeoff of reducing 60% TMF and increasing the allocation to unleveraged intermediate-term treasuries, while keeping the 40% UPRO allocation the same. Doing so reduces the problems during the pre-1980s period in all three ways: it shifts the portfolio towards a higher equity risk to term risk ratio (by reducing the term risk in bonds), it shifts the bond allocation to include more intermediate duration, and it reduces the leverage in the portfolio. The basic idea here is that equities have good real returns in almost any 20-30 year period, but the 60% TMF is a liability that can drag the real return of the whole portfolio down a lot for a long time, even on a 20-30 year time scale.

All comparisons are based on the entire time period 1955-2018, including the 27 years 1955-1981 and the 37 years 1982-2018. This is just the entire data set available in Simba's spreadsheet with the data from siamond (any cherry picking of start/end years is unintentional). Everything is based on year-to-year calculations. Since the funds didn't exist back then, these results are simulated from a model that fits the ETFs in the recent past.

UPRO is a 3x SP500 fund, TMF is a 3x 20+ year treasury fund, and IEF is a 7-10 year treasury fund.

40% UPRO, 60% IEF -> 11.49% CAGR, 20.89% std. dev., 0.42 Sharpe, 2.18 Sortino, 39.09% max drawdown.
40% UPRO, 10% TMF, 50% IEF -> 11.75% CAGR, 21.65% std. dev., 0.42 Sharpe, 2.16 Sortino, 41.78% max drawdown.
40% UPRO, 20% TMF, 40% IEF -> 11.91% CAGR, 22.9% std. dev., 0.41 Sharpe, 2.15 Sortino, 44.41% max drawdown.
40% UPRO, 30% TMF, 30% IEF -> 11.98% CAGR, 24.55% std. dev., 0.40 Sharpe, 2.03 Sortino, 46.98% max drawdown.
40% UPRO, 40% TMF, 20% IEF -> 11.95% CAGR, 26.54% std. dev., 0.38 Sharpe, 1.89 Sortino, 49.49% max drawdown.
40% UPRO, 50% TMF, 10% IEF -> 11.84% CAGR, 28.79% std. dev., 0.37 Sharpe, 1.77 Sortino, 54.13% max drawdown.
40% UPRO, 60% TMF -> 11.65% CAGR, 31.24% std. dev., 0.35 Sharpe, 1.67 Sortino, 63.05% max drawdown.

Of these options, anything with at least 30% IEF (and, at most, 30% TMF) looks reasonable to me.

The last three options seem to have lower expected return along with higher, uncompensated risk.
I dabbled with the idea of using a combo of EDV and TMF to reduce the volatility decay on the bond side. One implementation could be maintaining a fixed UPRO exposure (maybe 30%) and blending TMF and EDV to achieve risk parity every month. Not sure how I feel about constant equity absolute exposure but varying allocation. Maybe more efficient but obviously less leverage so less return.

Curious how your options did 1982 to present.
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

Noting again: UPRO is a 3x SP500 fund, TMF is a 3x 20+ year treasury fund, and IEF is a 7-10 year treasury fund. All numbers (including, particularly, max drawdown, standard deviation, & risk metrics) are only from year to year calculations. Someone's impression of risk might be different if they check the news more than once a year. :)
MotoTrojan wrote: Sat Jul 13, 2019 6:22 pm Curious how your options did 1982 to present.
Jan 1982 - Dec 2018

40% UPRO, 00% TMF, 60% IEF -> 13.29% CAGR, 19.75% std. dev., 0.56 Sharpe, 2.60 Sortino, 35.07% max drawdown.
40% UPRO, 10% TMF, 50% IEF -> 14.68% CAGR, 20.69% std. dev., 0.61 Sharpe, 3.91 Sortino, 30.39% max drawdown.
40% UPRO, 20% TMF, 40% IEF -> 15.94% CAGR, 22.26% std. dev., 0.64 Sharpe, 4.18 Sortino, 25.6% max drawdown.
40% UPRO, 30% TMF, 30% IEF -> 17.09% CAGR, 24.34% std. dev., 0.64 Sharpe, 4.05 Sortino, 22.76% max drawdown.
40% UPRO, 40% TMF, 20% IEF -> 18.13% CAGR, 26.81% std. dev., 0.64 Sharpe, 3.94 Sortino, 20.88% max drawdown.
40% UPRO, 50% TMF, 10% IEF -> 19.06% CAGR, 29.57% std. dev., 0.63 Sharpe, 3.85 Sortino, 23.39% max drawdown.
40% UPRO, 60% TMF, 00% IEF -> 19.9% CAGR, 32.55% std. dev., 0.62 Sharpe, 3.78 Sortino, 26.34% max drawdown.

100% SPY (S&P 500 index) -> 11.19% CAGR, 16.31% std. dev., 0.53 Sharpe, 1.33 Sortino, 37.80% max drawdown.

Jan 1955 - Dec 1981

40% UPRO, 00% TMF, 60% IEF -> 9.08% CAGR, 23.72% std. dev., 0.25 Sharpe, 0.95 Sortino, 39.09% max drawdown.
40% UPRO, 10% TMF, 50% IEF -> 7.86% CAGR, 22.77% std. dev., 0.20 Sharpe, 0.73 Sortino, 41.78% max drawdown.
40% UPRO, 20% TMF, 40% IEF -> 6.61% CAGR, 23.14% std. dev., 0.15 Sharpe, 0.54 Sortino, 44.41% max drawdown.
40% UPRO, 30% TMF, 30% IEF -> 5.33% CAGR, 23.65% std. dev., 0.10 Sharpe, 0.34 Sortino, 46.98% max drawdown.
40% UPRO, 40% TMF, 20% IEF -> 4.01% CAGR, 24.28% std. dev., 0.06 Sharpe, 0.17 Sortino, 49.49% max drawdown.
40% UPRO, 50% TMF, 10% IEF -> 2.66% CAGR, 25.03% std. dev., 0.01 Sharpe, 0.03 Sortino, 54.13% max drawdown.
40% UPRO, 60% TMF, 00% IEF -> 1.26% CAGR, 25.89% std. dev., -0.03 Sharpe, -0.08 Sortino, 63.05% max drawdown.

100% SPY (S&P 500 index) -> 8.57% CAGR, 17.51% std. dev., 0.26 Sharpe, 0.71 Sortino, 37.39% max drawdown.

Jan 1955 - Dec 2018

40% UPRO, 00% TMF, 60% IEF -> 11.49% CAGR, 20.89% std. dev., 0.42 Sharpe, 2.18 Sortino, 39.09% max drawdown.
40% UPRO, 10% TMF, 50% IEF -> 11.75% CAGR, 21.65% std. dev., 0.42 Sharpe, 2.16 Sortino, 41.78% max drawdown.
40% UPRO, 20% TMF, 40% IEF -> 11.91% CAGR, 22.9% std. dev., 0.41 Sharpe, 2.15 Sortino, 44.41% max drawdown.
40% UPRO, 30% TMF, 30% IEF -> 11.98% CAGR, 24.55% std. dev., 0.40 Sharpe, 2.03 Sortino, 46.98% max drawdown.
40% UPRO, 40% TMF, 20% IEF -> 11.95% CAGR, 26.54% std. dev., 0.38 Sharpe, 1.89 Sortino, 49.49% max drawdown.
40% UPRO, 50% TMF, 10% IEF -> 11.84% CAGR, 28.79% std. dev., 0.37 Sharpe, 1.77 Sortino, 54.13% max drawdown.
40% UPRO, 60% TMF, 00% IEF -> 11.65% CAGR, 31.24% std. dev., 0.35 Sharpe, 1.67 Sortino, 63.05% max drawdown.

100% SPY (S&P 500 index) -> 10.08% CAGR, 17.05% std. dev., 0.41 Sharpe, 1.42 Sortino, 37.80% max drawdown.
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

MotoTrojan wrote: Sat Jul 13, 2019 6:22 pmMaybe more efficient but obviously less leverage so less return.
More leverage doesn't always give more return. I'm not an expert on bonds, so I can only give a simple take on this, but I think it should be pretty obvious that you can get negative returns with leverage... just by not making enough on something that would otherwise have positive yield (unleveraged).

Right now there are relatively-low rates for the 20-to-30-year yield and a compressed yield curve.

Image

The difference between the yield on the 1-month T-bill and average duration 25-years (around 2.48%) is only 0.32%.

The fund itself has a 1.09% expense ratio and 1.16% yield, mostly cancelling out in a tax-advantaged account.

I think the basic concept of TMF is that you want the volatility of getting duration risk but you don't really care about inherent expected return in an environment where interest rates are flat, because you've given up that in order to be in an expensive leveraged fund.

You only really profit from TMF in two scenarios: (1) rates go down, or, even better, (2) rates go down when stocks go down.

Those scenarios apply more often than not post-1980s, so you're sitting pretty if rates keep going down or if, at the very least, rates go down when stocks go down. And, since this is a big bet on sideways or declining rates, you're also hosed if rates go up and up.

But... all the unleveraged ETFs like IEF, TLT, and EDV also benefit from (1) and (2). And they pay out significant net interest in a sideways rates environment (or, any environment) even if you don't get any particular great advantage from the volatility in rates.

You only sometimes get more return with more leverage. You often get less, especially with TMF, which is designed as a play on the change in the long-term treasury yield and mostly lives up to performing that function even in a long-term buy-and-hold portfolio. I'm not sure why there's such strong recency bias in favor of TMF when it's very clear that these super-strong-performing periods were just going from higher rates to lower rates.

Play 60% TMF only if you believe you know the direction of interest rates over the time period in which you're invested, IMO.
Topic Author
HEDGEFUNDIE
Posts: 4801
Joined: Sun Oct 22, 2017 2:06 pm

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

Post by HEDGEFUNDIE »

TMF also benefits from the actual yield on long treasuries, not just the change in yield...
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

HEDGEFUNDIE wrote: Sat Jul 13, 2019 8:04 pm TMF also benefits from the actual yield on long treasuries, not just the change in yield...
Right. I was saying that a lot of that benefit is eaten up by other costs in TMF.
NMBob
Posts: 382
Joined: Thu Apr 23, 2015 8:13 pm

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

Post by NMBob »

"This correlation matrix makes NUGT in particular look troubled, with extreme volatility and movement less correlated with the rest."

NUGT is gold mining industry, not gold spot price. Might explain some differences.
Kbg
Posts: 85
Joined: Thu Mar 23, 2017 11:33 am

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

Post by Kbg »

Good to see the thread evolving and learning some important execution things. I’ve been doing something similar to this going on 4 years now.

A couple of points/food for thought.

Leaps: if you are not figuring the cost of leaps into your performance expectations I suggest you do so. The options market is highly efficient...and if you believe your long run returns will be the same, have I got a bridge for you.

Futures: To the person who posted earlier...please do the math. Futures (I’ve used them for years) are great but they are unforgiving, very unforgiving. A quick eyeball suggested to me you are quite undercapitalized. Read your futures agreement closely, depending on the firm you could lose your entire margin balance and get a bill for more depending on how and when you were liquidated. Do not forget futures are marked to market continuously the entire time they are active - Sunday evening until Friday afternoon and there can be huge overnight moves due to lack of liquidity combined with major news. Without sufficient cash you could easily wake up in the morning with zero in your account. However, you will be able to pull up an overnight chart and see the precise moment you were liquidated. It will be doubly painful as by the morning you probably would have been fine.

A dab of gold, a worthy pursuit if one would like to have a reasonable chance of ending this adventure successfully. If you go here stay away from gold mining stocks and use actual gold/gold derivatives.

Good discussion on using derivatives vs. actual cash/bonds...a critical and overlooked nuance you guys are dialing into. It’s complex but on one level very simple (financing a position vs. getting the interest from holding it).
no simpler
Posts: 109
Joined: Sat Jul 13, 2019 4:54 pm

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

Post by no simpler »

Just got through this thread.

