HEDGEFUNDIE's excellent adventure Part II: The next journey

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
corpgator
Posts: 217
Joined: Wed Jul 22, 2015 12:00 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by corpgator »

akxc wrote: Thu Jan 06, 2022 10:04 pm The following chart shows the difference in annualized return for quarterly compared to daily rebalancing based on which day of the quarter is used for rebalancing. I also deducted a 0.1% expense on the amount being transferred between the two funds in each rebalance (current 30 day bid ask spread is .04% for TMF and .02% for UPRO). Sure enough, rebalancing right on the quarter has significantly outperformed other rebalancing periods. But on average, there is a significant benefit to rebalancing daily. Over the simulated Hedgefundie 1986-2019 dataset, this works out to a 92 basis point premium for rebalancing daily over quarterly. I'm inclined to believe the benefit of rebalancing on/around the quarter over the actual life of UPRO/TMF is just by chance based on this analysis.

Image
Why do you believe it's by chance? There are other stock market quirks that exist that make it possible that rebalancing on the quarter is simply better due to when people deposit money in or withdraw from the market.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

corpgator wrote: Fri Jan 07, 2022 3:55 am
akxc wrote: Thu Jan 06, 2022 10:04 pm The following chart shows the difference in annualized return for quarterly compared to daily rebalancing based on which day of the quarter is used for rebalancing. I also deducted a 0.1% expense on the amount being transferred between the two funds in each rebalance (current 30 day bid ask spread is .04% for TMF and .02% for UPRO). Sure enough, rebalancing right on the quarter has significantly outperformed other rebalancing periods. But on average, there is a significant benefit to rebalancing daily. Over the simulated Hedgefundie 1986-2019 dataset, this works out to a 92 basis point premium for rebalancing daily over quarterly. I'm inclined to believe the benefit of rebalancing on/around the quarter over the actual life of UPRO/TMF is just by chance based on this analysis.

Image
Why do you believe it's by chance? There are other stock market quirks that exist that make it possible that rebalancing on the quarter is simply better due to when people deposit money in or withdraw from the market.
The red chart represents only 12 years, or 48 quarters. The blue vs. red chart comparison suggests that about the entire "outperformance" of on-the-quarter rebalancing is from the 2009-2021 period.
On the other hand, since a rebalancing event is only with a relatively small portion of the assets, the 5% p.a. outperformance seems huge.

Let's look at the chart.
The biggest stock market drawdown during the 2009-2021 period (March 2020) bottomed out almost exactly at the end of March 2020. The drawdown from the previous high to 03/31/2020 was ca. 25%. By contrast, an unlucky guy who used 02/28/2020 and every quarter before/after that date as rebalancing days, would have almost completely lost out on his rebalancing opportunity from the drawdown, and also on the rebalancing opportunity from the previous peak.
Sep 2018 marked almost exactly a local peak. Dec 31 2018 marked almost exactly a local trough.
The spring 2009 bottom lasted more than a quarter, so no bias.
I can't really see how the rebalancing from those peaks and troughs could come even close to 5% p.a. Can you please check your data and calcs? Maybe 5% is the total return difference over the period, not p.a.?
Last edited by comeinvest on Fri Jan 07, 2022 3:08 pm, edited 1 time in total.
Hydromod
Posts: 1051
Joined: Tue Mar 26, 2019 10:21 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Hydromod »

akxc wrote: Thu Jan 06, 2022 10:04 pm The following chart shows the difference in annualized return for quarterly compared to daily rebalancing based on which day of the quarter is used for rebalancing. I also deducted a 0.1% expense on the amount being transferred between the two funds in each rebalance (current 30 day bid ask spread is .04% for TMF and .02% for UPRO). Sure enough, rebalancing right on the quarter has significantly outperformed other rebalancing periods. But on average, there is a significant benefit to rebalancing daily. Over the simulated Hedgefundie 1986-2019 dataset, this works out to a 92 basis point premium for rebalancing daily over quarterly. I'm inclined to believe the benefit of rebalancing on/around the quarter over the actual life of UPRO/TMF is just by chance based on this analysis.
This is pretty much what I had found. To be clear on the details, each curve represents 63 separate simulations, one for each trading day in the first quarter, with each simulation run for the same number of trading days and rebalancing on the same trading day of the quarter?

To get at timing luck, a series of fixed-duration simulations are nice. There was a post on reddit I linked to here that looked at a whole series of simulations over time. I haven't tried the calculation, but if the work holds up it is some strong evidence that there is a systematic benefit based on day of the quarter (in other words, not entirely timing luck).

I'm surprised that the daily trade still won over quarterly with the slippage included, even though it is better without (at least based on closing prices). Now I'm wondering whether some issue arises by the fluctuations between open and close for the two funds for real trading (i.e., TMF systematically lags UPRO during the day), since the daily close is not practical. M1 limits trades to two windows, for example, which may not have prices that are all that correlated to the daily close.
12casual
Posts: 1
Joined: Fri Jan 07, 2022 6:30 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by 12casual »

Admittedly a new poster here, but I've been a long time reader, and felt compelled to share thoughts that I'd appreciate other perspectives on:

I'm starting to see more posts here suggesting short-term enhancements/modifications to the traditional HFEA strategy (i.e. adding or substituting in VIX-correlated assets, Cash, CWEB, FAS, etc) based on current market conditions. Some of these have been debated more extensively with valuable data provided for thesis justification (i.e. VIX & Cash).

While I read this forum daily in anticipation that someone will suggest an exciting modification to the strategy, my humble opinion is that it's important for contributors to remember that the strategy has largely become popular due to its simplicity (2 ETFs) and ability to produce incredible risk-adjusted returns through extensive simulated backtests (including multiple recessions)... even if only rebalancing quarterly/annually.

I see nothing wrong with investing outside of UPRO/TMF (many contributors here only allocate a small % of their portfolio to HFEA). I often do this myself with other 3X funds like TECL/TQQQ/SOXL/FNGU/etc for a tech edge. However, if those additions to the strategy wouldn't produce outperformance vs. traditional HFEA in backtests (or simulated backtests), then I expect it will be hard for many HFEA followers to have faith in making them a permanent modification to their HFEA strategy.

We all acknowledge that past performance of HFEA doesn't guarantee future returns. Contributors have made strong arguments for adding TQQQ to the original strategy (despite essentially getting wiped out in a simulated '00 crash), and the rate increase concern has been justifiably debated as well although I still haven't seen a time-tested substitute for TMF that performs as a better hedge. However, these debates are typically coupled with extensive backtests to keep them in longer-term perspective. I personally feel like that's been missing in some of the more recent suggestions to adjust the strategy.

I'm curious to hear thoughts from others. Even if there's strong reason to believe that added funds might outperform traditional HFEA over the next year or two or more, would you consider those funds to be part of your HFEA strategy even if you don't believe you'll be holding them for your entire HFEA investment timeline? Do we need a "Modified HFEA" forum for that? (FWIW, I'd probably read that daily too :happy )

Or is this sentiment counterproductive to the value that comes from healthy discourse on this forum?
User avatar
tomphilly
Posts: 178
Joined: Wed Aug 05, 2020 11:29 am
Location: Philadelphia

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by tomphilly »

akxc wrote: Thu Jan 06, 2022 10:04 pm The following chart shows the difference in annualized return for quarterly compared to daily rebalancing based on which day of the quarter is used for rebalancing. I also deducted a 0.1% expense on the amount being transferred between the two funds in each rebalance (current 30 day bid ask spread is .04% for TMF and .02% for UPRO). Sure enough, rebalancing right on the quarter has significantly outperformed other rebalancing periods. But on average, there is a significant benefit to rebalancing daily. Over the simulated Hedgefundie 1986-2019 dataset, this works out to a 92 basis point premium for rebalancing daily over quarterly. I'm inclined to believe the benefit of rebalancing on/around the quarter over the actual life of UPRO/TMF is just by chance based on this analysis.

