HEDGEFUNDIE's excellent adventure Part II: The next journey

[orvanik]

i've started to play around w/ adding the VIX to this strategy and it certainly is appealing. However, the only way to invest in the VIX that I can find is through an ETN like VXX or VIXY, or use the leveraged version TVIX. all these options seem to drastically underperform the VIX index itself. Is there any better way to gain long-term exposure to the VIX? Thanks

A user above posted this paper which shows that most indexes tracking the VIX are a great way to lose money, but tracking the VIX yourself might be worth it if you are willing to go through the hassle. I'm still reading through it, but this is one of my personal interests so I will likely try to undertake the work of building my own VIX if its feasible.

https://papers.ssrn.com/sol3/papers.cfm ... id=3514141

[Hydromod]

I'm about to implement this strategy and wanted to clarify the intended rebalancing — I believe it's quarterly, but are there any other rebalance conditions e.g. 5/25 rebalancing?

I believe quarterly works best in backtests and in live data, but I don't want to take my eye off 5/25.

I don't have access to the portfoliovisualizer sims to backtest rebalancing, so am not 100% sure, but if I test periods in live data where rates were basically flat, 5/25 appears to be a slight winner. Maybe that's just for my allocation though.

https://www.portfoliovisualizer.com/bac ... tion2_1=50

https://www.portfoliovisualizer.com/bac ... tion2_1=50

In any case, PV can only rebalance once a month. Even if you put in a ridiculously small rebalancing band, the most frequently you will see is every month. So I don't think PV is really telling us how 5/25 works. For example, I believe 5/25 would have been triggered more than once in March but PV won't show that.

The backtesting presented in this thread was performed outside of PV. The analyses swept through intervals using many different start days, which clarified that there is quite a bit of luck related to start date relative to a few big events.

In general, shorter intervals tended to give systematically better results, but the effect was small for intervals between weekly and quarterly. Longer intervals between rebalances tended to have more spread in the returns for sequences starting on different days, even though the median returns were similar.

The results suggest that rebalancing effectiveness has depended on the time of month, with beginning/end of month and beginning/end of quarter historically doing significantly better than the middle of the month. This appears to be related to monthly (and perhaps quarterly) cycles for equities and treasuries that were out of phase. PV has historically used end of month/end of quarter, so backtesting comparisons of different intervals are biased by using an auspicious time.

Rebalancing with a 5/25 would have historically had rather frequent trades (averaging approximately 2 times a month but quite clustered). It seemed to me that a 10 or 15 band may have been a sweet spot, combined with the restriction that the maximum interval between rebalances wasn't too long (i.e., no longer than semiannual or annual).

None of these backtests considered very rapid shocks like COVID, so the optimal short-term approach is hazy.

YMMV.

[physixfan]

i've started to play around w/ adding the VIX to this strategy and it certainly is appealing. However, the only way to invest in the VIX that I can find is through an ETN like VXX or VIXY, or use the leveraged version TVIX. all these options seem to drastically underperform the VIX index itself. Is there any better way to gain long-term exposure to the VIX? Thanks

A user above posted this paper which shows that most indexes tracking the VIX are a great way to lose money, but tracking the VIX yourself might be worth it if you are willing to go through the hassle. I'm still reading through it, but this is one of my personal interests so I will likely try to undertake the work of building my own VIX if its feasible.

https://papers.ssrn.com/sol3/papers.cfm ... id=3514141

Can you provide a summary? I can't understand why all VIX ETNs are losing money due to huge contango but one can avoid it by DIY.

[Lee_WSP]

i've started to play around w/ adding the VIX to this strategy and it certainly is appealing. However, the only way to invest in the VIX that I can find is through an ETN like VXX or VIXY, or use the leveraged version TVIX. all these options seem to drastically underperform the VIX index itself. Is there any better way to gain long-term exposure to the VIX? Thanks

A user above posted this paper which shows that most indexes tracking the VIX are a great way to lose money, but tracking the VIX yourself might be worth it if you are willing to go through the hassle. I'm still reading through it, but this is one of my personal interests so I will likely try to undertake the work of building my own VIX if its feasible.

https://papers.ssrn.com/sol3/papers.cfm ... id=3514141

Can you provide a summary? I can't understand why all VIX ETNs are losing money due to huge contango but one can avoid it by DIY.

Since vix has no underlying value, it serves purely as a hedge against bad times and like insurance premiums, you will lose over the long haul.

[langlands]

i've started to play around w/ adding the VIX to this strategy and it certainly is appealing. However, the only way to invest in the VIX that I can find is through an ETN like VXX or VIXY, or use the leveraged version TVIX. all these options seem to drastically underperform the VIX index itself. Is there any better way to gain long-term exposure to the VIX? Thanks

A user above posted this paper which shows that most indexes tracking the VIX are a great way to lose money, but tracking the VIX yourself might be worth it if you are willing to go through the hassle. I'm still reading through it, but this is one of my personal interests so I will likely try to undertake the work of building my own VIX if its feasible.

https://papers.ssrn.com/sol3/papers.cfm ... id=3514141

Can you provide a summary? I can't understand why all VIX ETNs are losing money due to huge contango but one can avoid it by DIY.

Since vix has no underlying value, it serves purely as a hedge against bad times and like insurance premiums, you will lose over the long haul.

Yes, that's true, but doesn't answer the question of why replicating VIX with put and call options would outperform VIX ETNs as claimed by the white paper.

VIX is priced to predict future volatility. Essentially, it represents the implied volatility of SPY OTM call and put options at certain strikes. Because there's a premium for hedging downside risk ever since 1987, implied volatility reliably overpredicts realized volatility and there's a volatility risk premium. I remember reading somewhere that there's an additional risk premium of the futures VIX over spot VIX. I have no idea if it's really true or why that would be the case though.

[Lee_WSP]

People have posted plenty of data suggesting that futures are much better than upro/tmf. Nonetheless, there is value in not having to do it yourself. I imagine it’s a similar reason DIY will net you better returns or rather lower costs.

