HEDGEFUNDIE's excellent adventure Part II: The next journey

[willthrill81]

Is this strategy still worth doing if the leveraged long bonds have so little upside left given ultra low interest rates?

The TLDR answer to that question is yes because the strategy does not depend on good yields from long-term Treasuries, only the relationship between stocks and LTT. Also, due to bond convexity, long-term bonds have more relative capital appreciation potential when interest rates are low rather than high.

[mistermojorizin]

long-term bonds have more relative capital appreciation potential when interest rates are low rather than high.

I thought that they have more capital appreciation when interest rates decrease, not just by virtue of already being low.

[NMBob]

long-term bonds have more relative capital appreciation potential when interest rates are low rather than high.

I thought that they have more capital appreciation when interest rates decrease, not just by virtue of already being low.

https://www.google.com/search?sitesearc ... +convexity

[am]

Is this strategy still worth doing if the leveraged long bonds have so little upside left given ultra low interest rates?

i suggest you look through the two hef threads. there must be at least 1500-3500 previous posts on that issue, not all in agreement.

Can you summarize? Seems to me like TMF gets killed if rates go up and not much room left going down assuming rates don’t go negative.

[perfectuncertainty]

Deleted

[willthrill81]

I thought that they have more capital appreciation when interest rates decrease, not just by virtue of already being low.

The graph below from Tyler9000, one of the posters here and the creator of the excellent site PortfolioCharts.com, depicts the relationship between bond maturity, total returns, and interest rate changes. You can see that the single year return is much more strongly affected by interest rate declines when interest rates are low rather than high. If anything, interest rates being low right now means that the HF strategy may do better than before.


https://portfoliocharts.com/2019/05/27/ ... convexity/

[kjm]

Strategies involving upro produce great results and all but higher standard deviations. The spy/tmf strategy seems to produce better results and lower standard deviation compared to spy alone.

If you're interested in lowering standard deviation, you could rebalance more often (weekly) and scale the leverage based on the trailing volatility. So, you wouldn't be trading a fixed 2x (or whatever) leverage. You'd scale between 1x and 3x, for example.

[mistermojorizin]

Strategies involving upro produce great results and all but higher standard deviations. The spy/tmf strategy seems to produce better results and lower standard deviation compared to spy alone.

If you're interested in lowering standard deviation, you could rebalance more often (weekly) and scale the leverage based on the trailing volatility. So, you wouldn't be trading a fixed 2x (or whatever) leverage. You'd scale between 1x and 3x, for example.

[Imgur](https://i.imgur.com/OyPjEMI.png)

The UPRO gains are impressive, not gonna lie, but the spy/tmf is so much smoother, higher sharpe ratio, lower standard dev than the SPY alone and a nice little bump in value.

[am]

I thought that they have more capital appreciation when interest rates decrease, not just by virtue of already being low.

The graph below from Tyler9000, one of the posters here and the creator of the excellent site PortfolioCharts.com, depicts the relationship between bond maturity, total returns, and interest rate changes. You can see that the single year return is much more strongly affected by interest rate declines when interest rates are low rather than high. If anything, interest rates being low right now means that the HF strategy may do better than before.


https://portfoliocharts.com/2019/05/27/ ... convexity/

Thank you for posting. Learned a lot from that article. Very surprising results. Will stay the course with the strategy. Hopefully rates won’t go up any time soon.

[Semantics]

I thought that they have more capital appreciation when interest rates decrease, not just by virtue of already being low.

The graph below from Tyler9000, one of the posters here and the creator of the excellent site PortfolioCharts.com, depicts the relationship between bond maturity, total returns, and interest rate changes. You can see that the single year return is much more strongly affected by interest rate declines when interest rates are low rather than high. If anything, interest rates being low right now means that the HF strategy may do better than before.


https://portfoliocharts.com/2019/05/27/ ... convexity/

I find this article and chart incredibly misleading. It reverses cause and effect -- long term treasury prices going up cause bond yields to go down, not the other way around. There's no magic knob that adjusts the yield on a 30 year bond. The knob the Fed has is for the overnight rate, and yes, that could go negative, and that could drive up the prices of treasuries, but the market isn't likely to price in a drop in the overnight rate as being permanent. When the Fed lowers rates by 1%, they don't commit to keeping rates 1% lower than they otherwise would have been for 30 years, which is what that chart assumes. You can see as proof of this that in March, the overnight rate dropped -1.5%, but the 10-year yield only dropped by about -0.9% and the 30-year yield by about -0.6%. And that's with the Fed buying treasuries as part of their quantitative easing efforts, so in a truly free market the dip in yields would have been less.