Image
Thank you for this. Are you able to determine CAGR and max drawdown for a daily re-balanced UPRO/TMF 60/40? With PV we can only rebalance as frequently as monthly.
BayStater
Posts: 97
Joined: Sun Mar 29, 2020 11:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by BayStater »

12casual wrote: Fri Jan 07, 2022 7:43 am While I read this forum daily in anticipation that someone will suggest an exciting modification to the strategy, my humble opinion is that it's important for contributors to remember that the strategy has largely become popular due to its simplicity (2 ETFs) and ability to produce incredible risk-adjusted returns through extensive simulated backtests (including multiple recessions)... even if only rebalancing quarterly/annually.

Or is this sentiment counterproductive to the value that comes from healthy discourse on this forum?
I think the concerns are mainly around how well the strategy will hold in an environment with rising rates, inflation, low expected returns on equity.

Personally, I've thought about TQQQ/FNGU and the rest of it. I'm much more comfortable with the equity portion in a broad market index like 3x S&P 500. In any case, if I had to place my bets I'd say if the strategy fails going forward it will be because of bonds, not stocks.

Finding an alternative hedge for TMF is more important than an alternative for UPRO.
DMoogle
Posts: 549
Joined: Sat Oct 31, 2020 10:24 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DMoogle »

BayStater wrote: Fri Jan 07, 2022 9:03 amI think the concerns are mainly around how well the strategy will hold in an environment with rising rates, inflation, low expected returns on equity.

Personally, I've thought about TQQQ/FNGU and the rest of it. I'm much more comfortable with the equity portion in a broad market index like 3x S&P 500. In any case, if I had to place my bets I'd say if the strategy fails going forward it will be because of bonds, not stocks.

Finding an alternative hedge for TMF is more important than an alternative for UPRO.
@zkn put together a truly EXCELLENT post on different approaches to mitigating that risk: viewtopic.php?p=6391312#p6391312

Regarding equity underperformance, I'd argue that going a DIY route with futures/box spreads to leverage equities instead of LETFs is an excellent way to combat it, because you can avoid volatility decay pretty much completely, so that sideways or long-term mediocre performance won't hurt as bad.
DarkMatter731
Posts: 224
Joined: Sat Apr 25, 2020 8:28 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DarkMatter731 »

BayStater wrote: Fri Jan 07, 2022 9:03 am
12casual wrote: Fri Jan 07, 2022 7:43 am While I read this forum daily in anticipation that someone will suggest an exciting modification to the strategy, my humble opinion is that it's important for contributors to remember that the strategy has largely become popular due to its simplicity (2 ETFs) and ability to produce incredible risk-adjusted returns through extensive simulated backtests (including multiple recessions)... even if only rebalancing quarterly/annually.

Or is this sentiment counterproductive to the value that comes from healthy discourse on this forum?
I think the concerns are mainly around how well the strategy will hold in an environment with rising rates, inflation, low expected returns on equity.

Personally, I've thought about TQQQ/FNGU and the rest of it. I'm much more comfortable with the equity portion in a broad market index like 3x S&P 500. In any case, if I had to place my bets I'd say if the strategy fails going forward it will be because of bonds, not stocks.

Finding an alternative hedge for TMF is more important than an alternative for UPRO.
Finding an alternative hedge for TMF is more important than an alternative for UPRO.
Worst comes to worst, you can always just buy collars (buying protective puts and writing covered calls) that act as a hedge. It won't be as good as TMF but it will be a guaranteed hedge.
miabullitt
Posts: 1
Joined: Thu Jan 06, 2022 8:54 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by miabullitt »

privatefarmer wrote: Wed Jan 05, 2022 9:28 am
Maxx2 wrote: Tue Jan 04, 2022 10:16 pm Does anyone rebalance monthly using minimum variance? There are previous posts that suggest it outperforms 55/45 and has lower drawdowns.
I do. I use this strategy specifically :

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

It has tracked the unleveraged version quite well :

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

If using mutual funds instead of ETFs you can go back to the late 80s I’ve estimated that the leveraged version would’ve returned about a 45% CAGR with a max DD of about 50%.
Having gone through a lot of this post history, I like your thinking. Do you run this in a taxable account? Seems like would have a lot of tax leakage constantly rotating in and out of positions given pulling in 3 from a basket of [x+3]?
akxc
Posts: 19
Joined: Thu Jan 06, 2022 5:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by akxc »

Hydromod wrote: Fri Jan 07, 2022 7:20 am This is pretty much what I had found. To be clear on the details, each curve represents 63 separate simulations, one for each trading day in the first quarter, with each simulation run for the same number of trading days and rebalancing on the same trading day of the quarter?
Yes.
Hydromod wrote: Fri Jan 07, 2022 7:20 am To get at timing luck, a series of fixed-duration simulations are nice. There was a post on reddit I linked to here that looked at a whole series of simulations over time. I haven't tried the calculation, but if the work holds up it is some strong evidence that there is a systematic benefit based on day of the quarter (in other words, not entirely timing luck).
Don't have time to dig into too much of a deeper analysis, but here are the results from 1986-2021 data broken down into discrete ~7 year subsets:
Image
Hydromod wrote: Fri Jan 07, 2022 7:20 am I'm surprised that the daily trade still won over quarterly with the slippage included, even though it is better without (at least based on closing prices). Now I'm wondering whether some issue arises by the fluctuations between open and close for the two funds for real trading (i.e., TMF systematically lags UPRO during the day), since the daily close is not practical. M1 limits trades to two windows, for example, which may not have prices that are all that correlated to the daily close.
Agreed. I don't trade daily, so I don't have experience with how this would turn out in practice, and don't know if my 0.1% transaction cost is a good estimate. For example, I could imagine there being larger bid-ask spreads on high volatility days which would also likely be when you are transitioning larger amounts between the two funds.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

akxc wrote: Fri Jan 07, 2022 12:18 pm
Hydromod wrote: Fri Jan 07, 2022 7:20 am To get at timing luck, a series of fixed-duration simulations are nice. There was a post on reddit I linked to here that looked at a whole series of simulations over time. I haven't tried the calculation, but if the work holds up it is some strong evidence that there is a systematic benefit based on day of the quarter (in other words, not entirely timing luck).
Don't have time to dig into too much of a deeper analysis, but here are the results from 1986-2021 data broken down into discrete ~7 year subsets:
Image
Hydromod wrote: Fri Jan 07, 2022 7:20 am I'm surprised that the daily trade still won over quarterly with the slippage included, even though it is better without (at least based on closing prices). Now I'm wondering whether some issue arises by the fluctuations between open and close for the two funds for real trading (i.e., TMF systematically lags UPRO during the day), since the daily close is not practical. M1 limits trades to two windows, for example, which may not have prices that are all that correlated to the daily close.
Agreed. I don't trade daily, so I don't have experience with how this would turn out in practice, and don't know if my 0.1% transaction cost is a good estimate. For example, I could imagine there being larger bid-ask spreads on high volatility days which would also likely be when you are transitioning larger amounts between the two funds.
Did you ever run similar simulations on non-leveraged products, by any chance? Curious if this phenomenon is specific to the 3x leveraged products, or the underlying markets.
ceskejebenice
Posts: 1
Joined: Fri Jan 07, 2022 6:47 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by ceskejebenice »

Hi guys!