The fundamental premise is pretty sound. It's not dumb, and it doesn't even depend on alpha. You're simply levering betas. It doesn't really deviate from even a pretty strict interpretation of efficiency of market hypothesis. One sanity check is that while the return is high, so is the volatility; the sharpe ratio is nothing particularly out of this world. It's when you have a high return and low backtested vol that there's usually some hidden risk (think writing naked options, etc). I also like that its quite simple.

However, I would heed some advice from Aaron Brown, who I trust on this subject: https://www.quora.com/What-are-the-pitf ... ealize-the He suggests keeping vol at half the S&P.

I also agree with Kbg that it makes sense to add a gold position. If both treasuries and stocks dump, not many other places for money to go.The original paper on this type of strategy from Dalio also mentions EM debt and TIPS for a high inflation scenario. Not doing so because it doesn't do as well in a backtest is silly, as Dalio articulates: https://www.bridgewater.com/resources/a ... -story.pdf It's so easy to overfit.

Lastly, maybe I missed it, but has anybody played with adjusting leverage based on Shiller PE (or interest rates)? That is, increasing leverage when Shiller PE is low, and decreasing when high? Just a thought.
DriftWood
Posts: 31
Joined: Sun Mar 16, 2014 12:38 am

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

Post by DriftWood »

MoneyMarathon, thanks for the study of adding IEF to the portfolio. Excellent job! Improved standard deviation and max drawdown at the expense of CAGR is interesting. One question, is quarterly rebalance used in your study?
User avatar
privatefarmer
Posts: 725
Joined: Mon Sep 08, 2014 2:45 pm

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

Post by privatefarmer »

MoneyMarathon wrote: Sat Jul 13, 2019 7:56 pm
MotoTrojan wrote: Sat Jul 13, 2019 6:22 pmMaybe more efficient but obviously less leverage so less return.
More leverage doesn't always give more return. I'm not an expert on bonds, so I can only give a simple take on this, but I think it should be pretty obvious that you can get negative returns with leverage... just by not making enough on something that would otherwise have positive yield (unleveraged).

Right now there are relatively-low rates for the 20-to-30-year yield and a compressed yield curve.

Image

The difference between the yield on the 1-month T-bill and average duration 25-years (around 2.48%) is only 0.32%.

The fund itself has a 1.09% expense ratio and 1.16% yield, mostly cancelling out in a tax-advantaged account.

I think the basic concept of TMF is that you want the volatility of getting duration risk but you don't really care about inherent expected return in an environment where interest rates are flat, because you've given up that in order to be in an expensive leveraged fund.

You only really profit from TMF in two scenarios: (1) rates go down, or, even better, (2) rates go down when stocks go down.

Those scenarios apply more often than not post-1980s, so you're sitting pretty if rates keep going down or if, at the very least, rates go down when stocks go down. And, since this is a big bet on sideways or declining rates, you're also hosed if rates go up and up.

But... all the unleveraged ETFs like IEF, TLT, and EDV also benefit from (1) and (2). And they pay out significant net interest in a sideways rates environment (or, any environment) even if you don't get any particular great advantage from the volatility in rates.

You only sometimes get more return with more leverage. You often get less, especially with TMF, which is designed as a play on the change in the long-term treasury yield and mostly lives up to performing that function even in a long-term buy-and-hold portfolio. I'm not sure why there's such strong recency bias in favor of TMF when it's very clear that these super-strong-performing periods were just going from higher rates to lower rates.

Play 60% TMF only if you believe you know the direction of interest rates over the time period in which you're invested, IMO.
a narrow yield curve is of concern however I am not one to try and "time the market" so I would look at this strategy for the long-term as a buy-and-hold strategy. This strategy has worked very well since 1983 even though borrowing costs were significantly higher back in the 80s. So although you were benefiting from higher yields on LTTs you were paying more to borrow at the risk-free rate. Rates have been dropping since then but you have to remember that, due to bond convexity, durations are now much higher than they were then and so even a minor drop in interest rates today will have a pronounced effect on price return.

I personally am implementing a standard leveraged risk-parity strategy with my entire portfolio, not just a slice, and am trying to use a margin loan to obtain most of my leverage. I use LETFs to make up the difference, hoping in time to "pay down" those/convert them to non-leveraged ETFs and only have my margin loan remaining. But I think that we have to remember that the strategy is not relying on TMF or LTTs to generate returns. The strategy rather relies on LTTs/TMF to provide negative correlation to stocks when the market tanks, thus increasing our sharpe ratio. You have to remember that we are leveraging the heck out of stocks as well so even though stocks are "only" 40% of our portfolio because of the leverage we effectively have 120% exposure to stocks. So the LTTs are more to soften the downside risk vs provide actual return.
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

DriftWood wrote: Sat Jul 13, 2019 11:36 pm MoneyMarathon, thanks for the study of adding IEF to the portfolio. Excellent job! Improved standard deviation and max drawdown at the expense of CAGR is interesting.
Thanks! I think I might suggest that it's mostly at the expense of "Portfolio Visualizer CAGR." The suggestions to diversify the bond allocation (beyond just UPRO / TMF) all have higher CAGR for the time period 1955-2018. Someone can take a bet on 60% TMF and they might end up being right, but I personally don't think it actually provides higher expected CAGR.

I agree with the main pillars of Hedgefundie's OP, though. Long-term treasuries do provide good diversification to equities & deserve an allocation. Leverage can be used to improve returns and even to improve risk-adjusted returns (Sharpe) for a given level of expected return. And leveraged ETFs need a second look as a way for the individual investor to access cost-effective leverage.
DriftWood wrote: Sat Jul 13, 2019 11:36 pmOne question, is quarterly rebalance used in your study?
I've been using Simba's spreadsheet with siamond's data, which has annual numbers and uses annual rebalancing. You can download the spreadsheet and try out your own ideas too:

viewtopic.php?p=4425650#p4426315

https://drive.google.com/open?id=16ORud ... WfTP-0FggA
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

privatefarmer wrote: Sun Jul 14, 2019 12:08 ama narrow yield curve is of concern however I am not one to try and "time the market" so I would look at this strategy for the long-term as a buy-and-hold strategy. This strategy has worked very well since 1983 even though borrowing costs were significantly higher back in the 80s.
privatefarmer wrote: Sun Jul 14, 2019 12:08 amBut I think that we have to remember that the strategy is not relying on TMF or LTTs to generate returns. The strategy rather relies on LTTs/TMF to provide negative correlation to stocks when the market tanks, thus increasing our sharpe ratio. You have to remember that we are leveraging the heck out of stocks as well so even though stocks are "only" 40% of our portfolio because of the leverage we effectively have 120% exposure to stocks. So the LTTs are more to soften the downside risk vs provide actual return.
I would like to tease out an important point here. Consider these different goals and observations:
  • not "timing the market," long-term buy-and-hold strategy
  • this strategy has worked very well since 1983
  • the strategy is not relying on TMF or LTTs to generate returns
  • rather relies on LTTs/TMF to provide negative correlation to stocks when the market tanks
All of these points individually seem pretty intuitive:
  • it's a fixed allocation
  • since 1983 is a long time... and maybe a more relevant time period since it's more recent?
  • bonds are for safety
  • bonds have gone up when stocks go down
The apparently counter-intuitive thing here is that TMF has generated great returns... during 1987-2018. That's the most important part of what makes the backtest with a full TMF allocation so amazing. The "stock market tanks" anti-correlation-based rebalancing effects are also a factor, but a secondary one: if TMF was a drag or did little for the portfolio otherwise, but just had great anti-correlation, things would have been very different for a backtest on this range of dates.

Using Simba's spreadsheet:

40% UPRO rebalanced annually with plain cash (zero correlation, no interest) provided 7.77% CAGR and 10.97x return.
60% TMF rebalanced annually with plain cash (zero correlation, no interest) provided 8.07% CAGR and 11.96x return.
40% UPRO / 60% TMF rebalanced annually with each other provided 16.89% CAGR and 147.45x return.

Backing out the different factors, if there were no anti-correlation effect during annual rebalancing (just zero correlation), we might have expected to get a return of 10.97 x 11.96 = 131.2x. A return of 131.2x over 32 years is a CAGR of 16.46%. We actually got a CAGR of 16.89% which allowed us to get an extra 12.4% accumulated over a 32 year time period.

The added value of 60% TMF over the 32 years here was divided between 11.96x return, by itself with zero correlation to no-yield cash, and an extra 1.12x return thanks to the incremental benefit of being anti-correlated with UPRO. Understandably, then, I believe it should be plain that if TMF doesn't deliver strong positive yield all on its own, then you're not going to get it anything close to working like it did 1980s to present.

I think there is possibly an argument that TMF rebalancing with plain cash delivers its own premium, maybe by making good trades. But a starting 60% TMF (not even 100%) without rebalancing actually did even better. Without rebalancing into cash, a portfolio that drifted into a larger percentage allocation to TMF over time returned 15.6x or 8.96% CAGR. TMF just plain did well in this time period.

The main reason why TMF did very well (not middling, neutral, or negative) is simply linked to how the price of a bond is calculated.

That is, the yield on 30-year treasuries went down.

If the 30-year treasury yield had gone up during this time period, you'd get some very poor results like the ones that are hidden when choosing a time period like this that (yes) is long, (yes) is recent... but which, unfortunately, is also (pretty clearly, IMO, given that the entire long term, secular trend here was rates going down with only a bit of backtracking just to go more down) a biased / error-prone way to try to make an averaged estimate using historical data for the expected return for leveraged long-term treasuries. And, of course, accordingly, it's also not going to give an unbiased estimate for the risks and expected return of a two-fund portfolio with a 60% allocation to them.

It could still make a lot of money, of course. But to me, at least, it looks like (in part) a bet that interest rates and inflation won't go up significantly over the time period invested. If that's true, it pays off, at least a little. If that's not true, it could crash and burn. And if the 30-year yield actually keeps going down from here over the next 20-30 years, combined with strong economic growth, then sure, I guess someone might get similar results.
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

MoneyMarathon wrote: Sat Jul 13, 2019 7:56 pm
MotoTrojan wrote: Sat Jul 13, 2019 6:22 pmMaybe more efficient but obviously less leverage so less return.
More leverage doesn't always give more return. I'm not an expert on bonds, so I can only give a simple take on this, but I think it should be pretty obvious that you can get negative returns with leverage... just by not making enough on something that would otherwise have positive yield (unleveraged).

Right now there are relatively-low rates for the 20-to-30-year yield and a compressed yield curve.

Image

The difference between the yield on the 1-month T-bill and average duration 25-years (around 2.48%) is only 0.32%.

The fund itself has a 1.09% expense ratio and 1.16% yield, mostly cancelling out in a tax-advantaged account.

I think the basic concept of TMF is that you want the volatility of getting duration risk but you don't really care about inherent expected return in an environment where interest rates are flat, because you've given up that in order to be in an expensive leveraged fund.

You only really profit from TMF in two scenarios: (1) rates go down, or, even better, (2) rates go down when stocks go down.

Those scenarios apply more often than not post-1980s, so you're sitting pretty if rates keep going down or if, at the very least, rates go down when stocks go down. And, since this is a big bet on sideways or declining rates, you're also hosed if rates go up and up.

But... all the unleveraged ETFs like IEF, TLT, and EDV also benefit from (1) and (2). And they pay out significant net interest in a sideways rates environment (or, any environment) even if you don't get any particular great advantage from the volatility in rates.

You only sometimes get more return with more leverage. You often get less, especially with TMF, which is designed as a play on the change in the long-term treasury yield and mostly lives up to performing that function even in a long-term buy-and-hold portfolio. I'm not sure why there's such strong recency bias in favor of TMF when it's very clear that these super-strong-performing periods were just going from higher rates to lower rates.