Furthermore, because of this convexity effect, the Fed won't decrease the overnight rate as much now as they would have previously when it was higher, because they don't need to. Consider that in both 2000 and 2008 the Fed dropped the overnight rate by -5.0%. In March the change was -1.5%. It would take a pretty dire deflationary scenario for them to drop rates below 0%.

If anything, interest rates being low right now means that the HF strategy may do better than before.

This seems like the opposite of what I would expect. The price of long term treasuries depends entirely on how much investors are willing to pay for treasuries, and how much of them the Fed is willing to buy. Bond prices are so high right now that if there's another market crash, I'd expect fewer investors to flee to the treasuries than in the past, preferring gold or cash instead. So the HF strategy will likely do worse than before, unless the Fed shifts their QE program into a mode even beyond what they've been doing so far to make up the difference.

[Thereum]

I agree that overall, low rates are bad for the bond component of this strategy. I think long term treasury rates might grind upward for a few years, which is going to really hurt TMF. There is also no guarantee that investors will rush into bonds during a crash when yields are this low. So instead of buying TMF, I bought EDV using margin (financed with a box spread) and have long term put spreads in TMF as a hedge. If EDV falls at all, TMF is going to fall a lot more. I really have trouble envisioning a scenario in which TMF is trading higher a few years from now.

Then again, my predictions are usually wrong, so perhaps that should provide some comfort to people holding TMF

[coingaroo]

Low rates are double bad for TMF because its ER, as well as its futures financing premiums are fixed and do not go down as rates go down.

[LadyGeek]

I removed a post asking about the impact of a proposed stimulus bill on a few funds. This is "proposed legislation" and is off-topic in this forum. From: Political comments and proposed tax plan remain off-topic

Speculation about future legislation is prohibited by forum policy, see Unacceptable Topics:

Politics and Religion

In order to avoid the inevitable frictions that arise from these topics, political or religious posts and comments are prohibited. The only exceptions to this rule are:

• Common religious expressions such as sending your prayers to an ailing member.
• Usage of factual and non-derogatory political labels when necessary to the discussion at hand.
• Discussions about enacted laws or regulations that affect the individual investor. Note that discussions of proposed legislation are prohibited.
• Proposed regulations that are directly related to investing may be discussed if and when they are published for public comments.

This forum is focused on investing that is directly actionable to personal investors. We don't hold debates on conjecture.

The whole point of the policy is to (1) eliminate contentious disagreements that result from these discussions and (2) keep investors from making bad decisions. Proposed legislation changes many times between the time it's introduced and signed into law.

[willthrill81]

I thought that they have more capital appreciation when interest rates decrease, not just by virtue of already being low.

The graph below from Tyler9000, one of the posters here and the creator of the excellent site PortfolioCharts.com, depicts the relationship between bond maturity, total returns, and interest rate changes. You can see that the single year return is much more strongly affected by interest rate declines when interest rates are low rather than high. If anything, interest rates being low right now means that the HF strategy may do better than before.


https://portfoliocharts.com/2019/05/27/ ... convexity/

I find this article and chart incredibly misleading. It reverses cause and effect -- long term treasury prices going up cause bond yields to go down, not the other way around.

Prices are a function of yields, not vice versa.

The chart doesn't say anything about the Fed's actions, only the impact on single year returns as yields change.

[petulant]

Here's another question though. The graph indicates that the impact on bond prices from a 1% change in prevailing yields becomes much larger as yields go lower. But is a 1% change in yields just as likely in an environment where previous yields were 1% as they are in an environment where yields are 8%? That seems unlikely to me but I'm open to why it would be.

[Semantics]

I thought that they have more capital appreciation when interest rates decrease, not just by virtue of already being low.

The graph below from Tyler9000, one of the posters here and the creator of the excellent site PortfolioCharts.com, depicts the relationship between bond maturity, total returns, and interest rate changes. You can see that the single year return is much more strongly affected by interest rate declines when interest rates are low rather than high. If anything, interest rates being low right now means that the HF strategy may do better than before.


https://portfoliocharts.com/2019/05/27/ ... convexity/

I find this article and chart incredibly misleading. It reverses cause and effect -- long term treasury prices going up cause bond yields to go down, not the other way around.

Prices are a function of yields, not vice versa.

The chart doesn't say anything about the Fed's actions, only the impact on single year returns as yields change.

Assuming you mean effective yield, which is the only thing that makes sense in this conversation, that's like saying the price of a stock is a function of its P/E ratio.

The prices of bonds fluctuate constantly based on supply and demand in the market, the yield is simply the coupon payments divided by the market value. You've got it backwards.

[willthrill81]

I thought that they have more capital appreciation when interest rates decrease, not just by virtue of already being low.