[TwoIdenticalIndexes]

Speaking of using futures to avoid the ridiculous UPRO ER...

This article (https://articles/5-high-beta-etfs-stock ... 2017-02-16) lists some funds with high betas. Since UPRO's beta is ~3, and the primary point of the strategy is to increase exposure to the market, we should theoretically be able to replicate the strategy with exposure to some other high beta asset.

The betas hold water in PV. I tested SPHB and RZV. (https://www.portfoliovisualizer.com/bac ... ion3_3=100)

And yet, the Excellent Adventure crushes RZV, and SPHB fared no better, though I didn't include it in the link because the backtest doesn't go very far. (https://www.portfoliovisualizer.com/bac ... on5_2=-200)

Potential reasons for weaker performance:
1. RZV doesn't invese TMF sufficiently, even though they perform inversely vs. the market.
2. Some limitation of the CAPM?

[Uncorrelated]

Although you can avoid the high ER of UPRO and futures, you'll miss out on the auto adjusting leverage features. Although volatility decay often gets bad rep, rebalancing futures in monthly intervals isn't without it's faults. Depending on your target leverage it might be worth paying the ER on UPRO.

Full analysis here: viewtopic.php?p=5040888#p5040888

[IndexCore]

This is also why most discussions about long term treasuries end up focusing on the wrong arguments. The decision to invest in long term treasuries is not affected by the risk free rate being 10% of -5%. A bet on long term treasuries is a bet on the persistence of a term premium, which can only be indirectly observed. Frankly there is zero reason why the expected term premium is different from 10, 20 years ago. Pay close attention to the word expected here. There is no way the realized term premium the last 20 years is a good predictor of expected term premium, this is also why backtests involving treasuries are completely useless.

Long term bonds will (over long term) always outperform short term bonds. This is because of their massively higher interest rate and inflation risk. Higher risk = higher reward.
I'm not sure if I agree with the yields. To a lot of people/companies 5% safe vs. 10% equities is a lot better than 0% safe vs. 5% equities.

That's false, but might not be apparent in the past 30 years of data. Besides interest payments, bonds have capital gains when yields fall. A bond yielding 5% is worth more than a bond yielding 4%, which is reflected in it's sale price. But the opposite is also true, that rising yields make prior bonds look worse, and inflict a capital loss. Those capital losses can easily be greater than the bond's yield, in which case longer-term bonds lose more than short-term bonds.

For long-term bond funds, there's an important number called "bond duration" that acts as a multiplier on changes in yield. When interest rates go up... and bond yields follow, then you can see an impact proportional to the bond fund's duration. A short-term bond fund might have a duration of 3 years, while long-term bond funds can be 15-25 years.

[IndexCore]

According to the introduction of that paper, it's about market volatility rather than volatility drag on performance:

Actually, I think it is about performance. Market volatility includes both aspects of performance -- drag and boost. I mean LETFs can do either, under or over perform the multiple of its index. See this for example ... we are talking about price movement.

Re-reading, I suspect we're both right. On page 16, they mention how investors seeking consistent 3X performance could reduce volatility. It sounds like performance for the investor dovetails with reduced volatility for markets.

I believe it's simply an empirical fact that daily rebalancing will eliminate volatility drag. No one would do that though so it's a good thing that it needn't be done daily. Adding $2870 on day three is the same as adding $2k on day. 1, removing $900 on day 2 and adding 1890 on day 3.

Ah, excellent! I really like spreadsheet example of daily rebalancing. If I can round off for the sake of discussion, the maximum drop in that example is -15% with a +30% cash infusion (roughly). I've seen -15% drops in the current market, so I need to take that a bit further. In my calcs and here, I keep seeing a roughly "double the drop" need for cash. (Which pushes a 3X fund back up to where the 1X fund was trading, and then the cash is pulled out later).

I don't mean to suggest that using TMF (3X leveraged treasuries) to rebalance against UPRO (3X S&P500) will eliminate all volatility decay. It would seem though that if they are negatively correlated, that doing so will definitely reduce volatility decay. Thus I think it is pointless to look at volatility decay in each fund separately. It's simply an empirical fact that flows (such as those involved in rebalancing) can mitigate or eliminate decay.

Of course, if both TMF and UPRO were to go sideways such that money from TMF wasn't available to put into UPRO when it was down and money from UPRO wasn't available to put into TMF when it was down, then I think volatility drag could be a threat and there would be real risk of underperforming 3X the index. The only solution I see is to add more money and trust that TMF and UPRO will see better days.

I'm not that interested in TMF, but since you're helpful I want to sort of return the favor by pointing out an area that might be missed if you invest in TMF. My understanding is that 3X bond ETFs do not pay 3X the yield - they only make 3X the price movements. TMF moves inversely to bond yields, which since the late 1980s have headed mostly downwards. If all that is correct, then any back testing using the past 30 years of data is mostly testing the same situation: falling yields giving bonds a capital gain. To me, that's a risk factor with TMF: a rising rate environment hasn't happened in 30+ years, and so might not appear in back testing. In a rising rate environment, TMF should have consistent capital losses.

[typical.investor]

TMF moves inversely to bond yields, which since the late 1980s have headed mostly downwards. If all that is correct, then any back testing using the past 30 years of data is mostly testing the same situation: falling yields giving bonds a capital gain.

OK, I agree. Let's throw falling rates out the window.

To me, that's a risk factor with TMF: a rising rate environment hasn't happened in 30+ years, and so might not appear in back testing. In a rising rate environment, TMF should have consistent capital losses.

My expectation is simply this - rates will go up when the economy does well because the target is price stability and too loose of monetary policy is a threat to that.

And I expect that when things crash, as they will do from time to time, that rates will go down to fuel the economy.

And as such, I believe TMF will be a good counterweight to UPRO.

Are you sure the dividends aren't priced into TMF? With futures they are part of the price even though the dividend isn't received.

I'd like to use futures but don't want to watch margin levels all the time.