Play 60% TMF only if you believe you know the direction of interest rates over the time period in which you're invested, IMO.
Less leverage/exposure to equity was my point. You’d theoretically get more efficiency (but less return) by adjusting your UPRO allocation down to maintain risk parity with your mixed bond allocation.

TMF also pays out interest in neutral environments so not sure I see your point. Even if borrowing cost is as high as the long rate (flat or inverted curve) you’ll still get the 1x long rate, more than making up for the ER.
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

MotoTrojan wrote: Sun Jul 14, 2019 12:56 pm TMF also pays out interest in neutral environments so not sure I see your point.
I said: "I think it should be pretty obvious that you can get negative returns with leverage... just by not making enough on something that would otherwise have positive yield (unleveraged)."

Maybe it's easier to see with a simple graph. The point was relative return, net of all costs with dividends reinvested, when rates are flat. TMF has higher costs and doesn't really appreciate when rates go sideways over the time period looked at, while unleveraged ETFs have better returns (relatively) in that kind of environment.

In February 2015 and again in June 2019, the 30-year yield was 2.57, so that would be an overall neutral/sideways environment if investing over that time period. So it should illustrate whether and how much expenses in TMF create a drag relative to its unleveraged peers, independent of TMF's superior returns when rates go down.

This shows returns of EDV (in red) and TMF (in blue) from February 2015 to June 2019.

Image

Image

This shows annual rebalancing with 40% UPRO. (Sorry, I swapped the colors... 60% EDV is the one with lower volatility, now blue.)

Image

Quarterly rebalancing compensated for this effect a lot in this particular time period, interestingly enough.

Image
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

MoneyMarathon wrote: Sun Jul 14, 2019 2:10 pm
MotoTrojan wrote: Sun Jul 14, 2019 12:56 pm TMF also pays out interest in neutral environments so not sure I see your point.
I said: "I think it should be pretty obvious that you can get negative returns with leverage... just by not making enough on something that would otherwise have positive yield (unleveraged)."

Maybe it's easier to see with a simple graph. The point was relative return, net of all costs with dividends reinvested, when rates are flat. TMF has higher costs and doesn't really appreciate when rates go sideways over the time period looked at, while unleveraged ETFs have better returns (relatively) in that kind of environment.

In February 2015 and again in June 2019, the 30-year yield was 2.57, so that would be an overall neutral/sideways environment if investing over that time period. So it should illustrate whether and how much expenses in TMF create a drag relative to its unleveraged peers, independent of TMF's superior returns when rates go down.

This shows returns of EDV (in red) and TMF (in blue) from February 2015 to June 2019.

Image

Image

This shows annual rebalancing with 40% UPRO. (Sorry, I swapped the colors... 60% EDV is the one with lower volatility, now blue.)

Image

Quarterly rebalancing compensated for this effect a lot in this particular time period, interestingly enough.

Image
Very interesting indeed on the rebalancing. I wouldn’t consider using intermediate treasuries, not enough volatility, but I do find some form of including EDV intriguing.
EfficientInvestor
Posts: 392
Joined: Thu Nov 01, 2018 7:02 pm
Location: Alabama

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

Post by EfficientInvestor »

Kbg wrote: Sat Jul 13, 2019 8:25 pm Futures: To the person who posted earlier...please do the math. Futures (I’ve used them for years) are great but they are unforgiving, very unforgiving. A quick eyeball suggested to me you are quite undercapitalized. Read your futures agreement closely, depending on the firm you could lose your entire margin balance and get a bill for more depending on how and when you were liquidated. Do not forget futures are marked to market continuously the entire time they are active - Sunday evening until Friday afternoon and there can be huge overnight moves due to lack of liquidity combined with major news. Without sufficient cash you could easily wake up in the morning with zero in your account. However, you will be able to pull up an overnight chart and see the precise moment you were liquidated. It will be doubly painful as by the morning you probably would have been fine.
I assume you may have been referring to me with this comment. I currently have $32k invested in futures. My approximate notional exposures are $30k S&P 500 (/MES), $214k 2-year treasuries (/ZT), and $14k gold (/MGC). I have about $3k tied up in margin, $3k in liquid cash, and $26k in a ultra-short treasury ETF (MINT). My thought is I should always keep about 1X initial margin in cash and then sell some of the MINT etf if I need more cushion or if I get a margin call. Would you recommend something different? Any rules of thumb for how much buying power to maintain? Even if the notional value of my stock position dropped 10% in a single day, the $3k I have in cash could cover that. Maybe I should have an amount of cash equal to a 20% single-day drop? Or maybe I should talk to my broker, TastyWorks, and get a better understanding of what their protocol is in the event of a margin call. If I can get them to confirm that they will just sell some of my MINT etf to cover margin, then it's not a big issue.
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

EfficientInvestor wrote: Sun Jul 14, 2019 2:58 pm
Kbg wrote: Sat Jul 13, 2019 8:25 pm Futures: To the person who posted earlier...please do the math. Futures (I’ve used them for years) are great but they are unforgiving, very unforgiving. A quick eyeball suggested to me you are quite undercapitalized. Read your futures agreement closely, depending on the firm you could lose your entire margin balance and get a bill for more depending on how and when you were liquidated. Do not forget futures are marked to market continuously the entire time they are active - Sunday evening until Friday afternoon and there can be huge overnight moves due to lack of liquidity combined with major news. Without sufficient cash you could easily wake up in the morning with zero in your account. However, you will be able to pull up an overnight chart and see the precise moment you were liquidated. It will be doubly painful as by the morning you probably would have been fine.
I assume you may have been referring to me with this comment. I currently have $32k invested in futures. My approximate notional exposures are $30k S&P 500 (/MES), $214k 2-year treasuries (/ZT), and $14k gold (/MGC). I have about $3k tied up in margin, $3k in liquid cash, and $26k in a ultra-short treasury ETF (MINT). My thought is I should always keep about 1X initial margin in cash and then sell some of the MINT etf if I need more cushion or if I get a margin call. Would you recommend something different? Any rules of thumb for how much buying power to maintain? Even if the notional value of my stock position dropped 10% in a single day, the $3k I have in cash could cover that. Maybe I should have an amount of cash equal to a 20% single-day drop? Or maybe I should talk to my broker, TastyWorks, and get a better understanding of what their protocol is in the event of a margin call. If I can get them to confirm that they will just sell some of my MINT etf to cover margin, then it's not a big issue.
What is the “liquid cash” invested in? If it had a decent yield I’d just let it ride, or 1/2 of it.
EfficientInvestor
Posts: 392
Joined: Thu Nov 01, 2018 7:02 pm
Location: Alabama

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

Post by EfficientInvestor »

MotoTrojan wrote: Sun Jul 14, 2019 3:22 pm
EfficientInvestor wrote: Sun Jul 14, 2019 2:58 pm
Kbg wrote: Sat Jul 13, 2019 8:25 pm Futures: To the person who posted earlier...please do the math. Futures (I’ve used them for years) are great but they are unforgiving, very unforgiving. A quick eyeball suggested to me you are quite undercapitalized. Read your futures agreement closely, depending on the firm you could lose your entire margin balance and get a bill for more depending on how and when you were liquidated. Do not forget futures are marked to market continuously the entire time they are active - Sunday evening until Friday afternoon and there can be huge overnight moves due to lack of liquidity combined with major news. Without sufficient cash you could easily wake up in the morning with zero in your account. However, you will be able to pull up an overnight chart and see the precise moment you were liquidated. It will be doubly painful as by the morning you probably would have been fine.
I assume you may have been referring to me with this comment. I currently have $32k invested in futures. My approximate notional exposures are $30k S&P 500 (/MES), $214k 2-year treasuries (/ZT), and $14k gold (/MGC). I have about $3k tied up in margin, $3k in liquid cash, and $26k in a ultra-short treasury ETF (MINT). My thought is I should always keep about 1X initial margin in cash and then sell some of the MINT etf if I need more cushion or if I get a margin call. Would you recommend something different? Any rules of thumb for how much buying power to maintain? Even if the notional value of my stock position dropped 10% in a single day, the $3k I have in cash could cover that. Maybe I should have an amount of cash equal to a 20% single-day drop? Or maybe I should talk to my broker, TastyWorks, and get a better understanding of what their protocol is in the event of a margin call. If I can get them to confirm that they will just sell some of my MINT etf to cover margin, then it's not a big issue.
What is the “liquid cash” invested in? If it had a decent yield I’d just let it ride, or 1/2 of it.
Of the $29k that isn’t being used as margin, $3k isn’t in anything (I don’t think Tastyworks pays any interest on cash whereas Interactive Brokers does) and $26k is in MINT.
User avatar
Kevin M
Posts: 11730
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

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

Post by Kevin M »

privatefarmer wrote: Sun Jul 14, 2019 12:08 am <snip>
This strategy has worked very well since 1983 ...
<snip>
Rates have been dropping since then but you have to remember that, due to bond convexity, durations are now much higher than they were then and so even a minor drop in interest rates today will have a pronounced effect on price return.
Again, a technical note that this is not due to bond convexity, but due to bond duration being a function of both yield and coupon rate (and of course term to maturity as well). Convexity is related to the curvature of the yield sensitivity curve at a given yield, coupon rate, and maturity. Duration is related to the slope of that curve; it's the increasing magnitudes of the slopes of such curves as yields and coupon rates have declined that you are referencing.

Of more interest is that you also must consider that the duration benefit to bonds as yields have declined since the early 80s has been increasing along the way, not just recently, and that that benefit has compounded over the years. I thought this would be interesting to investigate.

As a simplified model, consider a 25-year constant maturity bond, starting at a yield and coupon rate of 15% (I'll use only par bonds in this example), with a decline of one percentage point per year in yield.

In year 1, with a drop in yield from 15% to 14%, the capital appreciation is 6.9% (incidentally, the modified duration at t0 is 6.5, and it's because of convexity that the price change for a one percentage point decrease in yield is slightly larger than the initial duration).

In year 2, with a drop in yield from 14% to 13%, the capital appreciation is 7.3%, so slightly larger because of the larger modified duration, with an initial value of 6.9. At this point the compound capital appreciation is (1 + 6.9%) * (1+7.3%) - 1 = 14.7%.

Continuing on with this, in the year in which yield drops from 4% to 3% the capital appreciation is 19.5% 17.4%, but more importantly, the compound capital appreciation from the beginning (call it the early 80s at 15%) to the end of this year is 315% 247%!

Now say we're starting fresh with a 25y constant maturity par bond at 3% (where we left off above). A drop in yield from 3% to 2% results in capital appreciation of 22.0% 19.5%. Of course this is much higher than the first year (15% to 14%), but it is only slightly higher than the preceding year (4% to 3%).

So we got significant benefit from the years toward the end of the first period (from 15% to 3%) due to higher durations, and those tail-end years contributed significantly to the compound appreciation over the entire period.

Continuing on with the second time period, starting at 3%, if we assume that the 25-year yield can't drop much below 0%, the maximum capital appreciation assuming the same one percentage point drop per year is 134% 82%. That's nothing to sneeze at, but it's a lot less than 315% 247%. Unless you believe that the 25-year yield can drop significantly below 0%, you simply can't get the returns going from 3% to 0% that you got going from 15% to 3%.

Of course this ignores the contribution of the higher coupon rates to the returns in earlier years. With my simplified model, the coupon returns would be higher than the capital returns until you got to a yield/rate of less than 10%. I'll probably follow up with an analysis including the coupons.

For what it's worth, here are the sensitivity curves of percentage price change to +/- one percentage point yield change for the bonds in the simplified model:

Image

If nothing else, I hope this reinforces the point that the increasing durations with decreasing yields and coupon rates are represented by the increasing steepness of the slopes of the curves, while convexity is related to the magnitude of curvature of each separate curve.