Can someone explain to me how exactly TMF and LTT work.

If rates go up, current bonds will be lowered price (because new ones will have more attractive rates). How does TMF then move when rates go up? I see negatively too and I see three things I need more clarification if anyone knows.

1. leverage used for TMF will be more expensive - what would the effect be?
2. current bonds under TMF will go down in value, but should be replaced with new ones? does TMF buy new ones and track those, or index of old?
3. if it actually moves with rates directly but is rebalanced daily, then it isn't really a long term hold, but more of a short-term hold while the rates consistently go down

Isn't TMV then a better hedge for TQQQ now?
DarkMatter731
Posts: 224
Joined: Sat Apr 25, 2020 8:28 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DarkMatter731 »

ceskejebenice wrote: Fri Jan 07, 2022 6:52 pm Hi guys!

Can someone explain to me how exactly TMF and LTT work.

If rates go up, current bonds will be lowered price (because new ones will have more attractive rates). How does TMF then move when rates go up? I see negatively too and I see three things I need more clarification if anyone knows.

1. leverage used for TMF will be more expensive - what would the effect be?
2. current bonds under TMF will go down in value, but should be replaced with new ones? does TMF buy new ones and track those, or index of old?
3. if it actually moves with rates directly but is rebalanced daily, then it isn't really a long term hold, but more of a short-term hold while the rates consistently go down

Isn't TMV then a better hedge for TQQQ now?
This has been rehashed so many times now. I really don't want to be rude but this has been rehashed so many times now.

If you don't have conviction in it, don't invest in it. You're being paid a risk premium for investing in UPRO/TMF because it isn't certain.

TMV isn't going to be a better hedge during a drawdown. Think logically - institutional investors need a safe place to put money during a market crash. US treasuries are the safest option.
Walkure
Posts: 1023
Joined: Tue Apr 11, 2017 9:59 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Walkure »

Hydromod wrote: Thu Jan 06, 2022 8:38 am
comeinvest wrote: Wed Jan 05, 2022 7:59 pm I read some of your posts and some of your thread regarding this. But you seem to contradict yourself. What is your final conclusion - is the starting date selection "timing luck", or do you have confidence in the turn-of-the-quarter starting date advantage? Obviously, the turn-of-the-quarter an only be chosen with quarterly rebalancing, so by implication, quarterly would then be superior to smaller intervals. Thanks for clarification.

It looks like there was another backtest showing a ca. 3% p.a. performance advantage of quarterly vs. monthly. That seems big. Is there any plausible explanation for this? I understand we have to subtract the part of it that is attributable to additional risk as more money stayed longer in higher-return asset classes; how large is that effect? Also, if there was high confidence in a quarterly rebalancing bonus vs alternate starting dates, I think that could be easily exploited beyond rebalancing via a strategy.
I guess a good place to illustrate the starting point issue is the first plot here. The gray line is the original 40/60 HFEA strategy.

I ran a separate simulation using UPROSIM and TMFSIM starting each day in the first quarter of the data. With 60-day rebalancing, the spread in CAGR for the next 32 years ranged from 14 to 16.5%. This isn't precisely quarterly rebalancing, but close.

The first figure here illustrates the timing luck. These were the same type of simulation, starting each day of the first quarter. All of these simulations have exactly the same rebalancing schedule, just offset by a single day one to the next. There are distinct jumps at certain events. Those jumps illustrate the timing luck of starting date, which widens the spread in returns over time.

So I think that historically there appears to have been an actual bias to quarterly rebalancing on the quarters, but I don't feel comfortable claiming that this can't be coincidence. I recommend folks rebalance on the turn of the quarter when doing quarterly rebalancing, but try to claim that this may just be superstition based on timing luck during a handful of consequential events experienced during backtesting.
In my case, I’ve settled into a pattern of “bucketed bi-quarterly” rebalancing. Because I’m running both a Roth and a taxable bucket, I alternate between rebalancing each sub-account every 45 days to guarantee I’ll never have a 30 day wash sale violation in my Roth. I’ve had to accept that this also means that I’m not fully at target allocation at any given point in time across the entire portfolio, and there is slight tracking error between the two buckets, but so far I haven’t noticed a huge difference.
akxc
Posts: 19
Joined: Thu Jan 06, 2022 5:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by akxc »

Did you ever run similar simulations on non-leveraged products, by any chance? Curious if this phenomenon is specific to the 3x leveraged products, or the underlying markets.
I assume you are referring to the spread in returns based on rebalance date? This is much more relevant to leveraged products, and especially a strategy like this using two negatively correlated assets where there are many days with large spreads between the two ETFs, which combined with the compounding effect of daily rebalanced leverage can lead to high sensitivity to input parameters.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

akxc wrote: Sat Jan 08, 2022 3:29 pm
Did you ever run similar simulations on non-leveraged products, by any chance? Curious if this phenomenon is specific to the 3x leveraged products, or the underlying markets.
I assume you are referring to the spread in returns based on rebalance date? This is much more relevant to leveraged products, and especially a strategy like this using two negatively correlated assets where there are many days with large spreads between the two ETFs, which combined with the compounding effect of daily rebalanced leverage can lead to high sensitivity to input parameters.
Got it, but still, I'm curious if this phenomenon is an intrinsic artifact of the 3x ETFs, or a (magnified) artifact of the underlying markets.

I personally use futures for a HFEA implementation. Not sure if the dependency of returns on rebalancing dates would affect a futures based implementation. But nevertheless I'm curious, as I'm trying to figure out my rebalancing strategy.
akxc
Posts: 19
Joined: Thu Jan 06, 2022 5:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by akxc »

comeinvest wrote: Sun Jan 09, 2022 1:12 am Got it, but still, I'm curious if this phenomenon is an intrinsic artifact of the 3x ETFs, or a (magnified) artifact of the underlying markets.
Easy to take a quick look at this in PV.

Calendar Aligned rebalance
Rebalance +1 month from calendar aligned

Much, much larger difference in CAGR for the levered vs. the unlevered portfolio over the life of UPRO & TMF.
Hydromod
Posts: 1051
Joined: Tue Mar 26, 2019 10:21 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Hydromod »

akxc wrote: Sun Jan 09, 2022 10:29 am
comeinvest wrote: Sun Jan 09, 2022 1:12 am Got it, but still, I'm curious if this phenomenon is an intrinsic artifact of the 3x ETFs, or a (magnified) artifact of the underlying markets.
Easy to take a quick look at this in PV.

Calendar Aligned rebalance
Rebalance +1 month from calendar aligned

Much, much larger difference in CAGR for the levered vs. the unlevered portfolio over the life of UPRO & TMF.
Be careful about generalizations. If you look at the ratio of the two UPRO/TMF portfolios, it seems clear that most or all of the discrepancy is timing luck on the rebalance.

Image
akxc
Posts: 19
Joined: Thu Jan 06, 2022 5:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by akxc »

Hydromod wrote: Sun Jan 09, 2022 12:52 pm Be careful about generalizations. If you look at the ratio of the two UPRO/TMF portfolios, it seems clear that most or all of the discrepancy is timing luck on the rebalance.
Right, I am still in the thinking that most of the historic out-performance of quarterly rebalancing is based on timing luck. Clearly there has not been a consistent "premium", but instead the outperformance has been a few well-timed rebalances.