The graph below from Tyler9000, one of the posters here and the creator of the excellent site PortfolioCharts.com, depicts the relationship between bond maturity, total returns, and interest rate changes. You can see that the single year return is much more strongly affected by interest rate declines when interest rates are low rather than high. If anything, interest rates being low right now means that the HF strategy may do better than before.


https://portfoliocharts.com/2019/05/27/ ... convexity/

I find this article and chart incredibly misleading. It reverses cause and effect -- long term treasury prices going up cause bond yields to go down, not the other way around.

Prices are a function of yields, not vice versa.

The chart doesn't say anything about the Fed's actions, only the impact on single year returns as yields change.

Assuming you mean effective yield, which is the only thing that makes sense in this conversation, that's like saying the price of a stock is a function of its P/E ratio.

The prices of bonds fluctuate constantly based on supply and demand in the market, the yield is simply the coupon payments divided by the market value. You've got it backwards.

I agree that prices fluctuate. And I agree that yield is coupon payments.

The only reason people are willing to pay for bonds is for their yield. Yields drive prices, not vice versa. When yields on new bonds go down, prices on old bonds paying higher yields go up to compensate.

If you disagree, start another thread. You'll quickly find that very few here will agree with you.

[willthrill81]

Here's another question though. The graph indicates that the impact on bond prices from a 1% change in prevailing yields becomes much larger as yields go lower. But is a 1% change in yields just as likely in an environment where previous yields were 1% as they are in an environment where yields are 8%? That seems unlikely to me but I'm open to why it would be.

The chart says nothing about the likelihood of a 1% change. You're right that such an event is more likely to occur when starting yields are higher rather than lower, but the chart is merely describing a mathematical relationship, not a probabilistic one.

[Thereum]

Is this strategy still worth doing if the leveraged long bonds have so little upside left given ultra low interest rates?

The TLDR answer to that question is yes because the strategy does not depend on good yields from long-term Treasuries, only the relationship between stocks and LTT. Also, due to bond convexity, long-term bonds have more relative capital appreciation potential when interest rates are low rather than high.

How do you know the inverse relationship between stocks and treasuries will continue to hold with interest rates so low? Perhaps investors will sell treasuries during the next crash and hold onto cash instead.

Second, if you don't seem especially concerned about overall capital losses in the leveraged long term treasuries, why not hold some volatility products instead? You would need to allocate a much smaller portion of the portfolio to this, since volatility products have a higher negative correlation with stocks and a greater degree of volatility.

TMF has essentially served as a put option with a 10% CAGR over the last decade. That is a free lunch. I am not so sure it's going to continue going forward.

[willthrill81]

Is this strategy still worth doing if the leveraged long bonds have so little upside left given ultra low interest rates?

The TLDR answer to that question is yes because the strategy does not depend on good yields from long-term Treasuries, only the relationship between stocks and LTT. Also, due to bond convexity, long-term bonds have more relative capital appreciation potential when interest rates are low rather than high.

How do you know the inverse relationship between stocks and treasuries will continue to hold with interest rates so low? Perhaps investors will sell treasuries during the next crash and hold onto cash instead.

Second, if you don't seem especially concerned about overall capital losses in the leveraged long term treasuries, why not hold some volatility products instead? You would need to allocate a much smaller portion of the portfolio to this, since volatility products have a higher negative correlation with stocks and a greater degree of volatility.

TMF has essentially served as a put option with a 10% CAGR over the last decade. That is a free lunch. I am not so sure it's going to continue going forward.

I don't know that the historic relationship will hold up. It very well might not. This strategy and its kin are not for the faint of heart nor for funds that cannot afford to be lost.

[privatefarmer]

I’ve made detailed posts several thousand pages back on this thread as to why LTTs can continue to perform as they have historically regardless of starting yield. But what I will just point out for now is this : VUSTX had a real CAGR of 5.4% from 1987-2003. It had a 5% real CAGR from 2004-present. The starting yield in 1987 was significantly higher than the starting yield in 2004. If you look at the growth of VUSTX since inception it has been remarkably consistent. For those of you arguing that the returns of bonds (or more specifically bond FUNDS) should decline with declining yields please look at the actual data.

Starting. Yields. Do. Not. Matter.

The starting yield is already “priced into” not only LTTs but corporate bonds, junk bonds, and even stocks. It’s already be accounted for, trying to pick an asset class bc of starting yield is a waste of time and energy bc EVERYTHING is “over priced” with regards to yield. Treasury funds make money by capital appreciation, rolling down the yield curve, gaining as yields continue to decline globally. Why do yields continue to decline? Because it’s like the gas in our global economic engine. Our global economy needs cheap money and it’s like cocaine, our economy becomes accustomed to current yields and needs even lower yields to boost/stimulate growth. Negative yields already exist across the globe and it’s naive to think the USA can prevent them here when the rest of the world goes deeper negative. Money will continue to flood the US treasury market as, when compared to other government debt, the USA actually has a positive yield so is extremely attractive, but this then drives down the USA yields...