EDIT NOTES. In the initial post of this reply, I was off by one year or added an extra year for some of the numbers. For example, the capital appreciation I quoted for a drop from 4% to 3% was actually for the drop from 3% to 2%, and I included the year of the drop from 3% to 2% in the cumulative compound capital return for 15% -> 3%. Similarly, for the example starting at 3%, I had included a drop from 0% to -1% in the cumulative compound capital appreciation. I've corrected those errors.

Kevin
Last edited by Kevin M on Sun Jul 14, 2019 5:56 pm, edited 1 time in total.
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

MotoTrojan wrote: Sun Jul 14, 2019 2:24 pm I wouldn’t consider using intermediate treasuries, not enough volatility
UGLD (3x gold) seems pretty interesting: it's got plenty of volatility, it's fairly cost effective for a leveraged ETF, its correlation profile to the other assets is very good (i.e., low and sometimes moving the opposite direction), the underlying asset has generally maintained value in real terms over long time periods, and there are Bridgewater-esque reasons to think that the asset class would do well if the others were both doing poorly for any extended period of time (lower growth, higher inflation, paper currency including USD holding less value). Using the spreadsheet and the time period for which the gold data is available (1969-2018):

40% UPRO, 40% TMF, and 20% UGLD had the highest CAGR of the 40% UPRO portfolios with a TMF/UGLD split. It had 14.25% CAGR and 0.42 Sharpe, compared to 12.39% CAGR and 0.35 Sharpe for 60% TMF.

From 1982-2018, a bull market for bonds, it also fared relatively well, among options seen so far that hedged bets on TMF a little. It had 16.35% CAGR (0.57 Sharpe, 26.41% std. dev.) in this time period, compared to 19.9% for 60% TMF (0.62 Sharpe, 32.55% standard deviation).

Very interesting to me that it even lowered standard deviation in the recent period. I'm skeptical of relying on only two asset classes for diversification, especially when leveraging up like this. Those who have any doubts about the risks of both going down at the same time might want to add a third asset class like gold.
User avatar
Kevin M
Posts: 11730
Joined: Mon Jun 29, 2009 3:24 pm
Contact:

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

Post by Kevin M »

Kevin M wrote: Sun Jul 14, 2019 4:00 pm Of course this ignores the contribution of the higher coupon rates to the returns in earlier years. With my simplified model, the coupon returns would be higher than the capital returns until you got to a yield/rate of less than 10%. I'll probably follow up with an analysis including the coupons.
Factoring the coupon rates into the returns makes the difference between yields falling from 15% to 3% and from 3% to 0% even more dramatic.

The compound return of just coupons alone with yields/coupons dropping from 15% to 3% at one percentage point per year is 195% (=(1+15%)*(1+14%)* ... * (1+4%) - 1). Starting at 3% the number is 6%.

But, looking at coupon and capital returns separately doesn't give us an accurate understanding, as these returns are added together each year to get total return, and then compounded with the previous years total return.

The cumulative total return for the 15% -> 3% scenario is 835%.

The cumulative return for the 3% -> 0% scenario is 92%.

You just can't have bond returns starting at 3% anything close to what you got starting at 15%. The higher duration at lower yields and coupon rates isn't nearly enough to make up for the larger fall in yields combined with the higher coupon rates.

Kevin
Wiki ||.......|| Suggested format for Asking Portfolio Questions (edit original post)
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

HEDGEFUNDIE wrote: Tue Feb 05, 2019 2:56 pm As of November 2018, the borrow rate of UPRO's swaps was 2.87%. This cost is on top of the 0.92% expense ratio.

As of October 2018, the borrow rate of TMF's swaps was 1.83%. This cost is on top of the 1.09% expense ratio.
Interesting. Quick note here - if you had to choose UPRO or TMF, which would be better? Rebalanced quarterly over the fund histories:

20% TMF / 60% SPY / 20% cash -> 12.84% CAGR, 1.53 Sharpe, 2.93 Sortino, 0.41 market correlation.
60% TLT / 20% UPRO / 20% cash -> 12.46% CAGR, 1.49 Sharpe, 2.72 Sortino, 0.48 market correlation.

Perhaps some modest strategies could be devised around this, getting more leverage from more efficient side, long bonds (TMF).
Topic Author
HEDGEFUNDIE
Posts: 4801
Joined: Sun Oct 22, 2017 2:06 pm

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

Post by HEDGEFUNDIE »

MoneyMarathon wrote: Sun Jul 14, 2019 8:05 pm
HEDGEFUNDIE wrote: Tue Feb 05, 2019 2:56 pm As of November 2018, the borrow rate of UPRO's swaps was 2.87%. This cost is on top of the 0.92% expense ratio.

As of October 2018, the borrow rate of TMF's swaps was 1.83%. This cost is on top of the 1.09% expense ratio.
Interesting. Quick note here - if you had to choose UPRO or TMF, which would be better? Rebalanced quarterly over the fund histories:

20% TMF / 60% SPY / 20% cash -> 12.84% CAGR, 1.53 Sharpe, 2.93 Sortino, 0.41 market correlation.
60% TLT / 20% UPRO / 20% cash -> 12.46% CAGR, 1.49 Sharpe, 2.72 Sortino, 0.48 market correlation.

Perhaps some modest strategies could be devised around this, getting more leverage from more efficient side, long bonds (TMF).
If I wanted more modest leverage I would probably just buy PSLDX and let the professionals handle everything.
Kbg
Posts: 85
Joined: Thu Mar 23, 2017 11:33 am

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

Post by Kbg »

EfficientInvestor wrote: Sun Jul 14, 2019 2:58 pm
Kbg wrote: Sat Jul 13, 2019 8:25 pm Futures: To the person who posted earlier...please do the math. Futures (I’ve used them for years) are great but they are unforgiving, very unforgiving. A quick eyeball suggested to me you are quite undercapitalized. Read your futures agreement closely, depending on the firm you could lose your entire margin balance and get a bill for more depending on how and when you were liquidated. Do not forget futures are marked to market continuously the entire time they are active - Sunday evening until Friday afternoon and there can be huge overnight moves due to lack of liquidity combined with major news. Without sufficient cash you could easily wake up in the morning with zero in your account. However, you will be able to pull up an overnight chart and see the precise moment you were liquidated. It will be doubly painful as by the morning you probably would have been fine.
I assume you may have been referring to me with this comment. I currently have $32k invested in futures. My approximate notional exposures are $30k S&P 500 (/MES), $214k 2-year treasuries (/ZT), and $14k gold (/MGC). I have about $3k tied up in margin, $3k in liquid cash, and $26k in a ultra-short treasury ETF (MINT). My thought is I should always keep about 1X initial margin in cash and then sell some of the MINT etf if I need more cushion or if I get a margin call. Would you recommend something different? Any rules of thumb for how much buying power to maintain? Even if the notional value of my stock position dropped 10% in a single day, the $3k I have in cash could cover that. Maybe I should have an amount of cash equal to a 20% single-day drop? Or maybe I should talk to my broker, TastyWorks, and get a better understanding of what their protocol is in the event of a margin call. If I can get them to confirm that they will just sell some of my MINT etf to cover margin, then it's not a big issue.
Why yes I was. :-) Good to see the 26K in MINT, I must have missed that. What TW may or may not do with MINT in a bummer situation is probably worth a phone call to them. Use the scenario I provided and see what they have to say. Any good firms worth having your money at are NOT going to mess around with futures risk as they have to make good on it if their customer(s) do not. And as mentioned, it is not day time risk you really need to be concerned about it's the risk that could happen while you are sleeping. During a day time unfortunate event a margin reserves level will be hit and they will just start selling your stuff off to maintain the level and your MINT will be liquid and can be sold. No idea on MINT, but most ETFs do not trade O/N and if they do the market probably sucks. With TW you may be fine and they will give you some credit for the MINT holding, but you need to know the details. At the end of the day your cash stash is not what is driving the returns in this strategy, accordingly you should posture to make sure you are not going to have a sudden terminal cash flow problem. This is actually a small advantage for using 3x ETFs (O/N risk is not on you).
User avatar
privatefarmer
Posts: 725
Joined: Mon Sep 08, 2014 2:45 pm

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

Post by privatefarmer »

https://images.aqr.com/-/media/AQR/Docu ... s-Rise.pdf

Good article on risk parity and how it’s performed in rising rate environments.
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

privatefarmer wrote: Sun Jul 14, 2019 10:32 pm https://images.aqr.com/-/media/AQR/Docu ... s-Rise.pdf

Good article on risk parity and how it’s performed in rising rate environments.
Thanks! This quote stood out to me:

"The simple risk parity strategy outperformed, in both total and risk-adjusted returns, both the U.S. 60/40 portfolio and the global 60/40 portfolio over the full sample, over the 1981 to 2013 period of falling rates, and, perhaps to the surprise of many, over the 1947 to 1981 period of rising rates. If we were to study a risk parity strategy that only invested in U.S. stocks and bonds, as some studies have done, we would find that the two market risk parity portfolio would have underperformed during this period. However, risk parity is not implemented with only U.S. stocks and bonds. Having global diversification within stocks and bonds and broad diversification into commodities is more representative of actual risk parity portfolios, and incorporating these other markets changes the results dramatically."

An allocation to commodities does seem appropriate for a portfolio otherwise heavily leveraged with US stock and nominal treasuries.
comeinvest
Posts: 302
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

EfficientInvestor wrote: Wed Jul 10, 2019 10:22 pm
RandomWord wrote: Wed Jul 10, 2019 5:18 pm
robertmcd wrote: Wed Jul 10, 2019 5:06 pm Days like today are what worry me for this strategy, granted your stocks did well, but long term yields went up 3 bps while short term went down 10 bps. Somebody please come out with a leveraged 2 yr treasury ETF so I can run a risk parity strategy with it.
If you're willing to use futures instead of an ETF, the futures market has this covered.

But I'm not sure how much difference it makes in the long term, seems like the sort of thing that flucuates from day to day but overall averages out.
I started implementing the strategy with futures contracts a few weeks ago. I'm using a Roth IRA account that has about $32k in it. I have it spread across 2 contracts of /MES (micro S&P), 1 contract of /ZT (2-year treasuries), and 1 contract of /MGC (micro gold). This roughly equates to a 15/80/5 portfolio leveraged by a factor of 8. I have about $4k that is covering the margin requirements and I keep another $2k or so as cash in order to cover any daily fluctuations. The remaining cash is in MINT (ultra-short term bond ETF) to help offset some of the borrowing cost associated with the futures. The backtest below (since Nov 1991) represents what I'm aiming for. I'm assuming a quarterly rebalance since I will do the rebalance every quarter when I need to roll the futures contracts. In times when I may need 2.5 contracts of /MES or 1.5 contracts of /ZT, I will likely supplement the futures with regular ETFs (SPY, SHY, GLD) in order to hit my target exposures and maintain an 8X 15/80/5 portfolio.

https://www.portfoliovisualizer.com/bac ... 0&total3=0

This is my first venture into the use of futures, so I'm still learning some of the intricacies. However, my hope is that this will be a better alternative than the leveraged ETFs because it will help me avoid the 1% expense ratios and avoid the volatility drag due to the daily leverage reset. That drag could add up to another 1.5-2% over time. Therefore, I think I could conservatively expect a 2% increase in return per year over using leveraged ETFs. In addition, the futures allow me to use shorter term bonds that help limit my interest rate risk. We shall see how it goes.
Have you factored the richness of futures into your equation, i.e. the spread between implied financing interest rate and Libor? You can look them up in the roll data on the CME and ICE web sites (NOT the immigration enforcement, lol). If a 3x ETF has 0.9% ER, the finance cost (above Libor) per leverage unit as a multiple of equity is ca. 0.45%. (I attribute the ER of LETFs to the financing cost, as unleveraged ETFs have ERs near zero or zero.) Futures unfortunately had similar spreads as of late. Futures would still lack the volatility drag, but I hear it is also very small for LETFs with monthly rebalancing.
Last edited by comeinvest on Mon Jul 15, 2019 7:59 pm, edited 4 times in total.
schismal
Posts: 203
Joined: Sat Apr 13, 2019 8:53 pm

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

Post by schismal »

MoneyMarathon wrote: Mon Jul 15, 2019 3:11 amAn allocation to commodities does seem appropriate for a portfolio otherwise heavily leveraged with US stock and nominal treasuries.
So I did a quick napkin comparison between 40/60 UPRO/TMF and 40/40/20 UPRO/TMF/UGL

40/60: CAGR 25.18%, SD 21.50%, max DD -25.14%
40/40/20: CAGR 22.50%, SD 18.43%, max DD -23.42%

There are 90 months of historical data on PV. UGL helped the portfolio when TMF was down 75% of the time (34/45 months). When UPRO is down, the effect of UGL depends almost entirely on whether TMF is also down (in which case UGL helps) or if TMF is up (in which case UGL hurts). In this time series, UGL helped the portfolio in 40% of months when UPRO was down (11/27), which closely mirrors the behavior of TMF.