To further answer comeinvest's question, here is the same chart, but also with the quarterly outperformance for the unlevered portfolio:
Image
The same bumps appear in the unlevered portfolio, but the nature of LEFTs make the effects of the timing of rebalances much, much larger.

So to sum up my thoughts on a few issues that have been gone over lately in this thread:
1) Rebalance timing has a much bigger impact on end performance with daily-rebalanced leveraged ETFs than with non-leveraged strategies.
2) Historically HFEA has performed better with rebalances on the calendar quarter. This is due to a select few rebalances that created most of the outsized gains and not from a consistent outperformance over time.
3) If we are to expect rebalances to continue to outperform on the quarter, we would need to believe that the rebalances that caused the outperformance in the past were due to the proximity of events around the calendar quarter. For example, nearly half of the outperformance of calendar quarter rebalancing is from the Covid decline, crash and recovery. I personally find no compelling reason that this was aligned with the quarter and would instead attribute it to luck.
4) Therefore I would not assume calendar rebalancing to outperform in the future, unless a compelling reason can be found that some of the other large quarterly swings were due to events that were directly attributable to happening in close proximity to the change in the calendar quarter.
5) In backtests, daily rebalancing outperforms quarterly rebalancing with an assumed 0.1% transaction cost. To take this further, I found the best results throughout time would be to rebalance anytime there is a spread between TMF & UPRO > ~3%, which happens on about 1/3 of days. However, I do not know whether the transaction cost I used is a good estimate, nor do I know how movements throughout the trading day may affect these results. Proceed at your own risk.
6) If someone is adding money to their HFEA, I would recommend using it to bring their allocation back towards/to their target allocation.
7) At the end of the day, these questions of rebalance timing lead to relatively small changes in CAGR and are not imperative in the big picture to the success/failure of a HFEA strategy.

Happy to hear any additional/alternative interpretations or analyses.
Hydromod
Posts: 1051
Joined: Tue Mar 26, 2019 10:21 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Hydromod »

akxc wrote: Sun Jan 09, 2022 6:35 pm 6) If someone is adding money to their HFEA, I would recommend using it to bring their allocation back towards/to their target allocation.
7) At the end of the day, these questions of rebalance timing lead to relatively small changes in CAGR and are not imperative in the big picture to the success/failure of a HFEA strategy.

Happy to hear any additional/alternative interpretations or analyses.
Just to hammer home the point about small changes, I did a little analysis here looking at systematic effects of contribution date within a quarter and rebalance date within a quarter. I used UPRO and TMF daily values between 1986 and present (synthetic prior to inception dates).

Each contribution is actually a separate small bet on performance between the contribution date and rebalance date; the main portfolio performance is completely unaffected by the contribution allocation (although highly correlated, of course). The simple test just compared different UPRO/TMF allocations for the contributions, looking at every combination of starting day within the quarter for several possible rebalance days within the quarter.

Basically the performance was a microcosm of HFEA performance writ small. The differences in return increase with duration before rebalancing, and the variability in returns increases as the contribution becomes more dominated by either UPRO or TMF. There's some hint that performance in the middle of the month is affected, perhaps by treasury auctions.

At the end of the day, the results suggest that it's completely justifiable to make contributions at the HFEA allocation (whatever allocation is selected), or to partially rebalance with contributions, or even to simply add 100% UPRO (accepting a more variability from contribution to contribution for a bit higher expected returns). One might even add 100% UPRO vs 100% TMF depending on recent volatility. Most of the differences from allocation and contribution date will be just due to timing luck.

Similar conclusions appear to hold using TQQQ or EURL vs. TMF or SPY vs. TLT (not shown).

I didn't show it, but TYD doesn't seem to have the issues with middle-of-the-month performance. Or SPY/GLD. IMO it would have been a very justifiable approach (and perhaps optimal) to contribute at 100% equity, taking advantage of the short opportunity for growth, and rebalance into ITTs/gold for safety.

With all that said, these issues are far down the list in the big picture.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

akxc wrote: Sun Jan 09, 2022 6:35 pm Right, I am still in the thinking that most of the historic out-performance of quarterly rebalancing is based on timing luck. Clearly there has not been a consistent "premium", but instead the outperformance has been a few well-timed rebalances.
Maybe it's just imagination, but I cannot get the thought out of my mind that the curves in your diagram in viewtopic.php?p=6433845#p6433845 that splits by decade, tend to have a U-shape. Maybe the leveraged ETFs rebalanced by calendar quarters amplify the so-called "Santa Claus Rallye" and similar anomalies?
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

akxc wrote: Sun Jan 09, 2022 6:35 pm So to sum up my thoughts on a few issues that have been gone over lately in this thread:
1) Rebalance timing has a much bigger impact on end performance with daily-rebalanced leveraged ETFs than with non-leveraged strategies.
2) Historically HFEA has performed better with rebalances on the calendar quarter. This is due to a select few rebalances that created most of the outsized gains and not from a consistent outperformance over time.
3) If we are to expect rebalances to continue to outperform on the quarter, we would need to believe that the rebalances that caused the outperformance in the past were due to the proximity of events around the calendar quarter. For example, nearly half of the outperformance of calendar quarter rebalancing is from the Covid decline, crash and recovery. I personally find no compelling reason that this was aligned with the quarter and would instead attribute it to luck.
4) Therefore I would not assume calendar rebalancing to outperform in the future, unless a compelling reason can be found that some of the other large quarterly swings were due to events that were directly attributable to happening in close proximity to the change in the calendar quarter.
5) In backtests, daily rebalancing outperforms quarterly rebalancing with an assumed 0.1% transaction cost. To take this further, I found the best results throughout time would be to rebalance anytime there is a spread between TMF & UPRO > ~3%, which happens on about 1/3 of days. However, I do not know whether the transaction cost I used is a good estimate, nor do I know how movements throughout the trading day may affect these results. Proceed at your own risk.
6) If someone is adding money to their HFEA, I would recommend using it to bring their allocation back towards/to their target allocation.
7) At the end of the day, these questions of rebalance timing lead to relatively small changes in CAGR and are not imperative in the big picture to the success/failure of a HFEA strategy.

Happy to hear any additional/alternative interpretations or analyses.
This is a very nice summary. Thanks for sharing your analysis. What are your thoughts on rebalancing bands - from previous discussions and this conversation I gather "it doesn't matter" i.e. rebalancing bands or periodic (monthly, quarterly, etc.) - basically any period longer than a few days - will have no systematic advantage in the long run? Or can we say with rebalancing bands we can be sure to capture things like the Covid crisis without relying on timing luck?
User avatar
Nicolas Perrault
Posts: 244
Joined: Thu Sep 27, 2018 12:33 pm
Location: Oxford, UK

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Nicolas Perrault »

hillclimber wrote: Wed Jan 05, 2022 12:04 am If you can't handle rebalancing into an asset that is going down, you shouldn't be messing around with 3x leveraged etfs.
Sure I can agree with that.