Anyhow don’t want to rant I’ve gone through this I don’t know how many times. This is a basic fundamental part of understanding risk parity. I would recommend the book by Dr Qian on risk parity he goes into great detail on why starting yields are irrelevant.

[langlands]

I’ve made detailed posts several thousand pages back on this thread as to why LTTs can continue to perform as they have historically regardless of starting yield. But what I will just point out for now is this : VUSTX had a real CAGR of 5.4% from 1987-2003. It had a 5% real CAGR from 2004-present. The starting yield in 1987 was significantly higher than the starting yield in 2004. If you look at the growth of VUSTX since inception it has been remarkably consistent. For those of you arguing that the returns of bonds (or more specifically bond FUNDS) should decline with declining yields please look at the actual data.

Starting. Yields. Do. Not. Matter.

The starting yield is already “priced into” not only LTTs but corporate bonds, junk bonds, and even stocks. It’s already be accounted for, trying to pick an asset class bc of starting yield is a waste of time and energy bc EVERYTHING is “over priced” with regards to yield. Treasury funds make money by capital appreciation, rolling down the yield curve, gaining as yields continue to decline globally. Why do yields continue to decline? Because it’s like the gas in our global economic engine. Our global economy needs cheap money and it’s like cocaine, our economy becomes accustomed to current yields and needs even lower yields to boost/stimulate growth. Negative yields already exist across the globe and it’s naive to think the USA can prevent them here when the rest of the world goes deeper negative. Money will continue to flood the US treasury market as, when compared to other government debt, the USA actually has a positive yield so is extremely attractive, but this then drives down the USA yields...

Anyhow don’t want to rant I’ve gone through this I don’t know how many times. This is a basic fundamental part of understanding risk parity. I would recommend the book by Dr Qian on risk parity he goes into great detail on why starting yields are irrelevant.

Are you saying that the expected return of a bond has nothing to do with the bond's yield?

[rascott]

I’ve made detailed posts several thousand pages back on this thread as to why LTTs can continue to perform as they have historically regardless of starting yield. But what I will just point out for now is this : VUSTX had a real CAGR of 5.4% from 1987-2003. It had a 5% real CAGR from 2004-present. The starting yield in 1987 was significantly higher than the starting yield in 2004. If you look at the growth of VUSTX since inception it has been remarkably consistent. For those of you arguing that the returns of bonds (or more specifically bond FUNDS) should decline with declining yields please look at the actual data.

Starting. Yields. Do. Not. Matter.

The starting yield is already “priced into” not only LTTs but corporate bonds, junk bonds, and even stocks. It’s already be accounted for, trying to pick an asset class bc of starting yield is a waste of time and energy bc EVERYTHING is “over priced” with regards to yield. Treasury funds make money by capital appreciation, rolling down the yield curve, gaining as yields continue to decline globally. Why do yields continue to decline? Because it’s like the gas in our global economic engine. Our global economy needs cheap money and it’s like cocaine, our economy becomes accustomed to current yields and needs even lower yields to boost/stimulate growth. Negative yields already exist across the globe and it’s naive to think the USA can prevent them here when the rest of the world goes deeper negative. Money will continue to flood the US treasury market as, when compared to other government debt, the USA actually has a positive yield so is extremely attractive, but this then drives down the USA yields...

Anyhow don’t want to rant I’ve gone through this I don’t know how many times. This is a basic fundamental part of understanding risk parity. I would recommend the book by Dr Qian on risk parity he goes into great detail on why starting yields are irrelevant.

Haven't commented on this thread for a few months..... but it's turning into a carousel discussion.

Readers/ posters would do well to actually read everything that's been discussed here over the 2 threads across 18+ months..... where basically every single question/ debate has been hashed out.

[Semantics]

The graph below from Tyler9000, one of the posters here and the creator of the excellent site PortfolioCharts.com, depicts the relationship between bond maturity, total returns, and interest rate changes. You can see that the single year return is much more strongly affected by interest rate declines when interest rates are low rather than high. If anything, interest rates being low right now means that the HF strategy may do better than before.


https://portfoliocharts.com/2019/05/27/ ... convexity/

I find this article and chart incredibly misleading. It reverses cause and effect -- long term treasury prices going up cause bond yields to go down, not the other way around.

Prices are a function of yields, not vice versa.

The chart doesn't say anything about the Fed's actions, only the impact on single year returns as yields change.

Assuming you mean effective yield, which is the only thing that makes sense in this conversation, that's like saying the price of a stock is a function of its P/E ratio.