In ~20% of months (18/90), UGL was the only asset that was down, which hurt the portfolio. And it depressed returns in ~60% of months (9/14) when all three assets were up.

Looking at the broad patterns, UGL tends to help when TMF is down, which may provide some insurance in situations where UPRO and TMF diverge from their normal relationship and drop together. However, if the assets move as expected (e.g. TMF is up when UPRO is down), the addition of UGL dampens returns.

EDIT:

I should note that 40/40/20 UPRO/TMF/GLD showed the same relationship with TMF, but overall performance was better. UGL has an ER of 5%, compared to 0.5% for GLD.

CAGR 22.56%, SD 17.18%, max DD -20.83%
EfficientInvestor
Posts: 392
Joined: Thu Nov 01, 2018 7:02 pm
Location: Alabama

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

Post by EfficientInvestor »

comeinvest wrote: Mon Jul 15, 2019 6:03 am Have you factored the richness of futures into your equation, i.e. the spread between implied financing interest rate and Libor? You can look them up in the roll data on the CME and ICE web sites. If a 3x ETF has 0.9% ER, the finance cost (above Libor) per leverage unit as a multiple of equity is ca. 0.45%. (I attribute the ER of LETFs to the financing cost, as unleveraged ETFs have ERs near zero or zero.) Futures unfortunately had similar spreads as of late. Futures would still lack the volatility drag, but I hear it is also very small for LETFs with monthly rebalancing.
I have not read up on it too much. My assumption is sometimes the spread will be against me and sometimes it will be in my favor. Over time, I just assume it will be a wash. As for comparing it against LETFs, isn't this spread cost also present in the financing of the swaps that are used by the LETFs? I believe HEDGEFUNDIE has a better understanding of the spread associated with the swaps than I do, so maybe he can chime in. If they are in fact similar spreads, then it would be the same benefit/cost to both products and futures would still be the better product in terms of cost.
samsdad
Posts: 758
Joined: Sat Jan 02, 2016 6:20 pm

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

Post by samsdad »

I wanted to take a look at what 7-10 year intermediate treasuries did via the iShares ETF (IEF) and the 3x version (TYD) when yields bounced around but ultimately came back to the same point.

For portfoliovisualizer purposes, finding a spot where they do this at the beginning of a month for the starting date, and again at the end of the month for the endpoint was a little challenging, but ultimately I found September 1st, 2011 and May 31st, 2019 to be a good fit. This is a spliced and diced image from https://www.treasury.gov/resource-cente ... a=yieldAll:

Image

Note, that the 7-year side actually went up 54 bp. According to iShares, https://www.ishares.com/us/products/239 ... bond-etf#/ the effective duration of the portfolio of the fund is 7.53 years, and the weighted average maturity is 8.47 years as of last Friday.

Visually, here's what happened around the start point back in 2011 through last Thursday (as close to the May 31, 2019 endpoint as I was going to futz with). I put the mouse on the graph to show the beginning date, or as close as I could get it---the graph is a little unweidly for getting down to the nitty gritty. The image below is from https://www.macrotrends.net/2016/10-ye ... ield-chart:

Image

Here's what portfoliovisualizer shows: https://www.portfoliovisualizer.com/bac ... 0&total3=0. And here's the result for those of you who don't click on links from a stranger:

Image

I threw in the Vanguard Total Bond Market ETF (BND) just to see what it did over this time period. While the 3x fund TYD was certainly a rollercoaster, it managed to beat, by a considerable margin IMHO, its unleveraged holding, iShares IEF. Now, is this cherrypicking the endpoints? Yup. But the point of this was to show that in a period of bouncing around but ultimately standing still on the treasury side (at least intermediate treasuries), good things can happen. Here's a hypothetical 40/60 UPRO/TYD over the same dates, now benchmarked with the Vanguard 500 https://www.portfoliovisualizer.com/bac ... 0&total3=0:

Image

Keep in mind, these dates allow me to use the real-life funds, not the simulated versions, so all the expenses/scary stuff is baked into these results. I'll try to find a 20-30 year treasury version next, to see if the same has held true for those treasury durations at least once, historically.
User avatar
firebirdparts
Posts: 2252
Joined: Thu Jun 13, 2019 4:21 pm

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

Post by firebirdparts »

MoneyMarathon wrote: Sun Jul 14, 2019 4:11 am The apparently counter-intuitive thing here is that TMF has generated great returns... during 1987-2018. That's the most important part of what makes the backtest with a full TMF allocation so amazing..
I am not really qualified to post in this thread, but after reading it, I do see that at 120% equities, if the leveraged return on the bonds used is mediocre going forward, the results going forward is still going to be a little better (or worse) than 100% equities. I feel funny saying that, considering TMF since November, but I think I don't expect that kind of continuing performance.

So I have been thinking about what an alternative might be to TMF. As we've seen above with IEF. Thanks for posting that.
A fool and your money are soon partners
samsdad
Posts: 758
Joined: Sat Jan 02, 2016 6:20 pm

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

Post by samsdad »

Here's a look at long-term treasuries and what they did via the iShares ETF (TLT) and the 3x version (TMF) when yields bounced around but ultimately came back to the same point.

Again, I tried to find a point at the beginning or end of a month so we could see the performance of iShares 20+ year treasury ETF (TLT) and the corresponding 3x version (TMF). Here, the endpoint is the same as the intermediate treasuries in the post above, May 31st, 2019, but now we're starting in August of 2012. Again, I spliced and diced from the treasury website:

Image
https://www.treasury.gov/resource-cente ... a=yieldAll

And, again, note that the shorter-side duration 20-year treasury went up a little, just as above for 7-year intermediates. According to iShares, https://www.ishares.com/us/products/239 ... y-bond-etf, the effective duration of the fund TLT is 17.66 years, and weighted average maturity is 25.45 years.

Visually, the long-term treasuries did the same sort of up and down as the intermediate treasuries in the post above:

Image
https://www.macrotrends.net/2521/30-yea ... ield-chart

I can't explain the small discrepancy between the chart's yield figure for July 30, 2012, and the treasury figure for the same day, but I'm going to assume the government got it right. :shock: Besides, the purpose of the chart is to see visually the bouncing around to compare it with what the funds did according to portfoliovisualizer. And without further ado, here's what happened:

Image
https://www.portfoliovisualizer.com/bac ... 0&total3=0

Well, you still would've beaten the total bond market by holding TMF! :D Let's see if the minor enlargement in volatility :o by holding TMF over TLT would've been worth it when coupled with UPRO:

Image
https://www.portfoliovisualizer.com/bac ... 0&total3=0

I honestly had no idea what the results of any of these tests (intermediate or long-term treasuries) would be before posting the results, FYI.

Thoughts?
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

samsdad wrote: Mon Jul 15, 2019 11:55 am Here's a look at long-term treasuries and what they did via the iShares ETF (TLT) and the 3x version (TMF) when yields bounced around but ultimately came back to the same point.

Again, I tried to find a point at the beginning or end of a month so we could see the performance of iShares 20+ year treasury ETF (TLT) and the corresponding 3x version (TMF). Here, the endpoint is the same as the intermediate treasuries in the post above, May 31st, 2019, but now we're starting in August of 2012. Again, I spliced and diced from the treasury website:

Image
https://www.treasury.gov/resource-cente ... a=yieldAll

And, again, note that the shorter-side duration 20-year treasury went up a little, just as above for 7-year intermediates. According to iShares, https://www.ishares.com/us/products/239 ... y-bond-etf, the effective duration of the fund TLT is 17.66 years, and weighted average maturity is 25.45 years.

Visually, the long-term treasuries did the same sort of up and down as the intermediate treasuries in the post above:

Image
https://www.macrotrends.net/2521/30-yea ... ield-chart

I can't explain the small discrepancy between the chart's yield figure for July 30, 2012, and the treasury figure for the same day, but I'm going to assume the government got it right. :shock: Besides, the purpose of the chart is to see visually the bouncing around to compare it with what the funds did according to portfoliovisualizer. And without further ado, here's what happened:

Image
https://www.portfoliovisualizer.com/bac ... 0&total3=0

Well, you still would've beaten the total bond market by holding TMF! :D Let's see if the minor enlargement in volatility :o by holding TMF over TLT would've been worth it when coupled with UPRO:

Image
https://www.portfoliovisualizer.com/bac ... 0&total3=0

I honestly had no idea what the results of any of these tests (intermediate or long-term treasuries) would be before posting the results, FYI.

Thoughts?
I think it’s flawed to compare different volatilities on the bond side during an equity bull. The reason you want duration and/or leverage is to save the day during an equity crash.

While looking at real results I’d use EDV over TLT.
samsdad
Posts: 758
Joined: Sat Jan 02, 2016 6:20 pm

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

Post by samsdad »

MotoTrojan wrote: Mon Jul 15, 2019 12:26 pm While looking at real results I’d use EDV over TLT.
Doesn't TMF hold TLT, not EDV?

EDIT: According to page 146 of the annual report, TMF held some 635,000 shares of TLT as of Halloween last year.

http://direxioninvestments.onlineprospe ... Prospectus

Why would I look at EDV?
robertmcd
Posts: 554
Joined: Tue Aug 09, 2016 9:06 am

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

Post by robertmcd »

samsdad wrote: Mon Jul 15, 2019 12:32 pm
MotoTrojan wrote: Mon Jul 15, 2019 12:26 pm While looking at real results I’d use EDV over TLT.
Doesn't TMF hold TLT, not EDV?

EDIT: According to page 146 of the annual report, TMF held some 46000 shares of TLT as of Halloween last year.

http://direxioninvestments.onlineprospe ... Prospectus

Why would I look at EDV?
He is saying if you are gonna go unlevered you are better off with EDV than TLT. Allows a higher stock allocation for risk parity than TLT.
samsdad
Posts: 758
Joined: Sat Jan 02, 2016 6:20 pm

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

Post by samsdad »

robertmcd wrote: Mon Jul 15, 2019 12:39 pm
samsdad wrote: Mon Jul 15, 2019 12:32 pm
MotoTrojan wrote: Mon Jul 15, 2019 12:26 pm While looking at real results I’d use EDV over TLT.
Doesn't TMF hold TLT, not EDV?