In March 2020, however, this is not the issue that most Bogleheads were struggling with. Sure, history could have turned out different. But most Bogleheads (Hedgefundie and myself included) were actually too eager to rebalance too early into the asset that was going down (UPRO), especially when TMF was skyrocketing around 8 March. Many, many rebalanced in the first ten days of March, which turned out to be considerably inferior to rebalancing on schedule on 1 April. The optimal rebalancing date was probably 23 March, but 1 April was better than what many Bogleheads ended up doing. One argument of the people rebalancing around 8 March was that TMF had just gone up so fast that it was likely it would go down soon. UPRO had just cratered so it looked cheaper. So, they thought, might as well rebalance from TMF into UPRO. As it turned out, the "TMF is likely to go down soon" part was correct. The problem is that UPRO also went down even more and and it went down much more still than TMF. So by 1 April, those that had rebalanced into the down-asset around 8 March had less money than they would have had if they hadn't touched anything.


Perhaps we could rephrase your sentence as:
If you can't handle sticking to a leveraged strategy as everything craters, you shouldn't be messing around with 3x leveraged etfs.
User avatar
Afrofreak
Posts: 223
Joined: Sat Feb 20, 2021 2:09 pm
Location: Ontario, Canada

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Afrofreak »

comeinvest wrote: Mon Jan 10, 2022 4:02 am
akxc wrote: Sun Jan 09, 2022 6:35 pm Right, I am still in the thinking that most of the historic out-performance of quarterly rebalancing is based on timing luck. Clearly there has not been a consistent "premium", but instead the outperformance has been a few well-timed rebalances.
Maybe it's just imagination, but I cannot get the thought out of my mind that the curves in your diagram in viewtopic.php?p=6433845#p6433845 that splits by decade, tend to have a U-shape. Maybe the leveraged ETFs rebalanced by calendar quarters amplify the so-called "Santa Claus Rallye" and similar anomalies?
Yes, I'm inclined to agree, this doesn't just look like luck to me. Just because we do not yet know the underlying reason for why rebalancing at the beginning of calendar quarters is optimal doesn't mean it just comes down to a few well-timed rebalances. If anything I'd argue that the graph shows that this premium is getting bigger and bigger as time goes on, the differential gets larger and larger by the decade.
DMoogle
Posts: 549
Joined: Sat Oct 31, 2020 10:24 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DMoogle »

Nicolas Perrault wrote: Mon Jan 10, 2022 9:08 am
hillclimber wrote: Wed Jan 05, 2022 12:04 am If you can't handle rebalancing into an asset that is going down, you shouldn't be messing around with 3x leveraged etfs.
Sure I can agree with that.

In March 2020, however, this is not the issue that most Bogleheads were struggling with. Sure, history could have turned out different. But most Bogleheads (Hedgefundie and myself included) were actually too eager to rebalance too early into the asset that was going down (UPRO), especially when TMF was skyrocketing around 8 March. Many, many rebalanced in the first ten days of March, which turned out to be considerably inferior to rebalancing on schedule on 1 April. The optimal rebalancing date was probably 23 March, but 1 April was better than what many Bogleheads ended up doing. One argument of the people rebalancing around 8 March was that TMF had just gone up so fast that it was likely it would go down soon. UPRO had just cratered so it looked cheaper. So, they thought, might as well rebalance from TMF into UPRO. As it turned out, the "TMF is likely to go down soon" part was correct. The problem is that UPRO also went down even more and and it went down much more still than TMF. So by 1 April, those that had rebalanced into the down-asset around 8 March had less money than they would have had if they hadn't touched anything.


Perhaps we could rephrase your sentence as:
If you can't handle sticking to a leveraged strategy as everything craters, you shouldn't be messing around with 3x leveraged etfs.
Although I agree with your main point, I wouldn't put too much weight into a single data point of where rebalancing early didn't work well.
Hydromod
Posts: 1051
Joined: Tue Mar 26, 2019 10:21 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Hydromod »

Afrofreak wrote: Mon Jan 10, 2022 9:28 am
comeinvest wrote: Mon Jan 10, 2022 4:02 am
akxc wrote: Sun Jan 09, 2022 6:35 pm Right, I am still in the thinking that most of the historic out-performance of quarterly rebalancing is based on timing luck. Clearly there has not been a consistent "premium", but instead the outperformance has been a few well-timed rebalances.
Maybe it's just imagination, but I cannot get the thought out of my mind that the curves in your diagram in viewtopic.php?p=6433845#p6433845 that splits by decade, tend to have a U-shape. Maybe the leveraged ETFs rebalanced by calendar quarters amplify the so-called "Santa Claus Rallye" and similar anomalies?
Yes, I'm inclined to agree, this doesn't just look like luck to me. Just because we do not yet know the underlying reason for why rebalancing at the beginning of calendar quarters is optimal doesn't mean it just comes down to a few well-timed rebalances. If anything I'd argue that the graph shows that this premium is getting bigger and bigger as time goes on, the differential gets larger and larger by the decade.
30-year treasuries are auctioned in the second week of February, May, August, and November and 20-year treasuries are auctioned on the next to last Wednesday of those months (i.e., the middle of the quarter). I strongly suspect that these LTT auctions mess with rebalancing. In the little analysis I pointed to here, the contribution effects appear to be messed up for LTTs linked to the middle of the quarter. I didn't see a similar effect with ITTs or gold.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

Hydromod wrote: Mon Jan 10, 2022 10:30 am
Afrofreak wrote: Mon Jan 10, 2022 9:28 am
comeinvest wrote: Mon Jan 10, 2022 4:02 am
akxc wrote: Sun Jan 09, 2022 6:35 pm Right, I am still in the thinking that most of the historic out-performance of quarterly rebalancing is based on timing luck. Clearly there has not been a consistent "premium", but instead the outperformance has been a few well-timed rebalances.
Maybe it's just imagination, but I cannot get the thought out of my mind that the curves in your diagram in viewtopic.php?p=6433845#p6433845 that splits by decade, tend to have a U-shape. Maybe the leveraged ETFs rebalanced by calendar quarters amplify the so-called "Santa Claus Rallye" and similar anomalies?
Yes, I'm inclined to agree, this doesn't just look like luck to me. Just because we do not yet know the underlying reason for why rebalancing at the beginning of calendar quarters is optimal doesn't mean it just comes down to a few well-timed rebalances. If anything I'd argue that the graph shows that this premium is getting bigger and bigger as time goes on, the differential gets larger and larger by the decade.
30-year treasuries are auctioned in the second week of February, May, August, and November and 20-year treasuries are auctioned on the next to last Wednesday of those months (i.e., the middle of the quarter). I strongly suspect that these LTT auctions mess with rebalancing. In the little analysis I pointed to here, the contribution effects appear to be messed up for LTTs linked to the middle of the quarter. I didn't see a similar effect with ITTs or gold.
If the treasury auctions are in the middle of the month and in the middle of the calendar quarter boundaries, wouldn't by this theory the rebalancing, which is most often from equities to treasuries, be cheaper in the middle between month or quarter boundaries, assuming that new treasury supply would slightly cheapen the treasury market? The opposite of what we see.
Hydromod
Posts: 1051
Joined: Tue Mar 26, 2019 10:21 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Hydromod »

comeinvest wrote: Mon Jan 10, 2022 12:23 pm
Hydromod wrote: Mon Jan 10, 2022 10:30 am 30-year treasuries are auctioned in the second week of February, May, August, and November and 20-year treasuries are auctioned on the next to last Wednesday of those months (i.e., the middle of the quarter). I strongly suspect that these LTT auctions mess with rebalancing. In the little analysis I pointed to here, the contribution effects appear to be messed up for LTTs linked to the middle of the quarter. I didn't see a similar effect with ITTs or gold.
If the treasury auctions are in the middle of the month and in the middle of the calendar quarter boundaries, wouldn't by this theory the rebalancing, which is most often from equities to treasuries, be cheaper in the middle between month or quarter boundaries, assuming that new treasury supply would slightly cheapen the treasury market? The opposite of what we see.
The data is pretty noisy, but it looks like the return from buying day to the end of the quarter is roughly the same for every day in the first month. Similarly for every day in the last month. Most of the decay in return is over the middle month. Which is consistent with the treasury market getting cheaper over the middle month and holding steady in the first and third month, if I understand correctly.