[IndexCore]

In trying to calculate it myself, I was a bit surprised how much cash is involved:
S&P 500 drops -10%, UPRO drops -30%, requiring +20% cash added to UPRO until recovery (and then removing the 20% cash)
S&P 500 drops -20%, UPRO drops -60%, requiring +40% cash loaned to UPRO until recovery

This is close enough for idle chitchat, but UPRO starts over every day. It’s not true at all that it drops 30% when the index drops 10%. FWIW. If it did that in one day, then yeah we’d say it tracked very well that one day.

Please go beyond idle chitchat and feel free to criticize where I have it wrong. I'll even start. In my goal of exploring volatility drag, I didn't provide realistic daily changes.

What if markets moved -8%, +5%, -5%, -10% ... like the S&P 500 did on March 9 to 12. Net loss -17%.
https://finance.yahoo.com/quote/VOO/history?p=VOO
https://finance.yahoo.com/quote/UPRO/history?p=UPRO
UPRO moved -23%, +16%, -15%, -29% during March 9 - 12. That's a net loss of -46%.

S&P 500 are real assets that can recover to the prior value. With one +20%, it reverts to the mean (from the above). But UPRO traces a path 3X that of the S&P 500, with no reversion to the mean. It goes up 3x20% = +60%, but still winds up -14% down even after S&P 500 has recovered. I'd like to understand and possibly mitigate that volatility drag.

One approach mentioned earlier was to inject the difference in 1X and 3X losses as a new cash infusion. With the numbers for Mar 9-12, it looks like about 1/3rd of UPRO's value needs to be kept in cash, and made available, to offset volatility drag.

When VOO / UPRO performs -8% and -23%, that requires +16% added cash (of UPRO's original value)
For VOO / UPRO gaining +5% and +16%, that allows withdrawing -11% from UPRO back into cash
Then VOO / UPRO lose -5% and -15%, requiring +10% cash injected to UPRO.
Finally, VOO / UPRO lose -10% and -29% and the cash need reaches a staggering +19% from one day (Mar 12).

It looks like canceling volatility drag in this case requires +36% additional cash. Scaling that from 136% down to 100%, and I get roughly 3/4ths invested and 1/4th waiting in cash. An even deeper worst case scenario might require 1/3rd or even 1/2 in cash to offset even deeper cumulative drops in UPRO over time.

[IndexCore]

To me, that's a risk factor with TMF: a rising rate environment hasn't happened in 30+ years, and so might not appear in back testing. In a rising rate environment, TMF should have consistent capital losses.

My expectation is simply this - rates will go up when the economy does well because the target is price stability and too loose of monetary policy is a threat to that.

And I expect that when things crash, as they will do from time to time, that rates will go down to fuel the economy.

And as such, I believe TMF will be a good counterweight to UPRO.

Are you sure the dividends aren't priced into TMF? With futures they are part of the price even though the dividend isn't received.

I'd like to use futures but don't want to watch margin levels all the time.

I assumed/guessed TMF didn't pay dividends, so they could be priced into derivatives TMF uses, as you mention.

Near a crash, I can see how UPRO/TMF makes sense: stocks drop sharply, causing the Fed to lower interest rates, which translates to lower yields, and TMF goes up. Unfortunately I tried that in March 9-12 and it went very badly... during the actual panic, all assets moved together (excepting short-term treasuries). I correctly predicted a massive drop, and got screwed out almost all of the gain by the correlations of other assets. At the time, I couldn't buy 3x bear S&P 500, since my brokerage didn't allow purchases of leveraged ETFs. I've since shifted some money elsewhere, and bought some shares of UPRO.

[typical.investor]

What if markets moved -8%, +5%, -5%, -10% ... like the S&P 500 did on March 9 to 12. Net loss -17%.

OK but go back and start from Feb 18 (or any point you choose really), and by March 9th (your starting point) TMF was up 62% and UPRO down about 50%.

I'm personally not injecting cash to combat volatility. TMF gains were more than sufficient.

[typical.investor]

Unfortunately I tried that in March 9-12 and it went very badly... during the actual panic, all assets moved together (excepting short-term treasuries). I correctly predicted a massive drop, and got screwed out almost all of the gain by the correlations of other assets. At the time, I couldn't buy 3x bear S&P 500, since my brokerage didn't allow purchases of leveraged ETFs. I've since shifted some money elsewhere, and bought some shares of UPRO.

I rebalanced 3.6 and 3.10. Nailing the 9th would have been best but such is life.

Long Term Treasuries definitely went up in the crisis. There was a period where liquidity hit and markets melted down. Perhaps it wasn't so evident in short term treasuries but municipal bonds had even worse trouble at least until the Fed restored liquidity.

Anyway, I'm up 24% after about 13 months. I have 3X international positions as well so probably others have done better.

[stormcrow]

This thread is so long that the same concerns or questions keep coming up. it’s like wack a mole, someone explains why this strategy has worked and can continue to work but 5-6 pages later someone else asks the same question.

Everyone’s concern is obviously on LTT yields. I am no expert but I did stay at holiday inn last night and my views are this :

- if the dollar continues to be strong and other currencies weaken, foreign investors will be happy to own treasuries even at negative yields bc they’ll be repaid 30 years from now in green backs which may be worth more than they are today, compared to their home currency

There is a lot of good stuff in your post, but I think this a crucial point in allaying fears regarding both US treasuries and US currency. Given how much global debt is dollar-denominated treasuries should like remain a highly favored asset.

[orvanik]

i've started to play around w/ adding the VIX to this strategy and it certainly is appealing. However, the only way to invest in the VIX that I can find is through an ETN like VXX or VIXY, or use the leveraged version TVIX. all these options seem to drastically underperform the VIX index itself. Is there any better way to gain long-term exposure to the VIX? Thanks

A user above posted this paper which shows that most indexes tracking the VIX are a great way to lose money, but tracking the VIX yourself might be worth it if you are willing to go through the hassle. I'm still reading through it, but this is one of my personal interests so I will likely try to undertake the work of building my own VIX if its feasible.

https://papers.ssrn.com/sol3/papers.cfm ... id=3514141

Can you provide a summary? I can't understand why all VIX ETNs are losing money due to huge contango but one can avoid it by DIY.