EDIT: According to page 146 of the annual report, TMF held some 46000 shares of TLT as of Halloween last year.

http://direxioninvestments.onlineprospe ... Prospectus

Why would I look at EDV?
He is saying if you are gonna go unlevered you are better off with EDV than TLT. Allows a higher stock allocation for risk parity than TLT.
Ah, I must have missed that discussion. I'll go back and look as to why. Thanks
no simpler
Posts: 109
Joined: Sat Jul 13, 2019 4:54 pm

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

Post by no simpler »

MoneyMarathon wrote: Sun Jul 14, 2019 4:20 pm
MotoTrojan wrote: Sun Jul 14, 2019 2:24 pm I wouldn’t consider using intermediate treasuries, not enough volatility
UGLD (3x gold) seems pretty interesting: it's got plenty of volatility, it's fairly cost effective for a leveraged ETF, its correlation profile to the other assets is very good (i.e., low and sometimes moving the opposite direction), the underlying asset has generally maintained value in real terms over long time periods, and there are Bridgewater-esque reasons to think that the asset class would do well if the others were both doing poorly for any extended period of time (lower growth, higher inflation, paper currency including USD holding less value). Using the spreadsheet and the time period for which the gold data is available (1969-2018):

40% UPRO, 40% TMF, and 20% UGLD had the highest CAGR of the 40% UPRO portfolios with a TMF/UGLD split. It had 14.25% CAGR and 0.42 Sharpe, compared to 12.39% CAGR and 0.35 Sharpe for 60% TMF.

From 1982-2018, a bull market for bonds, it also fared relatively well, among options seen so far that hedged bets on TMF a little. It had 16.35% CAGR (0.57 Sharpe, 26.41% std. dev.) in this time period, compared to 19.9% for 60% TMF (0.62 Sharpe, 32.55% standard deviation).

Very interesting to me that it even lowered standard deviation in the recent period. I'm skeptical of relying on only two asset classes for diversification, especially when leveraging up like this. Those who have any doubts about the risks of both going down at the same time might want to add a third asset class like gold.
I completely agree. Worth reposting Dalio's all weather paper: https://www.bridgewater.com/resources/a ... -story.pdf

"All Weather grew out of Bridgewater’s effort to make sense of the world, to hold the portfolio today that will do reasonably well 20
years from now even if no one can predict what form of growth and inflation will prevail. When investing over the long run, all you
can have confidence in is that (1) holding assets should provide a return above cash, and (2) asset volatility will be largely driven by
how economic conditions unfold relative to current expectations (as well as how these expectations change). That’s it. Anything else (asset class returns, correlations, or even precise volatilities) is an attempt to predict the future. In essence, All Weather can be
sketched out on a napkin. It is as simple as holding four different portfolios each with the same risk, each of which does well in a particular environment: when (1) inflation rises, (2) inflation falls, (3) growth rises, and (4) growth falls relative to expectations."

As a personal anecdote, I built a portfolio based on MPT that outperformed S&P for decades in backtest with way less risk back in 2012. Since 2012, it's completely lagged S&P with similar vol. It's hard to internalize the concept of overfitting until you've done it. In my case, correlations, variance and returns w.r.t domestic vs international equities and commodities completely changed out of sample. Look at Canadian equity and emerging market returns in particular.
Last edited by no simpler on Mon Jul 15, 2019 12:58 pm, edited 3 times in total.
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

samsdad wrote: Mon Jul 15, 2019 12:43 pm
robertmcd wrote: Mon Jul 15, 2019 12:39 pm
samsdad wrote: Mon Jul 15, 2019 12:32 pm
MotoTrojan wrote: Mon Jul 15, 2019 12:26 pm While looking at real results I’d use EDV over TLT.
Doesn't TMF hold TLT, not EDV?

EDIT: According to page 146 of the annual report, TMF held some 46000 shares of TLT as of Halloween last year.

http://direxioninvestments.onlineprospe ... Prospectus

Why would I look at EDV?
He is saying if you are gonna go unlevered you are better off with EDV than TLT. Allows a higher stock allocation for risk parity than TLT.
Ah, I must have missed that discussion. I'll go back and look as to why. Thanks
Correct. EDV will give you the max interest rate sensitivity without adverse leveraged impacts.

My other point is the key though; comparisons with fixed equity allocation during a bull are imho flawed. In a market downturn if treasuries pop, 60% TMF will offset the equity losses much better than 60% TLT. 60% EDV is somewhere in the middle. Neither 60% EDV or TMF is proper risk parity though as more risk is on the equity side.

I don't have data for 30 year strips going back to 1955 but using long-treasuries as a proxy it does actually seem that 50/50 TMF/LTT on the bond side gave a nice compromise. I believe it was a 4% CAGR for the 1955-1982 range, which while unfortunate compared to the S&P's 9% is not financial disaster.

Using UPRO & EDV with the 1-month risk-parity look-back also performed quite efficiently. Using 50/50 TMF/EDV on the bond side with a 1-month risk-parity exposure to UPRO is an area I am going to look at simulating a bit further.
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

Here's a paper out of JP Morgan Chase on the theory behind risk parity (understood as an equal-risk allocation) and other approaches:

https://am.jpmorgan.com/blobcontent/137 ... Parity.pdf

Basically, equal-risk allocation can do well when you assume that financial markets, in the long run, seek similar risk-adjusted returns (Sharpe) from all the asset classes in which you're invested. If so, you should just ignore historical data on relative performance, collect risk premiums across asset classes, and get the full benefits of diversification by building a portfolio with equal risk in each asset class. Investors can end up with an underperforming portfolio if some asset classes actually provide higher risk-adjusted returns than others.

Maximizing Sharpe on historical data does well when you assume that the historical data for that time period is a good representation of the expected return of the asset classes going forward, not just volatility. It attempts to learn the different Sharpe ratio of the different asset classes. It introduces an additional parameter to learn about each asset class -- expected return -- and so it is more sensitive to overfitting, biased data, and poor performance out of sample. Investors frequently end up overweighted and with an underperforming portfolio if their returns from different asset classes don't look like the historical returns they looked at.

Sharpe ratios change over time, since returns change over time. A truly optimal leveraged asset allocation would use the expected returns for each asset class in the future, but that's unknown and estimates are not precise or reliable. The conclusion of the paper, I think, is to deliberately avoid constructing a precisely optimal portfolio. Instead, make a good-enough portfolio that will still be good enough in different market conditions, since those conditions are unknown and unlikely to be exactly like the historical period looked at, and since most investors aren't going to be able to sit through 50-60 years of waiting for market conditions to become favorable to some highly concentrated portfolio again.

So suppose you wanted to construct a leveraged portfolio considering US large caps, long-term treasuries, and gold as the three asset classes. The non-simulated historical data going back to 1978 says these asset classes had these returns in this time period:

US large cap: 11.48% CAGR, 14.85% standard deviation, 0.51 Sharpe
Long term treasury: 8.46% CAGR, 10.98% standard deviation, 0.4 Sharpe
Gold: 5.15% CAGR, 18.85% standard deviation, 0.13 Sharpe

Maximizing Sharpe given this data for historical returns gives you: 47.51% US large caps, 45.81% LTT, 6.68% gold.

Assigning equal risk to each asset class given just the volatility in each: 32.12% US large caps, 42.57% LTT, 25.31% gold.

Both portfolios have a touch of gold in the 1978-2018 data but the equal-risk portfolio has a lot more. The equal-risk portfolio also has a much larger ratio of bonds to large cap stock (1.3x as much for equal risk compared to roughly 1:1 for maximizing Sharpe).

Both approaches assume the historical data is useful for measuring volatility. The maximizing-Sharpe portfolio will be better, going forward, if the relative returns from stocks, bonds, and gold are similar. Then it will pay to have more in stocks and less in gold, compared to the equal-risk portfolio.

As actually practiced, many investors who use "risk parity" ideas, instead of a traditional equity-risk-heavy portfolio, don't mechanically build portfolios that match volatility between asset classes. Instead they find a way to split the middle between the maximizing-historical-Sharpe approach (with its tendency to overfit the historical data and concentrate risk due to misjudged relative returns going forward) and the equalizing-risk approach (with its seemingly-flawed assumption that each asset class delivers the same risk-adjusted returns).

And sometimes they try to use a bit of thought about the nature of the asset classes and why they might deliver different, relative risk-adjusted returns. For example, commodities might be expected to return little in real terms, while investors in stocks and bonds are putting their capital at risk so that it can be put to productive use. So it may seem reasonable that commodities, even gold, over long periods of time would still have lower risk-adjusted returns; this doesn't seem like it's just a fluke in the historical data.

They also might think about the mechanics of why certain asset classes had certain returns, and what would happen if those things changed. For example, since the 1980s, the US dollar has generally been strong and has importances as the premier reserve currency, which weakens the value of gold in US dollar terms. The yield on nominal treasuries has fallen over that time period, boosting long term treasuries, something that might reverse. And the US has been relatively much more prosperous than the rest of the world, boosting US equity, something that might also reverse. If all of these trends reversed at the same time, you would expect gold to benefit. (As it did, in the early part of that period and earlier in the 1970s.)

Holding different asset classes that can benefit if the others suffer is a win from diversification. Rebalancing can be a boost to diversified portfolios. But the biggest boost is that you're more likely to actually get the returns you're hoping for. It doesn't just reduce volatility for the sake of having a less bumpy ride to riches. It actually reduces the chance that you're on the road to ruin while it improves your odds of getting a higher CAGR over the time period invested. The very-highly-concentrated portfolio with incredible-looking returns doesn't generally have higher expected CAGR; it generally has poorly-estimated risks and poorly-estimated returns.

That in mind, somebody might want to consider adding an allocation to commodities, such as gold, of at least 6.7% (since this was overall not a great time period for gold relative to stocks and bonds) and less than 25% (since equal-risk isn't really needed, just sufficiently-hedged risk from the other asset classes, US stocks and nominal long-term treasuries).

When evaluating this on recent data, I wouldn't just go for maximizing returns. We know that returns of different asset classes change over time, and some gain when others lose, so we want to hold some asset class that gains when both the S&P 500 and LTTs lose (and which at the same time may often lose when the other two make gains). Since Sharpe is a function of returns, I wouldn't just go for maximizing Sharpe over a historical time period. If you've already chosen asset classes that have some expected return and in which you want to invest for diversification, then some of the more interesting characteristics of a portfolio, over a short time period (even 40 years is short), are the risk characteristics (volatility of the portfolio) and how well the asset classes did in balancing out the others. When deciding a split between asset classes, the actual returns over a recent time period should be heavily discounted & you should, probably, want to do a little less than perfectly optimal on that data (trying to avoid overfitting). The perfect is the enemy of the good here.
comeinvest
Posts: 302
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest »

EfficientInvestor wrote: Mon Jul 15, 2019 6:41 am
comeinvest wrote: Mon Jul 15, 2019 6:03 am Have you factored the richness of futures into your equation, i.e. the spread between implied financing interest rate and Libor? You can look them up in the roll data on the CME and ICE web sites. If a 3x ETF has 0.9% ER, the finance cost (above Libor) per leverage unit as a multiple of equity is ca. 0.45%. (I attribute the ER of LETFs to the financing cost, as unleveraged ETFs have ERs near zero or zero.) Futures unfortunately had similar spreads as of late. Futures would still lack the volatility drag, but I hear it is also very small for LETFs with monthly rebalancing.
I have not read up on it too much. My assumption is sometimes the spread will be against me and sometimes it will be in my favor. Over time, I just assume it will be a wash. As for comparing it against LETFs, isn't this spread cost also present in the financing of the swaps that are used by the LETFs? I believe HEDGEFUNDIE has a better understanding of the spread associated with the swaps than I do, so maybe he can chime in. If they are in fact similar spreads, then it would be the same benefit/cost to both products and futures would still be the better product in terms of cost.
I don't know how LETFs are implemented or if they incur a spread to the benchmark interest rate. But I read a few articles that say the richness of futures started a few years ago as a result of increased government regulation in the wake of the financial crisis, and might be more or less permanent. Before the increased regulation, futures investors could indeed finance at close to the benchmark rate, and sometimes even with a small additional return from negative spread. My understanding is that the change is specific to futures. If this is the case, the benefit of using futures in terms of cost would be marginal. I would be happy to learn about the intrinsic financing cost of LETFs.
Last edited by comeinvest on Mon Jul 29, 2019 3:39 am, edited 1 time in total.
no simpler
Posts: 109
Joined: Sat Jul 13, 2019 4:54 pm

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

Post by no simpler »

MoneyMarathon wrote: Mon Jul 15, 2019 4:11 pm
When evaluating this on recent data, I wouldn't just go for maximizing returns. We know that returns of different asset classes change over time, and some gain when others lose, so we want to hold some asset class that gains when both the S&P 500 and LTTs lose (and which at the same time may often lose when the other two make gains). Since Sharpe is a function of returns, I wouldn't just go for maximizing Sharpe over a historical time period. If you've already chosen asset classes that have some expected return and in which you want to invest for diversification, then some of the more interesting characteristics of a portfolio, over a short time period (even 40 years is short), are the risk characteristics (volatility of the portfolio) and how well the asset classes did in balancing out the others. When deciding a split between asset classes, the actual returns over a recent time period should be heavily discounted & you should, probably, want to do a little less than perfectly optimal on that data (trying to avoid overfitting). The perfect is the enemy of the good here.
^This. If you backtest a portfolio of 90% tbills and 10% bitcoin, it looks amazing. Max drawdown is 10% and you'd still have a 500%+ return in 7 years.