The prices of bonds fluctuate constantly based on supply and demand in the market, the yield is simply the coupon payments divided by the market value. You've got it backwards.

I agree that prices fluctuate. And I agree that yield is coupon payments.

The only reason people are willing to pay for bonds is for their yield. Yields drive prices, not vice versa. When yields on new bonds go down, prices on old bonds paying higher yields go up to compensate.

If you disagree, start another thread. You'll quickly find that very few here will agree with you.

I don't know who else is disagreeing with me. To add to what privatefarmer posted about face value on new bonds not mattering, there's a simple thought experiment to see why this is the case: a new 10-year bond is issued with a 50% yield. What do you think is more likely to happen, a) prices on the millions of bonds in the market all crater to match that yield, or b) the single new bond becomes valued at 74.6x the original current price to match the 0.67% yield?

How about instead of starting another thread I just link some reading material: https://www.investopedia.com/terms/t/treasury-yield.asp

"The rate of return or yield required by investors for loaning their money to the government is determined by supply and demand."

[Alchemist]

Our global economy needs cheap money and it’s like cocaine, our economy becomes accustomed to current yields and needs even lower yields to boost/stimulate growth.

Perhaps we can coin this the privatefarmer "Cocaine theory of yield curves"



The discussion seems to be...

Someone points out that a 20% drop in equites and a simultaneous 1-2% rise in LTT yields blows up the adventure....adventure advocates say they don't think yields will go up for XYZ reasons.

Did I miss something?

[Semantics]

I’ve made detailed posts several thousand pages back on this thread as to why LTTs can continue to perform as they have historically regardless of starting yield. But what I will just point out for now is this : VUSTX had a real CAGR of 5.4% from 1987-2003. It had a 5% real CAGR from 2004-present. The starting yield in 1987 was significantly higher than the starting yield in 2004. If you look at the growth of VUSTX since inception it has been remarkably consistent. For those of you arguing that the returns of bonds (or more specifically bond FUNDS) should decline with declining yields please look at the actual data.

The 30-year yield was 7.3% in 1987, 5% in 2004, and is 1.4% at present. CAGR in the latter period was similar despite a lower starting yield because yields dropped by 3.6%, resulting in a much larger capital appreciation than the 2.3% drop from 1987-2003, and this offset the lower yield.

In order for VUSTX to achieve enough capital appreciation achieve a CAGR of 5% from 2020-2036 with the current near-zero yields, the 30-year yield would need to drop to around -3.5% if I did my math right. I suppose that could happen if we experience severe deflation, but do you think it's likely? (I suspect that whatever black swan would precipitate that would cause my HFEA performance to be the least of my worries.)

[CanaBogle24]

Plugging a separate thread, just created, to track / share progress of those who are executing HEA or some variant thereof:

viewtopic.php?p=5521765#p5521765

[coingaroo]

Are you saying that the expected return of a bond has nothing to do with the bond's yield?

An argument you can make is that starting yields also determine the equity risk premium, and so low bond yields don't necessarily mean it is less attractive than stocks: the market pricing mechanism should ensure the expected payoffs for treasury bonds, stocks, etc to be the same per unit of risk (and factor exposure; etc).

It is an argument I will concede for unleveraged, or cheaply leveraged bond funds; such as EDV or gaining leverage through futures markets. It is also an argument that I'd concede for long term TIPS bonds (see: Bridgewater's article on why inflation-linked bonds have no lower floor).

However, no one seems to be considering the fact that TMF has a real ER of about 4.4% p.a.; and then plus volatility drag. Calculations here: viewtopic.php?p=5511428#p5511428

This is a serious cost especially at the low interest rate environment of today.

[bling]

what are the downsides of this strategy getting too popular? this thread is referenced outside of BHs frequently, and any mention of UPRO is quickly followed with a mention of TMF. the backtests are impressive and with the pandemic, retail investors have nothing better to do than to gamble it all away by buying leveraged ETFs which are super easy to buy.

is it possible that the popularity of leveraged ETFs increase the probability that a loss of 34% actually happens in a single day and wipes everyone out?

[Mickelous]

Are you saying that the expected return of a bond has nothing to do with the bond's yield?

An argument you can make is that starting yields also determine the equity risk premium, and so low bond yields don't necessarily mean it is less attractive than stocks: the market pricing mechanism should ensure the expected payoffs for treasury bonds, stocks, etc to be the same per unit of risk (and factor exposure; etc).

It is an argument I will concede for unleveraged, or cheaply leveraged bond funds; such as EDV or gaining leverage through futures markets. It is also an argument that I'd concede for long term TIPS bonds (see: Bridgewater's article on why inflation-linked bonds have no lower floor).

