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

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Locked
User avatar
privatefarmer
Posts: 598
Joined: Mon Sep 08, 2014 2:45 pm

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

Post by privatefarmer » Fri Aug 02, 2019 9:26 pm

MotoTrojan wrote:
Fri Aug 02, 2019 9:22 pm
privatefarmer wrote:
Fri Aug 02, 2019 9:06 pm
snoopdoug1 wrote:
Fri Aug 02, 2019 7:26 pm
Does M1 automatically re balance for you? The risk / volatility method is confusing to me.
Seriously. I’ve been following this thread since inception and still can’t wrap my head around this “volatility targeting” strategy. So you basically rebalance at the end of each month, using the 20-day volatility of UPRO/TMF with the aim of having an equal amount of volatility from the two? So if UPRO had a volatility of 10% and TMF 5%, then you’d have essentially a 33/66 UPRO/TMF allocation?
To be clear, there are two methods:

Risk-parity w/ a look-back period (1 month often) would be 33/66 if UPRO was 10% and TMF 5%, for the past month, correct. Often it would be a much higher allocation to UPRO as well. This method backtests well, but most of the research actually favors specifically looking at volatility on the equity side. I have seen some research showing that bonds actually perform better during high volatility periods.

Volatility targeting is picking a target volatility for your equity holding, and resetting your allocation with a look-back period (usually equal to the rebalance period, and frequently discussed here as 1 month) to achieve that target volatility based on the past month. If your target was 9% and UPRO was 10%, then you'd just be 100% UPRO. If your target was 5%, then you'd be 50/50 UPRO/TMF. TMF has no impact on the decision. Numbers ranging from about 15-25% for targets have been explored here but I have found 20% to be the sweetspot.

This thread discusses it further. It works well for people who want 100/0 equity type returns, but with less drawdowns. It also historically worked well with this strategy, but depending on your target allocation you better be prepared for some 100% UPRO months; because of this some of us are implementing a cap such as 80% on UPRO for those months where it would be higher.

viewtopic.php?f=10&t=281691
Okay thanks! That makes sense. I’m still trying to figure out the math. Let’s say I wanted 20% annualized volatility. Where would I find last months volatility for UPRO and how would I use that number to come up with 20% annualized? Or do you simply use the tool on PV and plug in “20%” as your targeted volatility?

MotoTrojan
Posts: 6474
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan » Fri Aug 02, 2019 9:27 pm

privatefarmer wrote:
Fri Aug 02, 2019 9:24 pm
So did you guys use the “Target volatility” tool on PV to do this? You basically plug in UPRO as your asset and use TMF instead of cash for the “out of market asset”? Set the volatility to whatever annual std deviation you desire, And then run the test, click on “timing periods” and then scroll down to the bottom where it will tell you what your allocation for August 2019 should be?

I did this using a targeted 18% volatility and it’s telling me I should basically be 70/30 UPRO/TMF for the month of August?
Sounds about right, yup. Your choice of TMF shouldn't matter for seeing what the allocation needs to be. The backtest is very impressive with >20% CAGR compared to 17% for the OPs strategy and something like 18.5% for the risk-parity 1-month looback, but who knows what the future holds. Both of the lookback methods also drastically reduced the drawdown during the GFC.

samsdad
Posts: 744
Joined: Sat Jan 02, 2016 6:20 pm

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

Post by samsdad » Fri Aug 02, 2019 9:28 pm

privatefarmer wrote:
Fri Aug 02, 2019 9:24 pm
So did you guys use the “Target volatility” tool on PV to do this? You basically plug in UPRO as your asset and use TMF instead of cash for the “out of market asset”? Set the volatility to whatever annual std deviation you desire, And then run the test, click on “timing periods” and then scroll down to the bottom where it will tell you what your allocation for August 2019 should be?

I did this using a targeted 18% volatility and it’s telling me I should basically be 70/30 UPRO/TMF for the month of August?
Yes. Though I still don’t understand how I’m supposed to model/backtest when I don’t want to go above, say, 80% equities.