Since vix has no underlying value, it serves purely as a hedge against bad times and like insurance premiums, you will lose over the long haul.

Yes, that's true, but doesn't answer the question of why replicating VIX with put and call options would outperform VIX ETNs as claimed by the white paper.

VIX is priced to predict future volatility. Essentially, it represents the implied volatility of SPY OTM call and put options at certain strikes. Because there's a premium for hedging downside risk ever since 1987, implied volatility reliably overpredicts realized volatility and there's a volatility risk premium. I remember reading somewhere that there's an additional risk premium of the futures VIX over spot VIX. I have no idea if it's really true or why that would be the case though.

Again, haven't finished the paper, but my initial guess would be the costs associated with the ETF or ETN are even more impactful when you are talking about something like the VIX. People are paying effectively a large amount to gain exposure and because of the hassle of DIY they can afford to charge it. Also from what it looks like currently the VXX which I believe is the largest has only 600mm in AUM. I'm sure if they are going to run something like that they want to make their fair share off it.

[manlymatt83]

Is anyone else doing UPRO/TMF in taxable instead of tax advantaged? How are you rebalancing?

Due to my situation (long story) I only have one tax-advantaged account -- a traditional IRA.

Right now, I've been doing this in taxable, but just rebalanced and sold and took a profit on UPRO to rebalance into TMF and obviously will have to pay taxes on that. I don't intend to add much more to this strategy long term -- started with a specific dollar amount and happy to see that dollar amount grow over time, but don't intend on adding much more.

Will the long term benefits be negated by my need to pay taxes on gains every time I rebalance? If so, I can sell some bonds in my tax advantaged, move some of those to my taxable, and replace that amount with my fixed dollar amount of UPRO/TMF.

I guess I can also just put some in taxable and some in traditional IRA and just use the IRA allocation to rebalance, but I had hoped to keep this all in one place since my gamble amount is small.

[TwoIdenticalIndexes]

Don't do this in taxable.

[kmft]

Don't do this in taxable.

It's actually not as bad as you'd think.
viewtopic.php?f=10&t=288192&p=5192804#p5192804

[Lee_WSP]

Is anyone else doing UPRO/TMF in taxable instead of tax advantaged? How are you rebalancing?

Due to my situation (long story) I only have one tax-advantaged account -- a traditional IRA.

Right now, I've been doing this in taxable, but just rebalanced and sold and took a profit on UPRO to rebalance into TMF and obviously will have to pay taxes on that. I don't intend to add much more to this strategy long term -- started with a specific dollar amount and happy to see that dollar amount grow over time, but don't intend on adding much more.

Will the long term benefits be negated by my need to pay taxes on gains every time I rebalance? If so, I can sell some bonds in my tax advantaged, move some of those to my taxable, and replace that amount with my fixed dollar amount of UPRO/TMF.

I guess I can also just put some in taxable and some in traditional IRA and just use the IRA allocation to rebalance, but I had hoped to keep this all in one place since my gamble amount is small.

Well, if the UPRO portion had extreme losses, you could TLH and switch to TQQQ or the Direxion equivalent to offset the STCG of the TMF sale.

Otherwise, the tax drag isn't that bad and more or less like paying taxes on your wages if the strategy does well. Ie, it is what it is.

[fidream]

Don't do this in taxable.

It's actually not as bad as you'd think.
viewtopic.php?f=10&t=288192&p=5192804#p5192804

He mentions spec-id is best, but how to choose which ones to sell and which ones not in order to minimize taxes?

[occambogle]

Don't do this in taxable.

It's actually not as bad as you'd think.
viewtopic.php?f=10&t=288192&p=5192804#p5192804

That post is interesting. LIke @mjuszczak I also have minimal tax-advantaged space, so considering a bit of an adventure in taxable, although likely a milder form.

Here is the average tax drag percentage over the 10 years for each lot method:
Monthly Re-balancing of $100k lump sum, no added cash, from 1/1/2010 to 1/30/2010:

Am I interpreting the meaning of the dates wrong? 1/1/2010 to 1/30/2010 seems like 1 month not 10 years... or what does it mean?
It would have also been interesting to see equivalent data for 3-month (quarterly) rebalancing.

• FIFO: 3.84%
• LIFO: 1.58%
• Tax Efficient Loss Harvester: 1.73%
• Nearest Cost: 1.64%
• Highest Cost: 1.51%
• Lowest Taxes (Manual Spec-ID): 1.31% - Winner for this data set

I'm at Interactive Brokers... does anyone know how Dovahkiin's tax-lot methods match to IBKR's, and according to his data... which would be the optimum?

• First In, First Out (FIFO)
• Last In, First Out (LIFO)
• Maximize Long-Term Gain
• Maximize Long-Term Loss
• Maximize Short-Term Gain
• Maximize Short-Term Loss
• Highest Cost
• Specific Lot

https://www.interactivebrokers.com/en/s ... ethods.htm

[kmft]

Don't do this in taxable.

It's actually not as bad as you'd think.
viewtopic.php?f=10&t=288192&p=5192804#p5192804

He mentions spec-id is best, but how to choose which ones to sell and which ones not in order to minimize taxes?

I would use a tax efficient brokerage options, such as "Tax Sensitive: Short Term" at Fidelity or possibly "Maximize Short-Term Loss" at IBKR. I'm sure some other brokerages have similar asset disposal methods, but I'm not versed in all of them.

[kmft]

Don't do this in taxable.

It's actually not as bad as you'd think.
viewtopic.php?f=10&t=288192&p=5192804#p5192804

That post is interesting. LIke @mjuszczak I also have minimal tax-advantaged space, so considering a bit of an adventure in taxable, although likely a milder form.

Here is the average tax drag percentage over the 10 years for each lot method:
Monthly Re-balancing of $100k lump sum, no added cash, from 1/1/2010 to 1/30/2010:

Am I interpreting the meaning of the dates wrong? 1/1/2010 to 1/30/2010 seems like 1 month not 10 years... or what does it mean?
It would have also been interesting to see equivalent data for 3-month (quarterly) rebalancing.