However, no one seems to be considering the fact that TMF has a real ER of about 4.4% p.a.; and then plus volatility drag. Calculations here: viewtopic.php?p=5511428#p5511428

This is a serious cost especially at the low interest rate environment of today.

That last bit is what ultimately convinced me to go to EDV, along with having a 43/57 portfolio gives you total leverage of 1.86 instead of 3x. I feel like its much lower risk without compromising returns too badly. I am using TQQQ instead of UPRO to counteract the loss of gain potential with TMF by going slightly riskier on stocks and safer on the bond side. I feel like this gives me a well risk adjusted portfolio. I do not mind the weighting in the tech sector, as I wouldn't invest too much in cyclical and value on my own in the energy and financial sectors. The more and more I ponder my investments the more I lean towards liking what I currently see in the NASDAQ.

[Tingting1013]

what are the downsides of this strategy getting too popular?

I would think there are only upsides.

More assets flowing into LETFs means:

1. Lower expense ratios
2. Lower borrowing spreads for underlying swaps
3. Better liquidity for rebalancing

[Mickelous]

what are the downsides of this strategy getting too popular?

I would think there are only upsides.

More assets flowing into LETFs means:

1. Lower expense ratios
2. Lower borrowing spreads for underlying swaps
3. Better liquidity for rebalancing

Potentially creating more volatility in the market if all index investors suddenly leveraged their portfolio this way it could cause more gains but more volatility.

[occambogle]

what are the downsides of this strategy getting too popular?

UPRO AUM is 1.72 Billion, TQQQ AUM is 8.5 Billion, TMF AUM is 300 million. I find it hard to believe that followers of the excellent adventure are going to significantly affect anything.....

[privatefarmer]

Yeah, “risk parity” investors already make up a significant chunk of hedge fund investments. But they use options etc to obtain leverage, not LETFs. We are small fish but am hopeful soon to be fat fish

[firebirdparts]

what are the downsides of this strategy getting too popular?

Here's what I think: To get daily updated triple leverage, you prefer to have contracts with people who will pay you the return of the S&P 500 daily, and that is mostly what the fund holds. I personally would not want to get involved in this as a counterparty, so I think there is a risk of running out of willing counterparties at a price you're willing to pay. So I think the size of UPRO could have an effect on its actual structure. It appears it's nowhere near that now, though.

[Tingting1013]

what are the downsides of this strategy getting too popular?

Here's what I think: To get daily updated triple leverage, you prefer to have contracts with people who will pay you the return of the S&P 500 daily, and that is mostly what the fund holds. I personally would not want to get involved in this as a counterparty, so I think there is a risk of running out of willing counterparties at a price you're willing to pay. So I think the size of UPRO could have an effect on its actual structure. It appears it's nowhere near that now, though.

You would not personally want to be a counterparty but you also do not have the risk management needs of a mega bank.

If you are a mega bank with large holdings of the S&P 500, you might be eager to hedge some of that risk by entering into a swap agreement to redirect the S&P’s returns to a LETF (and get paid an attractive spread in the process).

[Dr. Long]

Rebal day! 60/40 OG TMF/UPRO blend quarterly rebals:

07/19 5k
10/19 5.4k
01/20 5.6k
04/20 6.6k
07/20 8.5k
10/20 9.3k

Chart: https://ibb.co/H4fqDTf

[RovenSkyfall]

It seems like minimum variance with 21 day look back is nearly the same as inverse volatility for this portfolio (both in returns, SD, max drawdown, rolling returns, sharpe). Any reason to use one over the other?

It appears using AAA with MV on UPRO/TMF is missing part of the whole process novel to MV AAA, given that there is no momentum aspect with just two funds. This leaves one to just comparing IV with the weighted covariance matrix of MV. Are these just very similar mathematically, and if so, why choose one over the other? IV seems much simpler to calculate than the covariance matrix of MV.

Trying to understand the differences better and I appreciate any information on their differences/strengths/weaknesses.

[axezine]

I agree. Without the diversity you'd have in the AAA porfolio as described, you lose the momentum side of that strategy. However, MV is still better than IV (at least from an information POV).

From my understanding, the whole point of IV as a strategy is that you use it when you *don't* have correlation information, or you believe such information to be wrong / misleading. In the case of UPRO/TMF, this whole strategy is pinned on the idea that TMF and UPRO are anti-correlated to some extent; BUT, that correlation mostly has to do with their price, and NOT their volatility, meaning we have reason to believe TMF goes up if UPRO goes down, but there's much less evidence that TMF is much less *volatile* when UPRO is more volatile and vice-versa. That being said, volatility for TMF is generally lower than for UPRO, and it seems historically that periods of high volatility correspond with UPRO going down much more than TMF.