I must admit that I don't completely understand what is going on.
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

Hydromod wrote: Mon Jan 10, 2022 1:21 pm
comeinvest wrote: Mon Jan 10, 2022 12:23 pm
Hydromod wrote: Mon Jan 10, 2022 10:30 am 30-year treasuries are auctioned in the second week of February, May, August, and November and 20-year treasuries are auctioned on the next to last Wednesday of those months (i.e., the middle of the quarter). I strongly suspect that these LTT auctions mess with rebalancing. In the little analysis I pointed to here, the contribution effects appear to be messed up for LTTs linked to the middle of the quarter. I didn't see a similar effect with ITTs or gold.
If the treasury auctions are in the middle of the month and in the middle of the calendar quarter boundaries, wouldn't by this theory the rebalancing, which is most often from equities to treasuries, be cheaper in the middle between month or quarter boundaries, assuming that new treasury supply would slightly cheapen the treasury market? The opposite of what we see.
The data is pretty noisy, but it looks like the return from buying day to the end of the quarter is roughly the same for every day in the first month. Similarly for every day in the last month. Most of the decay in return is over the middle month. Which is consistent with the treasury market getting cheaper over the middle month and holding steady in the first and third month, if I understand correctly.

I must admit that I don't completely understand what is going on.
Treasury market getting cheaper in the middle, would lead to higher returns, not lower, if we invest or rebalance into treasuries in the middle of the period.
Walkure
Posts: 1023
Joined: Tue Apr 11, 2017 9:59 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Walkure »

comeinvest wrote: Mon Jan 10, 2022 1:31 pm
Hydromod wrote: Mon Jan 10, 2022 1:21 pm
comeinvest wrote: Mon Jan 10, 2022 12:23 pm
Hydromod wrote: Mon Jan 10, 2022 10:30 am 30-year treasuries are auctioned in the second week of February, May, August, and November and 20-year treasuries are auctioned on the next to last Wednesday of those months (i.e., the middle of the quarter). I strongly suspect that these LTT auctions mess with rebalancing. In the little analysis I pointed to here, the contribution effects appear to be messed up for LTTs linked to the middle of the quarter. I didn't see a similar effect with ITTs or gold.
If the treasury auctions are in the middle of the month and in the middle of the calendar quarter boundaries, wouldn't by this theory the rebalancing, which is most often from equities to treasuries, be cheaper in the middle between month or quarter boundaries, assuming that new treasury supply would slightly cheapen the treasury market? The opposite of what we see.
The data is pretty noisy, but it looks like the return from buying day to the end of the quarter is roughly the same for every day in the first month. Similarly for every day in the last month. Most of the decay in return is over the middle month. Which is consistent with the treasury market getting cheaper over the middle month and holding steady in the first and third month, if I understand correctly.

I must admit that I don't completely understand what is going on.
Treasury market getting cheaper in the middle, would lead to higher returns, not lower, if we invest or rebalance into treasuries in the middle of the period.
If Treasuries have cyclical suppressed pricing not correlated to equities, would it suggest that there might one optimal timing for quarters when you are rebalancing out of stocks and into treasuries, and different optimal timing for quarters when you are rebalancing out of treasuries and into stocks?
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

Walkure wrote: Mon Jan 10, 2022 3:53 pm If Treasuries have cyclical suppressed pricing not correlated to equities, would it suggest that there might one optimal timing for quarters when you are rebalancing out of stocks and into treasuries, and different optimal timing for quarters when you are rebalancing out of treasuries and into stocks?
I assume that most rebalancing events would go net into treasuries, as equities had higher returns, especially during the last decade.
Walkure
Posts: 1023
Joined: Tue Apr 11, 2017 9:59 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Walkure »

comeinvest wrote: Mon Jan 10, 2022 5:05 pm
Walkure wrote: Mon Jan 10, 2022 3:53 pm If Treasuries have cyclical suppressed pricing not correlated to equities, would it suggest that there might one optimal timing for quarters when you are rebalancing out of stocks and into treasuries, and different optimal timing for quarters when you are rebalancing out of treasuries and into stocks?
I assume that most rebalancing events would go net into treasuries, as equities had higher returns, especially during the last decade.
Right, most of the time UPRO is doing the lifting and TMF is just around for insurance. I guess I'm suggesting something along a strategy of "rebalance mid-quarter, unless stocks are crashing, in which case you should hold off until the end of quarter."
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

Walkure wrote: Mon Jan 10, 2022 5:34 pm
comeinvest wrote: Mon Jan 10, 2022 5:05 pm
Walkure wrote: Mon Jan 10, 2022 3:53 pm If Treasuries have cyclical suppressed pricing not correlated to equities, would it suggest that there might one optimal timing for quarters when you are rebalancing out of stocks and into treasuries, and different optimal timing for quarters when you are rebalancing out of treasuries and into stocks?
I assume that most rebalancing events would go net into treasuries, as equities had higher returns, especially during the last decade.
Right, most of the time UPRO is doing the lifting and TMF is just around for insurance. I guess I'm suggesting something along a strategy of "rebalance mid-quarter, unless stocks are crashing, in which case you should hold off until the end of quarter."
Let's not get ahead of ourselves before we even have any explanation for the potential anomaly. The suggested explanation with treasury auctions schedules seems to explain the opposite of what we observed, if I'm not mistaken. I personally use not LETFs but futures anyways, where the natural rebalancing is at the time of futures expiration. I'm probably going to stick with this schedule, until I have more insight or evidence.
DarkMatter731
Posts: 224
Joined: Sat Apr 25, 2020 8:28 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DarkMatter731 »

I've got to say I do find it slightly interesting that market timing is seen as so controversial when the above discussion essentially boils down to market timing.

I think you guys might be reading too much into things. A few outsized returns could heavily influence your overall return and the timing of the rebalance just happens to catch that.
akxc
Posts: 19
Joined: Thu Jan 06, 2022 5:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by akxc »

DarkMatter731 wrote: Mon Jan 10, 2022 8:59 pm I've got to say I do find it slightly interesting that market timing is seen as so controversial when the above discussion essentially boils down to market timing.

I think you guys might be reading too much into things. A few outsized returns could heavily influence your overall return and the timing of the rebalance just happens to catch that.
I don't see how this discussion boils down to market timing? I think we are trying to look more at and understand the rebalance premium which is an integral part of a risk-parity strategy like HFEA. And yes, clearly a few outsized returns influence your overall return based on the analysis a few posts ago about the difference in performance that a one month lag in rebalancing creates, so if anything part of what we are looking at is trying to REFUTE market timing.
DarkMatter731
Posts: 224
Joined: Sat Apr 25, 2020 8:28 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DarkMatter731 »

akxc wrote: Mon Jan 10, 2022 9:42 pm
DarkMatter731 wrote: Mon Jan 10, 2022 8:59 pm I've got to say I do find it slightly interesting that market timing is seen as so controversial when the above discussion essentially boils down to market timing.