I work as a data scientist. So much of our work is just trying to prevent overfitting. It's incredibly hard to avoid. You need to take extreme precautions to avoid contamination of hold out sets when doing cross validation. I also have the very real world experience of building a portfolio according to MPT and then having the live portfolio behave nothing like the past. Based on both academic, professional, and just straight up hard knocks, I would not extrapolate based on historical time series. Nobody knows what will happen.

That said, I like the risk parity approaches in that JPM paper and the Bridgewater papers. It's based on a more qualitative understanding of how assets respond to economic conditions and makes very few assumptions. But it definitely makes sense to hold more than two assets using this strategy so that you can survive multiple environments.
no simpler
Posts: 109
Joined: Sat Jul 13, 2019 4:54 pm

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

Post by no simpler »

MoneyMarathon wrote: Mon Jul 15, 2019 4:11 pm Here's a paper out of JP Morgan Chase on the theory behind risk parity (understood as an equal-risk allocation) and other approaches:

https://am.jpmorgan.com/blobcontent/137 ... Parity.pdf

Basically, equal-risk allocation can do well when you assume that financial markets, in the long run, seek similar risk-adjusted returns (Sharpe) from all the asset classes in which you're invested. If so, you should just ignore historical data on relative performance, collect risk premiums across asset classes, and get the full benefits of diversification by building a portfolio with equal risk in each asset class. Investors can end up with an underperforming portfolio if some asset classes actually provide higher risk-adjusted returns than others.

Whoa, wait - I went through the whole paper. This is unwittingly the best argument for risk parity and not using return forecasts I've seen. Notice this was published in October 2012. They forecast that the excess return premium for commodities is 6%, with a lower vol (10%) than S&P 500. In other words, they forecast that commodities have a higher sharpe than equities, which is qualitatively insane considering they aren't assets that produce cashflows. they then advise much higher weightings to commodities than equities. And of course, had you been weighted heavily in commodities and then leveraged, you would have lost your shirt from 2012 -2019.
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

MoneyMarathon wrote: Sat Jul 13, 2019 7:00 pm Noting again: UPRO is a 3x SP500 fund, TMF is a 3x 20+ year treasury fund, and IEF is a 7-10 year treasury fund. All numbers (including, particularly, max drawdown, standard deviation, & risk metrics) are only from year to year calculations. Someone's impression of risk might be different if they check the news more than once a year. :)
MotoTrojan wrote: Sat Jul 13, 2019 6:22 pm Curious how your options did 1982 to present.
Jan 1982 - Dec 2018

40% UPRO, 00% TMF, 60% IEF -> 13.29% CAGR, 19.75% std. dev., 0.56 Sharpe, 2.60 Sortino, 35.07% max drawdown.
40% UPRO, 10% TMF, 50% IEF -> 14.68% CAGR, 20.69% std. dev., 0.61 Sharpe, 3.91 Sortino, 30.39% max drawdown.
40% UPRO, 20% TMF, 40% IEF -> 15.94% CAGR, 22.26% std. dev., 0.64 Sharpe, 4.18 Sortino, 25.6% max drawdown.
40% UPRO, 30% TMF, 30% IEF -> 17.09% CAGR, 24.34% std. dev., 0.64 Sharpe, 4.05 Sortino, 22.76% max drawdown.
40% UPRO, 40% TMF, 20% IEF -> 18.13% CAGR, 26.81% std. dev., 0.64 Sharpe, 3.94 Sortino, 20.88% max drawdown.
40% UPRO, 50% TMF, 10% IEF -> 19.06% CAGR, 29.57% std. dev., 0.63 Sharpe, 3.85 Sortino, 23.39% max drawdown.
40% UPRO, 60% TMF, 00% IEF -> 19.9% CAGR, 32.55% std. dev., 0.62 Sharpe, 3.78 Sortino, 26.34% max drawdown.

100% SPY (S&P 500 index) -> 11.19% CAGR, 16.31% std. dev., 0.53 Sharpe, 1.33 Sortino, 37.80% max drawdown.

Jan 1955 - Dec 1981

40% UPRO, 00% TMF, 60% IEF -> 9.08% CAGR, 23.72% std. dev., 0.25 Sharpe, 0.95 Sortino, 39.09% max drawdown.
40% UPRO, 10% TMF, 50% IEF -> 7.86% CAGR, 22.77% std. dev., 0.20 Sharpe, 0.73 Sortino, 41.78% max drawdown.
40% UPRO, 20% TMF, 40% IEF -> 6.61% CAGR, 23.14% std. dev., 0.15 Sharpe, 0.54 Sortino, 44.41% max drawdown.
40% UPRO, 30% TMF, 30% IEF -> 5.33% CAGR, 23.65% std. dev., 0.10 Sharpe, 0.34 Sortino, 46.98% max drawdown.
40% UPRO, 40% TMF, 20% IEF -> 4.01% CAGR, 24.28% std. dev., 0.06 Sharpe, 0.17 Sortino, 49.49% max drawdown.
40% UPRO, 50% TMF, 10% IEF -> 2.66% CAGR, 25.03% std. dev., 0.01 Sharpe, 0.03 Sortino, 54.13% max drawdown.
40% UPRO, 60% TMF, 00% IEF -> 1.26% CAGR, 25.89% std. dev., -0.03 Sharpe, -0.08 Sortino, 63.05% max drawdown.

100% SPY (S&P 500 index) -> 8.57% CAGR, 17.51% std. dev., 0.26 Sharpe, 0.71 Sortino, 37.39% max drawdown.

Jan 1955 - Dec 2018

40% UPRO, 00% TMF, 60% IEF -> 11.49% CAGR, 20.89% std. dev., 0.42 Sharpe, 2.18 Sortino, 39.09% max drawdown.
40% UPRO, 10% TMF, 50% IEF -> 11.75% CAGR, 21.65% std. dev., 0.42 Sharpe, 2.16 Sortino, 41.78% max drawdown.
40% UPRO, 20% TMF, 40% IEF -> 11.91% CAGR, 22.9% std. dev., 0.41 Sharpe, 2.15 Sortino, 44.41% max drawdown.
40% UPRO, 30% TMF, 30% IEF -> 11.98% CAGR, 24.55% std. dev., 0.40 Sharpe, 2.03 Sortino, 46.98% max drawdown.
40% UPRO, 40% TMF, 20% IEF -> 11.95% CAGR, 26.54% std. dev., 0.38 Sharpe, 1.89 Sortino, 49.49% max drawdown.
40% UPRO, 50% TMF, 10% IEF -> 11.84% CAGR, 28.79% std. dev., 0.37 Sharpe, 1.77 Sortino, 54.13% max drawdown.
40% UPRO, 60% TMF, 00% IEF -> 11.65% CAGR, 31.24% std. dev., 0.35 Sharpe, 1.67 Sortino, 63.05% max drawdown.

100% SPY (S&P 500 index) -> 10.08% CAGR, 17.05% std. dev., 0.41 Sharpe, 1.42 Sortino, 37.80% max drawdown.
So I have been looking more into incorporating some unleveraged bonds to improve efficiency in flat/rising rate environments and mildly reduce leverage overall; I think this will make my 1-month lookback risk parity a bit more palatable by reducing the extremes of the allocation that UPRO could be month to month, and also reduces the "bet" on rates continuing down. I do not have simulated data for 30 year strips, but using the same LTT data from Siamonds spreadsheet (run in PV though) I ran some results that I found quite interesting. I'll let you run your numbers to get a proper 1:1 comparison (my drawdown looks a bit more but it is monthly) but for all of the major time-periods the return of 40% UPRO, 30% TMF, and 30% 20-year unleveraged treasuries rebalanced annually comes quite close to the 30% IEF cases, within <<1% CAGR/Std. It seems more of the benefit came from the lack of leverage than the intermediate term.

What I am leaning towards is 50/50 EDV/TMF for the bond side, with it's variance balanced monthly against UPRO.
MoneyMarathon
Posts: 992
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon »

MotoTrojan wrote: Mon Jul 15, 2019 9:40 pmreduces the "bet" on rates continuing down
A little. Right now, effective duration is 7.53 years for IEF, but 17.66 years for TLT and 24.37 years for EDV. This suggests that the price of EDV is roughly three times more sensitive to changes in yield. And TMF has an effective duration of 54.66 years. So, right now, IEF is about 1/6th as sensitive as TMF while EDV is only 1/2 as sensitive. Also, since IEF is based on the nominal yield curve at 7-10 years out, it's less than 90% correlated to the others. TLT, EDV, and TMF are all very highly correlated (98%-100%).

Based on a fixed income factor model, with term and credit risk factors, IEF has 0.46 term risk, EDV has 1.7 term risk, and TMF has 3.4 term risk (all have no or even slightly negative correlation to credit risk). Also, similar to the way that people are willing to accept lower risk-adjusted return in order to get more beta (above 1) without any actual cost of leverage and seem to demand higher risk-adjusted return for lower beta... IEF according to this factor regression has 0.79% annual alpha (compensating for getting less term risk for capital invested), EDV has -0.02% annual alpha, and TMF has -2.85% annual alpha (directly attributable to the expense ratio and the expenses of leverage).

Along with its low volatility (5.58% since 2010, compared to 12.51% for TLT, 19.4% for EDV and 38.33% for TMF), this provides more opportunity for beneficial rebalancing (especially with the more-volatile EDV or TMF). I think the purpose of adding intermediate-term treasuries, if someone wants to, is just to have something that never has strong movement, holds value with a yield over time, benefits from a flight to quality (practically no liquidity or credit risk), contributes modestly to the duration risk of the portfolio, and always can rebalance into other, declining parts of the portfolio without relying on anti-correlation working every time.
MotoTrojan wrote: Mon Jul 15, 2019 9:40 pmI'll let you run your numbers to get a proper 1:1 comparison (my drawdown looks a bit more but it is monthly) but for all of the major time-periods the return of 40% UPRO, 30% TMF, and 30% 20-year unleveraged treasuries rebalanced annually comes quite close to the 30% IEF cases, within <<1% CAGR/Std. It seems more of the benefit came from the lack of leverage than the intermediate term.
It depends on whether you're measuring a period of rising or falling rates. With falling rates, it doesn't make a big difference and you'll be posting high CAGR with a big enough exposure to duration risk (TLT/EDV/TMF) and beta (UPRO). With rising rates, the very real possibility that all components of the portfolio perform very poorly, along with an annual drag on returns for long stretches from having such high duration risk and getting the bad side of that risk, makes things look very different.