So, if you can't be bothered to calculate the covariance, then IV is suitable, but it certainly leaves room for lowering volatility if you do go to the trouble of calculating it.

I've run some backtests of my own back to Jan 2010 with a number of strategies, mostly around a UPRO/TQQQ/TMF portfolio. So far, the best back test for me has been a risk-parity approach between the three, rebalanced monthly, using 15 day volatility. Meaning, calculating the covariance matrix and then computing allocations such that each of the three funds contributes the same amount to the overall volatility. From Jan 2010 to yesterday (09/30/20) this returned 41.05% CAGR with 25.83% anualized volatility, for a Sharpe ratio of 1.47 (assuming a 3.5% risk-free rate, which at this point is probably generous, but seems correct for the time period). Using 21-day volatility knocks back the return a bit, but volatility remains about the same.

Jorge

[kjm]

I agree. Without the diversity you'd have in the AAA porfolio as described, you lose the momentum side of that strategy. However, MV is still better than IV (at least from an information POV).

From my understanding, the whole point of IV as a strategy is that you use it when you *don't* have correlation information, or you believe such information to be wrong / misleading. In the case of UPRO/TMF, this whole strategy is pinned on the idea that TMF and UPRO are anti-correlated to some extent; BUT, that correlation mostly has to do with their price, and NOT their volatility, meaning we have reason to believe TMF goes up if UPRO goes down, but there's much less evidence that TMF is much less *volatile* when UPRO is more volatile and vice-versa. That being said, volatility for TMF is generally lower than for UPRO, and it seems historically that periods of high volatility correspond with UPRO going down much more than TMF.

So, if you can't be bothered to calculate the covariance, then IV is suitable, but it certainly leaves room for lowering volatility if you do go to the trouble of calculating it.

I've run some backtests of my own back to Jan 2010 with a number of strategies, mostly around a UPRO/TQQQ/TMF portfolio. So far, the best back test for me has been a risk-parity approach between the three, rebalanced monthly, using 15 day volatility. Meaning, calculating the covariance matrix and then computing allocations such that each of the three funds contributes the same amount to the overall volatility. From Jan 2010 to yesterday (09/30/20) this returned 41.05% CAGR with 25.83% anualized volatility, for a Sharpe ratio of 1.47 (assuming a 3.5% risk-free rate, which at this point is probably generous, but seems correct for the time period). Using 21-day volatility knocks back the return a bit, but volatility remains about the same.

Jorge

Jorge,

Thanks for sharing your approach. What standard deviation and max drawdown did you see for the overall portfolio for that period? Have you tried incorporating implied volatility into you models? I find adding gold and using implied volatility (via VIX, VXN, VXTLT and GVZ) along with historical volatility (standard deviation) results in very good risk-adjusted returns.

[axezine]

Hi!

I must admit I had not thought about this approach, which sounds quite interesting - I'll have to run some tests using these forward-looking volatility measures...

How are you transforming them to account for the leverage in TQQQ, UPRO and TMF? Do you just multiply by 3 (that's what the maths suggest) or are you doing something different?

And how are you incorporating the measure into your model?

Jorge

[kjm]

Hi!

I must admit I had not thought about this approach, which sounds quite interesting - I'll have to run some tests using these forward-looking volatility measures...

How are you transforming them to account for the leverage in TQQQ, UPRO and TMF? Do you just multiply by 3 (that's what the maths suggest) or are you doing something different?

And how are you incorporating the measure into your model?

Jorge

Yeah, definitely play around with it. I'm interested to get someone else's take. I don't do anything to account for the leverage. Since I'm looking at the volatilities relative to each other (and they're all leveraged), I figure it would be a wash.

What I'm doing is super rudimentary. For example, to calculate UPRO's portfolio weight, i do the following:

UPRO weight = (average standard deviation / trailing 6-month standard deviation) * (average VIX / trailing 15-day VIX)

Repeat for TQQQ, TMF and UGL.

I find dividing trailing standard deviation and VIX into their respective long-term averages gives better results than just inversing them.

There's nothing special about those look-back periods. I'm sure they can be optimized. Or, you could loop through multiple look-back periods and come up with a composite metric.

A couple caveats. There's no 3x gold ETF, so I'm using UGL. Also, VXTLT has only been around since 2015, so that's the furthest I can take the backtest. I guess I could calculate implied volatility using historical options premium, but that's beyond my abilities at the moment.

I find better absolute returns without gold (UGL). Better risk-adjusted returns with gold.

I've also played around with adjusting the total portfolio leverage, but it seems hit or miss.

[coingaroo]

Stop playing with inverse volatility, adaptive allocation, whatever, etc. You will get burned.

I'm saying this as someone who played around with inverse volatility (SPY/IEF, 8%). It looked GREAT in backtests. So I started doing it around the start of 2018. Two years later, my CAGR is half as buy and hold.