I think you guys might be reading too much into things. A few outsized returns could heavily influence your overall return and the timing of the rebalance just happens to catch that.
I don't see how this discussion boils down to market timing? I think we are trying to look more at and understand the rebalance premium which is an integral part of a risk-parity strategy like HFEA. And yes, clearly a few outsized returns influence your overall return based on the analysis a few posts ago about the difference in performance that a one month lag in rebalancing creates, so if anything part of what we are looking at is trying to REFUTE market timing.
Trying to optimize the timing of the rebalance is still an example of market timing, is it not?

I fail to see why there's a rebalancing premium outside of the benefits of diversification itself. AQR management wrote an article on 'rebalancing premium' a few years ago and they argued along the same lines.
akxc
Posts: 19
Joined: Thu Jan 06, 2022 5:20 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by akxc »

DarkMatter731 wrote: Mon Jan 10, 2022 9:54 pm Trying to optimize the timing of the rebalance is still an example of market timing, is it not?

I fail to see why there's a rebalancing premium outside of the benefits of diversification itself. AQR management wrote an article on 'rebalancing premium' a few years ago and they argued along the same lines.
I mean yes, the rebalancing premium is really about managing risk, which is a very important part of a leveraged strategy like HFEA. Part of managing risk is calculating a return.

From my understanding the "don't market time" refrain stems from the generally accepted finding that time in the market beats timing the market, and that future predictions about the market are generally shown to not be sources of alpha. "Timing" is an integral part of a risk-parity strategy like HFEA, and everyone engages in it when they rebalance their portfolio, with the majority choosing to default to the recommendation of quarterly timing. I believe a systematic mathematical examination of timing rebalances differs from traditional "market timing", which I would argue revolves more around decisions like "I'm going to reduce my bond allocation because the Fed is raising rates", "I'm going to reduce my equity allocation because markets are at all-time highs" or "I'm going to rebalance on the calendar quarter, because lots of events happen around the time of the change in the calendar quarter".

I try to attempt to understand things I invest in. And as someone who believes generally in the EMH, I think examinations like this can help give insights into the type of risks you are potentially being compensated for in various investment strategies. For example, I've heard a common refrain that you should quarterly rebalance because of "momentum". So is HFEA a momentum strategy? I've found that daily rebalancing leads to slightly higher risk adjusted returns by some metrics than quarterly rebalancing. Of course, daily rebalancing results in higher transaction costs, so quarterly rebalancing is likely to be favorable to daily rebalancing in most instances. But this tells me that quarterly rebalancing is perhaps more about transaction costs than momentum. Which helps provide me with at least a little more enlightenment on how HFEA / risk-parity strategies work.
User avatar
Afrofreak
Posts: 223
Joined: Sat Feb 20, 2021 2:09 pm
Location: Ontario, Canada

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Afrofreak »

I noticed recently that in March 2020 when the markets were in turmoil, the authorized participants seemed unable to maintain the 3x daily leverage both positively and negatively. On March 16, $TLT gained 6.48% and we would have expected $TMF to gain 3x as much, or 19.44%, instead we only got 16.76%. Similarly, $QQQ dropped 11.98% and we would have expected $TQQQ to drop 35.94% rather than only 34.47%. Is this cause for concern and why is this happening? If you think about it, you might actually want the leverage to concave like that so your equity portion is somewhat protected, but I suspect that we can observe this same phenomenon when $QQQ has a large upswing following a crash too.
DMoogle
Posts: 549
Joined: Sat Oct 31, 2020 10:24 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DMoogle »

Afrofreak wrote: Tue Jan 11, 2022 10:12 pm I noticed recently that in March 2020 when the markets were in turmoil, the authorized participants seemed unable to maintain the 3x daily leverage both positively and negatively. On March 16, $TLT gained 6.48% and we would have expected $TMF to gain 3x as much, or 19.44%, instead we only got 16.76%. Similarly, $QQQ dropped 11.98% and we would have expected $TQQQ to drop 35.94% rather than only 34.47%. Is this cause for concern and why is this happening? If you think about it, you might actually want the leverage to concave like that so your equity portion is somewhat protected, but I suspect that we can observe this same phenomenon when $QQQ has a large upswing following a crash too.
Not sure of the "why" here, but worth noting that the folks on here have come to the conclusion that Proshares (UPRO/TQQQ) is MUCH better at actually achieving their stated objective than Direxion is (TMF).

Since the delta is regularly in both directions, I don't think it's really too much cause for concern.
DarkMatter731
Posts: 224
Joined: Sat Apr 25, 2020 8:28 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DarkMatter731 »

Afrofreak wrote: Tue Jan 11, 2022 10:12 pm I noticed recently that in March 2020 when the markets were in turmoil, the authorized participants seemed unable to maintain the 3x daily leverage both positively and negatively. On March 16, $TLT gained 6.48% and we would have expected $TMF to gain 3x as much, or 19.44%, instead we only got 16.76%. Similarly, $QQQ dropped 11.98% and we would have expected $TQQQ to drop 35.94% rather than only 34.47%. Is this cause for concern and why is this happening? If you think about it, you might actually want the leverage to concave like that so your equity portion is somewhat protected, but I suspect that we can observe this same phenomenon when $QQQ has a large upswing following a crash too.
Leverage drifts on a daily basis.

It then resets back to 3x but your 3x ETF won't actually return 3x the daily return.
vimpindare
Posts: 9
Joined: Mon Aug 03, 2020 4:21 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by vimpindare »

OptimalFI wrote: Wed Nov 03, 2021 11:22 am A lot of the discussion here about UPRO seems to ignore the fact that it's not only a levered product, but also leverage with a daily rebalance. This makes a huge difference, and leverage with a daily rebalance does not give remotely close to the same long-term return as a long-term levered bet on SPX.

For example, SPY closed around 321 on Jan 31, 2020, and recovered to that level around July 15th. On Jan 31 2020 UPRO closed around 69,and on July 15 2020 it was priced around 50. A portfolio which borrowed on margin to buy 3X SPY would be roughly flat, but a portfolio which was long UPRO was down 30%. Some of this is leverage drag (e.g. https://www.etf.com/etf-education-cente ... nopaging=1). I cherry picked dates, but this will generally happen when volatility is high. The point here is that the return profile of a 3x levered product with daily reset is very different from a levered bet on long-term returns (e.g. buying long-dated futures or borrowing to buy SPY). Meaning, there can be a long period of time where the asset on which UPRO is seeking levered exposure (SPX) is flat, but UPRO itself has a substantially non-zero return.

Also, in terms of performance characteristics, products like UPRO have (1) relatively high expense ratios (2) underlying swaps with embedded financing costs (3) substantial turnover, especially in volatile markets, with associated transaction costs.

Given this, I feel like the suggestion made by cos
Why not switch to portfolio margin (in taxable) and futures (in tax-advantaged)
might be worth considering for investors seeking leverage.
Ultimately, the tradeoff in using daily reset vs monthly (for example) is that you get better performance in trending markets (both up and down), but worse performance in choppy markets. The choice of leverage reset period is really a bet on forward volatility adjusted performance. If you have a good model for predicting forward volatility you could dynamically adjust the leverage reset period to be longer/shorter depending on fwd vol.