• FIFO: 3.84%
• LIFO: 1.58%
• Tax Efficient Loss Harvester: 1.73%
• Nearest Cost: 1.64%
• Highest Cost: 1.51%
• Lowest Taxes (Manual Spec-ID): 1.31% - Winner for this data set

I'm at Interactive Brokers... does anyone know how Dovahkiin's tax-lot methods match to IBKR's, and according to his data... which would be the optimum?

• First In, First Out (FIFO)
• Last In, First Out (LIFO)
• Maximize Long-Term Gain
• Maximize Long-Term Loss
• Maximize Short-Term Gain
• Maximize Short-Term Loss
• Highest Cost
• Specific Lot

https://www.interactivebrokers.com/en/s ... ethods.htm

I've always assumed the 1/30/2010 was just a typo.

Regarding the tax method, I would think "Maximize Short-Term Loss" would be your best option outside of selecting the Specific Lots yourself (more work and probably very close to the aforementioned anyway.)

[occambogle]

I've always assumed the 1/30/2010 was just a typo.

I suspect you are right and it meant to say 1/30/2020.

Regarding the tax method, I would think "Maximize Short-Term Loss" would be your best option outside of selecting the Specific Lots yourself (more work and probably very close to the aforementioned anyway.)

Thank you... selecting the Specific Lots myself isn't going to happen, so seems Maximize Short-Term Loss would be next best option, followed by LIFO which seems to capture most of the benefit compared to FIFO.

I was thinking about the viability of TLH with pairs like UPRO/TQQQ, and if you did 3-month rebalancing you'd avoid any wash sales. But I would imagine most of the time you'd only be selling the part that had gained..... so unless both sides of the pair had gone down, seems unlikely you'd get much TLH going on.

I've been playing with PV's portfolio optimization tool which is very useful indeed, basically plugging in all the leveraged and non-leveraged ETFs and inputting different volatility and return figures and seeing what comes out, though noting of course it is very limited to the last 10 years (due to recency of many of the ETFs):

PV optimization link with lots of ETFs

What's interesting is looking at the efficient frontiers down below and the high Sharpe ratios of TQQQ and TECL compared to UPRO, but also surpisingly CURE. CURE also seems to work because of a lower correlation with the total market and when you include a little bit of it it seems you can increase some benefits. But of course it is a narrow sector like technology.

Some rough observations, which are unlikely to be a surprise I guess, are that parts of EDV and TYD can be used in parts to reduce leverage, volatility and increase Sharpe. Or VIIX as discussed previously but amounts seem to need to be very precise. While at the other end of the scale TQQQ and CURE (3x Healthcare) can be used to increase returns. Combinations of the two elements seem to be able to provide some interesting options. But I'm well aware there's a lot of recency-bias going on here and I'm certainly not one knowledgeable enough....

I've seen in previous posts people using custom data sets in PV named UPROSIM, TMFSIM, TYDSIM etc that go back further.... are these available to download somewhere?

Who else is currently doing the adventure (or a variant of it) in taxable? And what ETF combinations are you using? Presumably for tax-efficiency a milder form that had lower STDev would be more optimal.... right?

[LittleBitMore]

[/quote]
Please go beyond idle chitchat and feel free to criticize where I have it wrong. I'll even start. In my goal of exploring volatility drag, I didn't provide realistic daily changes.

What if markets moved -8%, +5%, -5%, -10% ... like the S&P 500 did on March 9 to 12. Net loss -17%.
https://finance.yahoo.com/quote/VOO/history?p=VOO
https://finance.yahoo.com/quote/UPRO/history?p=UPRO
UPRO moved -23%, +16%, -15%, -29% during March 9 - 12. That's a net loss of -46%.

S&P 500 are real assets that can recover to the prior value. With one +20%, it reverts to the mean (from the above). But UPRO traces a path 3X that of the S&P 500, with no reversion to the mean. It goes up 3x20% = +60%, but still winds up -14% down even after S&P 500 has recovered. I'd like to understand and possibly mitigate that volatility drag.

One approach mentioned earlier was to inject the difference in 1X and 3X losses as a new cash infusion. With the numbers for Mar 9-12, it looks like about 1/3rd of UPRO's value needs to be kept in cash, and made available, to offset volatility drag.

When VOO / UPRO performs -8% and -23%, that requires +16% added cash (of UPRO's original value)
For VOO / UPRO gaining +5% and +16%, that allows withdrawing -11% from UPRO back into cash
Then VOO / UPRO lose -5% and -15%, requiring +10% cash injected to UPRO.
Finally, VOO / UPRO lose -10% and -29% and the cash need reaches a staggering +19% from one day (Mar 12).

It looks like canceling volatility drag in this case requires +36% additional cash. Scaling that from 136% down to 100%, and I get roughly 3/4ths invested and 1/4th waiting in cash. An even deeper worst case scenario might require 1/3rd or even 1/2 in cash to offset even deeper cumulative drops in UPRO over time.
[/quote]

I don't know if this is actually sustainable though. Let's say you had a sustained drawdown over the course of several months (more like 2008). Because you are "resetting" UPRO by adding cash, each drop takes a material amount of additional cash. Once you start talking about 1/2 cash then you're just talking about market timing during bad times and sitting in cash during good times

[TwoIdenticalIndexes]

If I'm reading these comments correctly, if you go through a lot of trouble to be tax efficient, you will spend ~1% in unnecessary taxes, on top of already paying ~1% on expense ratios. In addition, TEA is presumably the highest expected return part of your portfolio, maximizing your expected taxable capital gains.

Even if you have little space in your IRA, why not devote it all to this strategy? I'm still not seeing the appeal of putting this in taxable.

[Gufomel]

Please go beyond idle chitchat and feel free to criticize where I have it wrong. I'll even start. In my goal of exploring volatility drag, I didn't provide realistic daily changes.