Jan 1982 - Dec 2018

40% UPRO, 60% IEF ----------> 13.29% CAGR, 0.56 Sharpe, 2.60 Sortino
40% UPRO, 30% EDV, 30% IEF -> 15.23% CAGR, 0.61 Sharpe, 3.64 Sortino
40% UPRO, 30% TMF, 30% IEF -> 17.09% CAGR, 0.64 Sharpe, 4.05 Sortino
40% UPRO, 30% TMF, 30% EDV -> 18.48% CAGR, 0.62 Sharpe, 3.48 Sortino
40% UPRO, 60% TMF ----------> 19.9% CAGR, 0.62 Sharpe, 3.78 Sortino

100% SPY (S&P 500 index) -> 11.19% CAGR, 0.53 Sharpe, 1.33 Sortino

Jan 1955 - Dec 1981

40% UPRO, 60% IEF ----------> 9.08% CAGR, 0.25 Sharpe, 0.95 Sortino
40% UPRO, 30% EDV, 30% IEF -> 6.78% CAGR, 0.16 Sharpe, 0.54 Sortino
40% UPRO, 30% TMF, 30% IEF -> 5.33% CAGR, 0.10 Sharpe, 0.34 Sortino
40% UPRO, 30% TMF, 30% EDV -> 2.87% CAGR, 0.02 Sharpe, 0.04 Sortino
40% UPRO, 60% TMF ----------> 1.26% CAGR, -0.03 Sharpe, -0.08 Sortino

100% SPY (S&P 500 index) -> 8.57% CAGR, 0.26 Sharpe, 0.71 Sortino

Jan 1955 - Dec 2018

40% UPRO, 60% IEF ----------> 11.49% CAGR, 0.42 Sharpe, 2.18 Sortino
40% UPRO, 30% EDV, 30% IEF -> 11.58% CAGR, 0.40 Sharpe, 2.02 Sortino
40% UPRO, 30% TMF, 30% IEF -> 11.98% CAGR, 0.40 Sharpe, 2.03 Sortino
40% UPRO, 30% TMF, 30% EDV -> 11.62% CAGR, 0.36 Sharpe, 1.65 Sortino
40% UPRO, 60% TMF ----------> 11.65% CAGR, 0.35 Sharpe, 1.67 Sortino

100% SPY (S&P 500 index) -> 10.08% CAGR, 0.41 Sharpe, 1.42 Sortino

Of these options, the ones that (over the time period 1955-2018) get Sharpe above 0.4 and Sortino above 2.0 and the ones that (over the time period 1955-1981) get CAGR over 5% all have added the allocation to intermediate-term treasuries (30% IEF). So if someone is looking at these 4 assets only (ignoring, for example, commodities) and wants to reduce the risks of loss and negative real returns in a rising rates environment, I think intermediate term treasuries are necessary. (Not necessarily with this 40/30/30 split, of course.)

Whether to go EDV or TMF is much more optional.
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

MoneyMarathon wrote: Tue Jul 16, 2019 1:26 am
MotoTrojan wrote: Mon Jul 15, 2019 9:40 pmreduces the "bet" on rates continuing down
A little. Right now, effective duration is 7.53 years for IEF, but 17.66 years for TLT and 24.37 years for EDV. This suggests that the price of EDV is roughly three times more sensitive to changes in yield. And TMF has an effective duration of 54.66 years. So, right now, IEF is about 1/6th as sensitive as TMF while EDV is only 1/2 as sensitive. Also, since IEF is based on the nominal yield curve at 7-10 years out, it's less than 90% correlated to the others. TLT, EDV, and TMF are all very highly correlated (98%-100%).

Based on a fixed income factor model, with term and credit risk factors, IEF has 0.46 term risk, EDV has 1.7 term risk, and TMF has 3.4 term risk (all have no or even slightly negative correlation to credit risk). Also, similar to the way that people are willing to accept lower risk-adjusted return in order to get more beta (above 1) without any actual cost of leverage and seem to demand higher risk-adjusted return for lower beta... IEF according to this factor regression has 0.79% annual alpha (compensating for getting less term risk for capital invested), EDV has -0.02% annual alpha, and TMF has -2.85% annual alpha (directly attributable to the expense ratio and the expenses of leverage).

Along with its low volatility (5.58% since 2010, compared to 12.51% for TLT, 19.4% for EDV and 38.33% for TMF), this provides more opportunity for beneficial rebalancing (especially with the more-volatile EDV or TMF). I think the purpose of adding intermediate-term treasuries, if someone wants to, is just to have something that never has strong movement, holds value with a yield over time, benefits from a flight to quality (practically no liquidity or credit risk), contributes modestly to the duration risk of the portfolio, and always can rebalance into other, declining parts of the portfolio without relying on anti-correlation working every time.
MotoTrojan wrote: Mon Jul 15, 2019 9:40 pmI'll let you run your numbers to get a proper 1:1 comparison (my drawdown looks a bit more but it is monthly) but for all of the major time-periods the return of 40% UPRO, 30% TMF, and 30% 20-year unleveraged treasuries rebalanced annually comes quite close to the 30% IEF cases, within <<1% CAGR/Std. It seems more of the benefit came from the lack of leverage than the intermediate term.
It depends on whether you're measuring a period of rising or falling rates. With falling rates, it doesn't make a big difference and you'll be posting high CAGR with a big enough exposure to duration risk (TLT/EDV/TMF) and beta (UPRO). With rising rates, the very real possibility that all components of the portfolio perform very poorly, along with an annual drag on returns for long stretches from having such high duration risk and getting the bad side of that risk, makes things look very different.

Jan 1982 - Dec 2018

40% UPRO, 60% IEF ----------> 13.29% CAGR, 0.56 Sharpe, 2.60 Sortino
40% UPRO, 30% EDV, 30% IEF -> 15.23% CAGR, 0.61 Sharpe, 3.64 Sortino
40% UPRO, 30% TMF, 30% IEF -> 17.09% CAGR, 0.64 Sharpe, 4.05 Sortino
40% UPRO, 30% TMF, 30% EDV -> 18.48% CAGR, 0.62 Sharpe, 3.48 Sortino
40% UPRO, 60% TMF ----------> 19.9% CAGR, 0.62 Sharpe, 3.78 Sortino

100% SPY (S&P 500 index) -> 11.19% CAGR, 0.53 Sharpe, 1.33 Sortino

Jan 1955 - Dec 1981

40% UPRO, 60% IEF ----------> 9.08% CAGR, 0.25 Sharpe, 0.95 Sortino
40% UPRO, 30% EDV, 30% IEF -> 6.78% CAGR, 0.16 Sharpe, 0.54 Sortino
40% UPRO, 30% TMF, 30% IEF -> 5.33% CAGR, 0.10 Sharpe, 0.34 Sortino
40% UPRO, 30% TMF, 30% EDV -> 2.87% CAGR, 0.02 Sharpe, 0.04 Sortino
40% UPRO, 60% TMF ----------> 1.26% CAGR, -0.03 Sharpe, -0.08 Sortino

100% SPY (S&P 500 index) -> 8.57% CAGR, 0.26 Sharpe, 0.71 Sortino

Jan 1955 - Dec 2018

40% UPRO, 60% IEF ----------> 11.49% CAGR, 0.42 Sharpe, 2.18 Sortino
40% UPRO, 30% EDV, 30% IEF -> 11.58% CAGR, 0.40 Sharpe, 2.02 Sortino
40% UPRO, 30% TMF, 30% IEF -> 11.98% CAGR, 0.40 Sharpe, 2.03 Sortino
40% UPRO, 30% TMF, 30% EDV -> 11.62% CAGR, 0.36 Sharpe, 1.65 Sortino
40% UPRO, 60% TMF ----------> 11.65% CAGR, 0.35 Sharpe, 1.67 Sortino

100% SPY (S&P 500 index) -> 10.08% CAGR, 0.41 Sharpe, 1.42 Sortino

Of these options, the ones that (over the time period 1955-2018) get Sharpe above 0.4 and Sortino above 2.0 and the ones that (over the time period 1955-1981) get CAGR over 5% all have added the allocation to intermediate-term treasuries (30% IEF). So if someone is looking at these 4 assets only (ignoring, for example, commodities) and wants to reduce the risks of loss and negative real returns in a rising rates environment, I think intermediate term treasuries are necessary. (Not necessarily with this 40/30/30 split, of course.)

Whether to go EDV or TMF is much more optional.
Thank you for the numbers. Interesting indeed. If it isn't much effort I would be curious to see the performance of 40% UPRO, 30% TMF, 30% TLT/similar long treasury fund, as that is what I was directly comparing and found performed not far from the UPRO, TMF, IEF instance.

While duration is a good way to compare the interest rate sensitivity, it ignores the offsetting gains due to a higher interest rate which is amplified in the unleveraged funds relative to the leveraged TMF's price change. When rates go up 1% TMF will be expected to lose ~2x as much as EDV's price, but the interest rate improvement is roughly equivalent after accounting for the short-term leverage costs. My simplistic way of thinking about it is that TMF's interest payments are roughly equivalent to the unleveraged bond, but it's price movements are amplified.
no simpler
Posts: 109
Joined: Sat Jul 13, 2019 4:54 pm

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

Post by no simpler »

The secret is a 10% allocation to bitcoin:
https://www.portfoliovisualizer.com/bac ... 0&total3=0

Mostly kidding to make a point about the dangers of backtests. Bitcoin does have a nice risk/reward asymmetry when held in small amounts, and has no correlation with market.
MotoTrojan
Posts: 10776
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan »

no simpler wrote: Tue Jul 16, 2019 9:34 am The secret is a 10% allocation to bitcoin:
https://www.portfoliovisualizer.com/bac ... 0&total3=0

Mostly kidding to make a point about the dangers of backtests. Bitcoin does have a nice risk/reward asymmetry when held in small amounts, and has no correlation with market.
Good point. I am using backtests to inform me of sensitivities but in general am basing the construction on core beliefs, namely those mentioned in the OP along with a belief that momentum in volatility is stronger than in expected return (hence the frequent rebalancing based on the past months volatility). Backtesting is still a valuable way to get an idea of how different constructions (in my case really, amounts of leverage) would perform in different market conditions, at-least those that have happened to date.
no simpler
Posts: 109
Joined: Sat Jul 13, 2019 4:54 pm

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

Post by no simpler »

MotoTrojan wrote: Tue Jul 16, 2019 10:39 am
no simpler wrote: Tue Jul 16, 2019 9:34 am The secret is a 10% allocation to bitcoin:
https://www.portfoliovisualizer.com/bac ... 0&total3=0

Mostly kidding to make a point about the dangers of backtests. Bitcoin does have a nice risk/reward asymmetry when held in small amounts, and has no correlation with market.
Good point. I am using backtests to inform me of sensitivities but in general am basing the construction on core beliefs, namely those mentioned in the OP along with a belief that momentum in volatility is stronger than in expected return (hence the frequent rebalancing based on the past months volatility). Backtesting is still a valuable way to get an idea of how different constructions (in my case really, amounts of leverage) would perform in different market conditions, at-least those that have happened to date.
Agree on the rebalancing based on volatility. Unlike price series, volatility clustering/autocorrelation is a well known phenomenon. Not an options guy, but pretty sure you could also use implied volatility from options market, which is highly efficient and takes the collective wisdom of the market w.r.t future volatility. And ultimately future volatility is what we care about (vs historical/statistical volatility).
P90XBC
Posts: 3
Joined: Sat Jun 15, 2019 5:10 pm

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

Post by P90XBC »

I have my Vanguard IRA switched over to M1 finance and the transfer was executed right as the s&p crossed 3000. Now that cash is sitting in the M1 roth IRA and ready to go, I'm wondering if i should be a bit cautious on the buy in for the leveraged fund. I'm not one to try and time things but buying in "high" and getting crushed on a 3x fund right from the start seems counter intuitive. Should I wait until at least a small dip before starting this up? Am I looking at this wrong? The funds i'm playing with here are very minimal roughly 10k. Thanks.
Locked