A daily reset also prevents wipeouts.
User avatar
hillclimber
Posts: 136
Joined: Wed Mar 10, 2021 2:32 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by hillclimber »

DMoogle wrote: Tue Jan 11, 2022 11:28 pm
Afrofreak wrote: Tue Jan 11, 2022 10:12 pm I noticed recently that in March 2020 when the markets were in turmoil, the authorized participants seemed unable to maintain the 3x daily leverage both positively and negatively. On March 16, $TLT gained 6.48% and we would have expected $TMF to gain 3x as much, or 19.44%, instead we only got 16.76%. Similarly, $QQQ dropped 11.98% and we would have expected $TQQQ to drop 35.94% rather than only 34.47%. Is this cause for concern and why is this happening? If you think about it, you might actually want the leverage to concave like that so your equity portion is somewhat protected, but I suspect that we can observe this same phenomenon when $QQQ has a large upswing following a crash too.
Not sure of the "why" here, but worth noting that the folks on here have come to the conclusion that Proshares (UPRO/TQQQ) is MUCH better at actually achieving their stated objective than Direxion is (TMF).

Since the delta is regularly in both directions, I don't think it's really too much cause for concern.
If you compare the Proshares 2x LTT product holdings, to the Direxion 3x LTT product, you'll see that the Proshares one has futures in its daily holdings. I think that Proshares uses futures to smooth out its exposure.

This might also explain some of the difficulties with international leveraged etfs. EFO doesn't have EAFE futures to work with, so it's less efficient.
DarkMatter731
Posts: 224
Joined: Sat Apr 25, 2020 8:28 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by DarkMatter731 »

hillclimber wrote: Wed Jan 12, 2022 8:53 pm
DMoogle wrote: Tue Jan 11, 2022 11:28 pm
Afrofreak wrote: Tue Jan 11, 2022 10:12 pm I noticed recently that in March 2020 when the markets were in turmoil, the authorized participants seemed unable to maintain the 3x daily leverage both positively and negatively. On March 16, $TLT gained 6.48% and we would have expected $TMF to gain 3x as much, or 19.44%, instead we only got 16.76%. Similarly, $QQQ dropped 11.98% and we would have expected $TQQQ to drop 35.94% rather than only 34.47%. Is this cause for concern and why is this happening? If you think about it, you might actually want the leverage to concave like that so your equity portion is somewhat protected, but I suspect that we can observe this same phenomenon when $QQQ has a large upswing following a crash too.
Not sure of the "why" here, but worth noting that the folks on here have come to the conclusion that Proshares (UPRO/TQQQ) is MUCH better at actually achieving their stated objective than Direxion is (TMF).

Since the delta is regularly in both directions, I don't think it's really too much cause for concern.
If you compare the Proshares 2x LTT product holdings, to the Direxion 3x LTT product, you'll see that the Proshares one has futures in its daily holdings. I think that Proshares uses futures to smooth out its exposure.

This might also explain some of the difficulties with international leveraged etfs. EFO doesn't have EAFE futures to work with, so it's less efficient.
You've hit the nail on the head.

UPRO and TQQQ use futures and swaps as laid out in their prospectus. It's one of the reasons I decided to just use futures myself - why do I need to pay a middle-man when I can get more leverage myself and adjust my leverage when I need to rather than them adjusting their exposure on a daily basis.

https://www.proshares.com/globalassets/ ... pectus.pdf
comeinvest
Posts: 2709
Joined: Mon Mar 12, 2012 6:57 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by comeinvest »

hillclimber wrote: Wed Jan 12, 2022 8:53 pm
DMoogle wrote: Tue Jan 11, 2022 11:28 pm
Afrofreak wrote: Tue Jan 11, 2022 10:12 pm I noticed recently that in March 2020 when the markets were in turmoil, the authorized participants seemed unable to maintain the 3x daily leverage both positively and negatively. On March 16, $TLT gained 6.48% and we would have expected $TMF to gain 3x as much, or 19.44%, instead we only got 16.76%. Similarly, $QQQ dropped 11.98% and we would have expected $TQQQ to drop 35.94% rather than only 34.47%. Is this cause for concern and why is this happening? If you think about it, you might actually want the leverage to concave like that so your equity portion is somewhat protected, but I suspect that we can observe this same phenomenon when $QQQ has a large upswing following a crash too.
Not sure of the "why" here, but worth noting that the folks on here have come to the conclusion that Proshares (UPRO/TQQQ) is MUCH better at actually achieving their stated objective than Direxion is (TMF).

Since the delta is regularly in both directions, I don't think it's really too much cause for concern.
If you compare the Proshares 2x LTT product holdings, to the Direxion 3x LTT product, you'll see that the Proshares one has futures in its daily holdings. I think that Proshares uses futures to smooth out its exposure.

This might also explain some of the difficulties with international leveraged etfs. EFO doesn't have EAFE futures to work with, so it's less efficient.
They could use MSCI EAFE futures if they wanted https://www.theice.com/products/3119684 ... dex-Future
corpgator
Posts: 217
Joined: Wed Jul 22, 2015 12:00 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by corpgator »

UPRO and TQQQ split yesterday at close. Took me a minute to figure out why my options were suddenly worth 10x - unfortunately they won't be tomorrow.
User avatar
typical.investor
Posts: 5263
Joined: Mon Jun 11, 2018 3:17 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by typical.investor »

corpgator wrote: Thu Jan 13, 2022 3:48 am UPRO and TQQQ split yesterday at close. Took me a minute to figure out why my options were suddenly worth 10x - unfortunately they won't be tomorrow.
Thanks for posting that. The ETF is down 50% and I was like *WHAT*?!?!?!?
kbourgu
Posts: 23
Joined: Tue Sep 28, 2021 10:05 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by kbourgu »

typical.investor wrote: Thu Jan 13, 2022 4:08 am
corpgator wrote: Thu Jan 13, 2022 3:48 am UPRO and TQQQ split yesterday at close. Took me a minute to figure out why my options were suddenly worth 10x - unfortunately they won't be tomorrow.
Thanks for posting that. The ETF is down 50% and I was like *WHAT*?!?!?!?
Does anyone know when things will normalize with these stock splits in M1 Finance? Because right now it's displaying as if my accounts crashed.
Kbg
Posts: 355
Joined: Thu Mar 23, 2017 11:33 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Kbg »

kbourgu wrote: Thu Jan 13, 2022 8:39 am
typical.investor wrote: Thu Jan 13, 2022 4:08 am
corpgator wrote: Thu Jan 13, 2022 3:48 am UPRO and TQQQ split yesterday at close. Took me a minute to figure out why my options were suddenly worth 10x - unfortunately they won't be tomorrow.
Thanks for posting that. The ETF is down 50% and I was like *WHAT*?!?!?!?
Does anyone know when things will normalize with these stock splits in M1 Finance? Because right now it's displaying as if my accounts crashed.
Likely NLT tomorrow.
Centurion
Posts: 29
Joined: Wed Apr 08, 2020 10:49 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Centurion »

Oh wow - first time I have split for a position in my portfolio. For somewhat volatile assets like UPRO this is something to watch out for in the future I guess, as it seems it takes some time (depending on your broker of course) until the new amount is reflected in the portfolio. In the meantime it would not be possible to trade and I also wonder what would happen to pre-existing stop loss orders? :shock:
huzaing
Posts: 292
Joined: Mon May 03, 2021 2:59 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by huzaing »

I was reading some discussions on quarterly rebalancing and the recommendation seems to be the first trading day of Jan, April, July, October.
Does the time of day matter at all, like morning or afternoon?

Thanks.
Post Reply