What if markets moved -8%, +5%, -5%, -10% ... like the S&P 500 did on March 9 to 12. Net loss -17%.
https://finance.yahoo.com/quote/VOO/history?p=VOO
https://finance.yahoo.com/quote/UPRO/history?p=UPRO
UPRO moved -23%, +16%, -15%, -29% during March 9 - 12. That's a net loss of -46%.

S&P 500 are real assets that can recover to the prior value. With one +20%, it reverts to the mean (from the above). But UPRO traces a path 3X that of the S&P 500, with no reversion to the mean. It goes up 3x20% = +60%, but still winds up -14% down even after S&P 500 has recovered. I'd like to understand and possibly mitigate that volatility drag.

One approach mentioned earlier was to inject the difference in 1X and 3X losses as a new cash infusion. With the numbers for Mar 9-12, it looks like about 1/3rd of UPRO's value needs to be kept in cash, and made available, to offset volatility drag.

When VOO / UPRO performs -8% and -23%, that requires +16% added cash (of UPRO's original value)
For VOO / UPRO gaining +5% and +16%, that allows withdrawing -11% from UPRO back into cash
Then VOO / UPRO lose -5% and -15%, requiring +10% cash injected to UPRO.
Finally, VOO / UPRO lose -10% and -29% and the cash need reaches a staggering +19% from one day (Mar 12).

It looks like canceling volatility drag in this case requires +36% additional cash. Scaling that from 136% down to 100%, and I get roughly 3/4ths invested and 1/4th waiting in cash. An even deeper worst case scenario might require 1/3rd or even 1/2 in cash to offset even deeper cumulative drops in UPRO over time.


I don't know if this is actually sustainable though. Let's say you had a sustained drawdown over the course of several months (more like 2008). Because you are "resetting" UPRO by adding cash, each drop takes a material amount of additional cash. Once you start talking about 1/2 cash then you're just talking about market timing during bad times and sitting in cash during good times

What about moving 100% from UPRO to SPY if VIX hits a certain level (let’s say 25)? The intention is not to time the market (you’re still staying in the market) but to protect against extreme volatility drag or the need for extreme infusion of cash to negate the volatility drag of a 3x position. The possibility of having to buy back in at a higher UPRO price (whenever VIX drops back below 25 in this example) could be worth the downside protection against being leveraged during huge declines like in March where volatility crushes UPRO or requires massive infusion of cash to counteract the volatility drag. Only having looked at this briefly, it seems that having to buy back into UPRO at a much higher price is fairly uncommon because the high volatility that typically occurs when VIX is high eats into UPRO even if SPY rises during the period where VIX is above 25. The advantage of this is you wouldn’t have to rely on treasuries (TMF) being up while stocks are down.

When VIX is lower, it should be more feasible to negate the volatility drag with cash infusion.

I’m pretty much a novice at this. Just talking ideas.

[KSActuary]

This appears to be the preferred strategy as Exposure remains, just the dent is decreased if the market turns down.

The original risk parity approach that had widespread commercial success was the All Weather strategy which supposedly only leveraged the fixed income side to be in parity with the volatilities of the equity components. Given the questions surrounding the usefulness of fixed income as a future volatility dampener, deleveraging the equity side is certainly another alternative.

[LEVERAGED INVESTOR]

This strategy works very well with UOPIX/TMF or even better with TQQQ at higher Sharpe and Sortino. It makes money even during correction because it assigns a larger portion of assets to TMF during corrections. You need flexibility not just 50/50 all the time.

https://www.portfoliovisualizer.com/tes ... odWeight=0

[Gufomel]

I think I’ve seen discussion here and elsewhere about targeting volatility by using last 20 days (as an example) volatility. Is that correct? Has that been shown to be predictive of upcoming volatility? How does that compare to using something like VIX? Seems like VIX would be more forward looking.

[kmft]

I've seen in previous posts people using custom data sets in PV named UPROSIM, TMFSIM, TYDSIM etc that go back further.... are these available to download somewhere?

From the very first post of the adventure; viewtopic.php?f=10&t=272007#p4364001

How confident are you that your simulated backtest actually tracks the real UPRO and TMF?
UPRO tracks extremely well. TMF has some issues, but siamond's analysis indicates that since 2013 it has also "fallen in line" with the formula. I have uploaded the data here, if you would like to play with it yourself. Samsdad has provided step-by-step directions on how to run this dataset in PortfolioVisualizer

You'll need the paid version of PV.

[occambogle]

From the very first post of the adventure; viewtopic.php?f=10&t=272007#p4364001

Thanks a lot. My bad... I forgot to look at the original MkI thread. Got them now. But I also see mention in this post of a TYDSIM... but don't find it. In any case, I guess this would be the right thread to ask the question so will go there....

[Lee_WSP]

If I'm reading these comments correctly, if you go through a lot of trouble to be tax efficient, you will spend ~1% in unnecessary taxes, on top of already paying ~1% on expense ratios. In addition, TEA is presumably the highest expected return part of your portfolio, maximizing your expected taxable capital gains.

Even if you have little space in your IRA, why not devote it all to this strategy? I'm still not seeing the appeal of putting this in taxable.

You're only paying taxes if you make money. Is there a higher yielding strategy that would only incur ltcg?

[kmft]

From the very first post of the adventure; viewtopic.php?f=10&t=272007#p4364001

Thanks a lot. My bad... I forgot to look at the original MkI thread. Got them now. But I also see mention in this post of a TYDSIM... but don't find it. In any case, I guess this would be the right thread to ask the question so will go there....

This file has TYD simulated back to 1955, see the "(click here to download it)" at the bottom of this post; viewtopic.php?f=10&t=288192&sid=924191e ... 0#p5027049 Not sure how you'd import into PV though, as it's yearly data.

Interestly enough you can backtest EDV all the way back to 1871, here it is compared to total stock market (through 2018);

[occambogle]

Interestly enough you can backtest EDV all the way back to 1871, here it is compared to total stock market (through 2018);

Wow, that's some serious period of backtesting... and thanks for the link to the LETF version of Simba sheet.