What happened to the Bogleheads manta of "no market timing"?

[privatefarmer]

Stop playing with inverse volatility, adaptive allocation, whatever, etc. You will get burned.

I'm saying this as someone who played around with inverse volatility (SPY/IEF, 8%). It looked GREAT in backtests. So I started doing it around the start of 2018. Two years later, my CAGR is half as buy and hold.

What happened to the Bogleheads manta of "no market timing"?

What specifically was your strategy? I use min variance, monthly rebalancing.

[hilink73]

Stop playing with inverse volatility, adaptive allocation, whatever, etc. You will get burned.

I'm saying this as someone who played around with inverse volatility (SPY/IEF, 8%). It looked GREAT in backtests. So I started doing it around the start of 2018. Two years later, my CAGR is half as buy and hold.

What happened to the Bogleheads manta of "no market timing"?

Would you mind discussing what strategy you followed?

I was on a target volatility strategy for UPRO/TMF and had difficult times selling into losses.
Last month, it said 80/20, for Oktober it said 35/65!
I now rebalanced to 55/45 (the original strategy) and probably stay with it (fixed allocation) for good.

[perfectuncertainty]

Stop playing with inverse volatility, adaptive allocation, whatever, etc. You will get burned.

I'm saying this as someone who played around with inverse volatility (SPY/IEF, 8%). It looked GREAT in backtests. So I started doing it around the start of 2018. Two years later, my CAGR is half as buy and hold.

What happened to the Bogleheads manta of "no market timing"?

What period were you looking at SPY and IEF for? If I backtest it using the adaptive allocation model with inverse volatility the CAGR was 6.89% from Jan 2003 - Dec 2017 whereas the SP500 was 9.79%.

[golongterm]

Stop playing with inverse volatility, adaptive allocation, whatever, etc. You will get burned.

I'm saying this as someone who played around with inverse volatility (SPY/IEF, 8%). It looked GREAT in backtests. So I started doing it around the start of 2018. Two years later, my CAGR is half as buy and hold.

What happened to the Bogleheads manta of "no market timing"?

Can you expand on this? What is SPY/IEF, 8%? 92SPY/8IEF? Or are you saying whatever SPY/IEF allocation backtested to 8%/year?

Also, I'm assuming you did a multi-decade backtest. In the case of this topic's allocation, the 55/45 lags SPY for what, like 6-8 years before pulling massively ahead?

[Semantics]

Stop playing with inverse volatility, adaptive allocation, whatever, etc. You will get burned.

I'm saying this as someone who played around with inverse volatility (SPY/IEF, 8%). It looked GREAT in backtests. So I started doing it around the start of 2018. Two years later, my CAGR is half as buy and hold.

I think it's unlikely for these techniques to enhance your CAGR. The point is not to increase CAGR, it's to reduce volatility - and inverse volatility with SPY/IEF since 2018 has trounced 50/50 or 100/0 buy and hold in both volatility and risk-adjusted returns.

I also don't think it's worthwhile to use inverse volatility on an unleveraged portfolio. High volatility is not as detrimental as it is to a portfolio subject to compounded daily leverage resets. You also lose a lot of dividends by being underweight in equities that a leveraged portfolio compensates for with compounding during the low volatility periods.

[langlands]

Stop playing with inverse volatility, adaptive allocation, whatever, etc. You will get burned.

I'm saying this as someone who played around with inverse volatility (SPY/IEF, 8%). It looked GREAT in backtests. So I started doing it around the start of 2018. Two years later, my CAGR is half as buy and hold.

I think it's unlikely for these techniques to enhance your CAGR. The point is not to increase CAGR, it's to reduce volatility - and inverse volatility with SPY/IEF since 2018 has trounced 50/50 or 100/0 buy and hold in both volatility and risk-adjusted returns.

I also don't think it's worthwhile to use inverse volatility on an unleveraged portfolio. High volatility is not as detrimental as it is to a portfolio subject to compounded daily leverage resets. You also lose a lot of dividends by being underweight in equities that a leveraged portfolio compensates for with compounding during the low volatility periods.

I was completely with you until the last sentence. Why are dividends in particular important? Is there anything about dividend paying stocks vs. non-dividend paying that interact differently with a volatility-based asset allocation scheme?

[Thereum]

Very glad I bought puts on TMF! As I said earlier, I see very few ways this ETF won't be down over the next few years, even if rates fall. I suspect it's going to be down 30% or more. I also think the correlation between treasuries and stocks will be weaker than ever, due to bonds being less a safe haven at these rates. But as long as we don't see a stock market crash, those running the Hedgefundie strategy probably won't see drawdowns larger than 40% or so.