[IndexCore]

Unfortunately I tried that in March 9-12 and it went very badly... during the actual panic, all assets moved together (excepting short-term treasuries). I correctly predicted a massive drop, and got screwed out almost all of the gain by the correlations of other assets. At the time, I couldn't buy 3x bear S&P 500, since my brokerage didn't allow purchases of leveraged ETFs. I've since shifted some money elsewhere, and bought some shares of UPRO.

I rebalanced 3.6 and 3.10. Nailing the 9th would have been best but such is life.

Long Term Treasuries definitely went up in the crisis. There was a period where liquidity hit and markets melted down. Perhaps it wasn't so evident in short term treasuries but municipal bonds had even worse trouble at least until the Fed restored liquidity.

Anyway, I'm up 24% after about 13 months. I have 3X international positions as well so probably others have done better.

If you already held long-term treasuries when the market closed on Mar 6, that was ideal: the Fed cut interest rates on Mar 8, from 1% to 0%. Look at the bond fund's duration, and that's your profit! But for me, entering the market March 9, it was already too late to buy before that gain. A few minutes after the market opened on March 9 (and before the circuit breaker halted trading), Vanguard Extended Duration ETF (EDV) was already up +10%.

From there, minutes into March 9, until March 20, it was mostly capital losses. I recall taking a very small position in long-term zero coupon bonds (ZROZ) on March 20 since the panic was about to break. From Mar 20-27, yields dropped and long-term bonds profited. I would say long-term bonds were very bad during the panic which lasted from Mar 9-20, but during the overall crisis they had some great moments (Mar 6-9, after Mar 20).

[Mickelous]

If you were to add UGLD to this portfolio for combating hyper inflation, what amount would you recommend? It's a 3x leveraged gold bullion tracking ETF. So a little under 7% of the portfolio in UGLD would give you 20% exposure to gold in case of hyper inflation. I wonder what back tests show with draw downs and affecting total returns as a whole.

[keith6014]

If I'm reading these comments correctly, if you go through a lot of trouble to be tax efficient, you will spend ~1% in unnecessary taxes, on top of already paying ~1% on expense ratios. In addition, TEA is presumably the highest expected return part of your portfolio, maximizing your expected taxable capital gains.

Even if you have little space in your IRA, why not devote it all to this strategy? I'm still not seeing the appeal of putting this in taxable.

You're only paying taxes if you make money. Is there a higher yielding strategy that would only incur ltcg?

Exactly. If you are making money why be afraid of taxes? If you boss gives you a raise do you say, sorry I don't want it because I pay more on taxes? Do the strategy using futures. Much more tax friendly.

[HEDGEFUNDIE]

If I'm reading these comments correctly, if you go through a lot of trouble to be tax efficient, you will spend ~1% in unnecessary taxes, on top of already paying ~1% on expense ratios. In addition, TEA is presumably the highest expected return part of your portfolio, maximizing your expected taxable capital gains.

Even if you have little space in your IRA, why not devote it all to this strategy? I'm still not seeing the appeal of putting this in taxable.

You're only paying taxes if you make money. Is there a higher yielding strategy that would only incur ltcg?

Exactly. If you are making money why be afraid of taxes? If you boss gives you a raise do you say, sorry I don't want it because I pay more on taxes? Do the strategy using futures. Much more tax friendly.

You don't "make money" until you cash out of the strategy. The key question is how much are you losing in taxes along the ride.

[keith6014]

You make money in rebalancing. Hopefully.

[Rad Dad]

If I'm reading these comments correctly, if you go through a lot of trouble to be tax efficient, you will spend ~1% in unnecessary taxes, on top of already paying ~1% on expense ratios. In addition, TEA is presumably the highest expected return part of your portfolio, maximizing your expected taxable capital gains.

Even if you have little space in your IRA, why not devote it all to this strategy? I'm still not seeing the appeal of putting this in taxable.

You're only paying taxes if you make money. Is there a higher yielding strategy that would only incur ltcg?

Exactly. If you are making money why be afraid of taxes? If you boss gives you a raise do you say, sorry I don't want it because I pay more on taxes? Do the strategy using futures. Much more tax friendly.

Is it really more tax friendly with futures? If you use micro e-minis aren’t you going to have to pay 60% LTCG and 40% STCG every 3 months when you roll over a contract?

[privatefarmer]

If you were to add UGLD to this portfolio for combating hyper inflation, what amount would you recommend? It's a 3x leveraged gold bullion tracking ETF. So a little under 7% of the portfolio in UGLD would give you 20% exposure to gold in case of hyper inflation. I wonder what back tests show with draw downs and affecting total returns as a whole.

I’ve been researching the effects of adding UGLD to this strategy. From what I’ve seen however, it seems that UGLD has more volatility drag. It seems that’s the leveraged portfolio does not track the unleveraged version as well when UGLD is added. But if this changes I would think adding gold would make this strategy more diversified and more prepared to handle unexpected inflation.

[GiveTendies]

This is also why most discussions about long term treasuries end up focusing on the wrong arguments. The decision to invest in long term treasuries is not affected by the risk free rate being 10% of -5%. A bet on long term treasuries is a bet on the persistence of a term premium, which can only be indirectly observed. Frankly there is zero reason why the expected term premium is different from 10, 20 years ago. Pay close attention to the word expected here. There is no way the realized term premium the last 20 years is a good predictor of expected term premium, this is also why backtests involving treasuries are completely useless.

Long term bonds will (over long term) always outperform short term bonds. This is because of their massively higher interest rate and inflation risk. Higher risk = higher reward.
I'm not sure if I agree with the yields. To a lot of people/companies 5% safe vs. 10% equities is a lot better than 0% safe vs. 5% equities.

That's false, but might not be apparent in the past 30 years of data. Besides interest payments, bonds have capital gains when yields fall. A bond yielding 5% is worth more than a bond yielding 4%, which is reflected in it's sale price. But the opposite is also true, that rising yields make prior bonds look worse, and inflict a capital loss. Those capital losses can easily be greater than the bond's yield, in which case longer-term bonds lose more than short-term bonds.

For long-term bond funds, there's an important number called "bond duration" that acts as a multiplier on changes in yield. When interest rates go up... and bond yields follow, then you can see an impact proportional to the bond fund's duration. A short-term bond fund might have a duration of 3 years, while long-term bond funds can be 15-25 years.

I don't get how your explanation contradicts mine, so what is incorrect that i said?

A bonds long-term gain can be quoted by the yield to maturity, which is a combination of coupon payments + capital gain/loss when the principal is repaid. Clearly, bonds today have much lower YTM than 20 years ago. This is bad news for companies, who are legally forced to invest in bonds and not allowed to invest in equity.

[IndexCore]

This is also why most discussions about long term treasuries end up focusing on the wrong arguments. The decision to invest in long term treasuries is not affected by the risk free rate being 10% of -5%. A bet on long term treasuries is a bet on the persistence of a term premium, which can only be indirectly observed. Frankly there is zero reason why the expected term premium is different from 10, 20 years ago. Pay close attention to the word expected here. There is no way the realized term premium the last 20 years is a good predictor of expected term premium, this is also why backtests involving treasuries are completely useless.

Long term bonds will (over long term) always outperform short term bonds. This is because of their massively higher interest rate and inflation risk. Higher risk = higher reward.
I'm not sure if I agree with the yields. To a lot of people/companies 5% safe vs. 10% equities is a lot better than 0% safe vs. 5% equities.

That's false, but might not be apparent in the past 30 years of data. Besides interest payments, bonds have capital gains when yields fall. A bond yielding 5% is worth more than a bond yielding 4%, which is reflected in it's sale price. But the opposite is also true, that rising yields make prior bonds look worse, and inflict a capital loss. Those capital losses can easily be greater than the bond's yield, in which case longer-term bonds lose more than short-term bonds.

For long-term bond funds, there's an important number called "bond duration" that acts as a multiplier on changes in yield. When interest rates go up... and bond yields follow, then you can see an impact proportional to the bond fund's duration. A short-term bond fund might have a duration of 3 years, while long-term bond funds can be 15-25 years.

I don't get how your explanation contradicts mine, so what is incorrect that i said?

A bonds long-term gain can be quoted by the yield to maturity, which is a combination of coupon payments + capital gain/loss when the principal is repaid. Clearly, bonds today have much lower YTM than 20 years ago. This is bad news for companies, who are legally forced to invest in bonds and not allowed to invest in equity.

Since future capital gains and losses are unknown, they cannot be reflected in yield to maturity. A bond fund may show 2% yield to maturity now, but if interest rates go up, it gets an unexpected capital loss. Yield to maturity doesn't reflect that.

If your "long term" is decades, then I don't have specific examples of short-term bonds beating long-term bonds over decades. For specific years, certainly (2013, 2015, 2018)... and my view is those years will be more common in a rising interest rate environment, possibly leading to a long time where short-term bonds beat long-term bonds.

[kim.gold]

If I'm reading these comments correctly, if you go through a lot of trouble to be tax efficient, you will spend ~1% in unnecessary taxes, on top of already paying ~1% on expense ratios. In addition, TEA is presumably the highest expected return part of your portfolio, maximizing your expected taxable capital gains.

Even if you have little space in your IRA, why not devote it all to this strategy? I'm still not seeing the appeal of putting this in taxable.

You're only paying taxes if you make money. Is there a higher yielding strategy that would only incur ltcg?

Exactly. If you are making money why be afraid of taxes? If you boss gives you a raise do you say, sorry I don't want it because I pay more on taxes? Do the strategy using futures. Much more tax friendly.

Is it really more tax friendly with futures? If you use micro e-minis aren’t you going to have to pay 60% LTCG and 40% STCG every 3 months when you roll over a contract?

Future contracts are marked to market and must be settled daily as the price changes. That means that every trading day at 5pm EDT your unrealized P&L becomes realized.

[firebirdparts]

I don't get how your explanation contradicts mine, so what is incorrect that i said?

A bonds long-term gain can be quoted by the yield to maturity, which is a combination of coupon payments + capital gain/loss when the principal is repaid. Clearly, bonds today have much lower YTM than 20 years ago. This is bad news for companies, who are legally forced to invest in bonds and not allowed to invest in equity.

More fundamentally than what indexcore responded to, your idea that long term bonds have to outperform is not true at all, but it may well turn out to be true. They have been performing. The premium for holding 30 years is only about 1% and that's just not enough to guarantee they will outperform. They have been.

Everybody here knows that, and I think it was just not interesting for others to answer you on that point. We've been talking about all this for a year.
There is plenty of concern with using this strategy in a situation where long term bonds are declining. Out of the 10,000 or so posts total on this strategy, I would say well over 1000 of them are about that concern. There are several threads on this subject, and some of them are totally devoted to using something other than TMF.

[boglerdude]

Would this strategy work with German/Japan 30yrs or is their volatility too low because they're near zero?

[keith6014]

If I'm reading these comments correctly, if you go through a lot of trouble to be tax efficient, you will spend ~1% in unnecessary taxes, on top of already paying ~1% on expense ratios. In addition, TEA is presumably the highest expected return part of your portfolio, maximizing your expected taxable capital gains.

Even if you have little space in your IRA, why not devote it all to this strategy? I'm still not seeing the appeal of putting this in taxable.

You're only paying taxes if you make money. Is there a higher yielding strategy that would only incur ltcg?

Exactly. If you are making money why be afraid of taxes? If you boss gives you a raise do you say, sorry I don't want it because I pay more on taxes? Do the strategy using futures. Much more tax friendly.

Is it really more tax friendly with futures? If you use micro e-minis aren’t you going to have to pay 60% LTCG and 40% STCG every 3 months when you roll over a contract?

Future contracts are marked to market and must be settled daily as the price changes. That means that every trading day at 5pm EDT your unrealized P&L becomes realized.

I had no idea. So everyday its a taxable event?