MotoTrojan
Posts: 6474
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan » Fri Aug 02, 2019 9:29 pm

privatefarmer wrote:
Fri Aug 02, 2019 9:26 pm
MotoTrojan wrote:
Fri Aug 02, 2019 9:22 pm
privatefarmer wrote:
Fri Aug 02, 2019 9:06 pm
snoopdoug1 wrote:
Fri Aug 02, 2019 7:26 pm
Does M1 automatically re balance for you? The risk / volatility method is confusing to me.
Seriously. I’ve been following this thread since inception and still can’t wrap my head around this “volatility targeting” strategy. So you basically rebalance at the end of each month, using the 20-day volatility of UPRO/TMF with the aim of having an equal amount of volatility from the two? So if UPRO had a volatility of 10% and TMF 5%, then you’d have essentially a 33/66 UPRO/TMF allocation?
To be clear, there are two methods:

Risk-parity w/ a look-back period (1 month often) would be 33/66 if UPRO was 10% and TMF 5%, for the past month, correct. Often it would be a much higher allocation to UPRO as well. This method backtests well, but most of the research actually favors specifically looking at volatility on the equity side. I have seen some research showing that bonds actually perform better during high volatility periods.

Volatility targeting is picking a target volatility for your equity holding, and resetting your allocation with a look-back period (usually equal to the rebalance period, and frequently discussed here as 1 month) to achieve that target volatility based on the past month. If your target was 9% and UPRO was 10%, then you'd just be 100% UPRO. If your target was 5%, then you'd be 50/50 UPRO/TMF. TMF has no impact on the decision. Numbers ranging from about 15-25% for targets have been explored here but I have found 20% to be the sweetspot.

This thread discusses it further. It works well for people who want 100/0 equity type returns, but with less drawdowns. It also historically worked well with this strategy, but depending on your target allocation you better be prepared for some 100% UPRO months; because of this some of us are implementing a cap such as 80% on UPRO for those months where it would be higher.

viewtopic.php?f=10&t=281691
Okay thanks! That makes sense. I’m still trying to figure out the math. Let’s say I wanted 20% annualized volatility. Where would I find last months volatility for UPRO and how would I use that number to come up with 20% annualized? Or do you simply use the tool on PV and plug in “20%” as your targeted volatility?
A few posts back someone posted some example equations for finding the volatility in the last 20 trading days but that is quite simple. From there you simply take your target/realized and that is your UPRO allocation. PV also does this and will show you the volatility as well as the allocation based on that simple ratio.

MotoTrojan
Posts: 6474
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan » Fri Aug 02, 2019 9:31 pm

samsdad wrote:
Fri Aug 02, 2019 9:28 pm
privatefarmer wrote:
Fri Aug 02, 2019 9:24 pm
So did you guys use the “Target volatility” tool on PV to do this? You basically plug in UPRO as your asset and use TMF instead of cash for the “out of market asset”? Set the volatility to whatever annual std deviation you desire, And then run the test, click on “timing periods” and then scroll down to the bottom where it will tell you what your allocation for August 2019 should be?

I did this using a targeted 18% volatility and it’s telling me I should basically be 70/30 UPRO/TMF for the month of August?
Yes. Though I still don’t understand how I’m supposed to model/backtest when I don’t want to go above, say, 80% equities.
Backtesting should be a tool but I would not rely on it for a decision like that which is more behavioral than anything.

no simpler
Posts: 71
Joined: Sat Jul 13, 2019 4:54 pm

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

Post by no simpler » Fri Aug 02, 2019 9:40 pm

Forester wrote:
Fri Aug 02, 2019 6:39 pm
The leveraged bond ETF doesn't appear to add much value (?) vs TLT https://www.portfoliovisualizer.com/bac ... total3=100

The concept looks very promising, wondering what the catch is :D
Check the correlation to US market on the two strategies. The above link is only during this mega bull run.

samsdad
Posts: 744
Joined: Sat Jan 02, 2016 6:20 pm

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

Post by samsdad » Fri Aug 02, 2019 9:54 pm

MotoTrojan wrote:
Fri Aug 02, 2019 9:31 pm
samsdad wrote:
Fri Aug 02, 2019 9:28 pm
privatefarmer wrote:
Fri Aug 02, 2019 9:24 pm
So did you guys use the “Target volatility” tool on PV to do this? You basically plug in UPRO as your asset and use TMF instead of cash for the “out of market asset”? Set the volatility to whatever annual std deviation you desire, And then run the test, click on “timing periods” and then scroll down to the bottom where it will tell you what your allocation for August 2019 should be?

I did this using a targeted 18% volatility and it’s telling me I should basically be 70/30 UPRO/TMF for the month of August?
Yes. Though I still don’t understand how I’m supposed to model/backtest when I don’t want to go above, say, 80% equities.
Backtesting should be a tool but I would not rely on it for a decision like that which is more behavioral than anything.
I agree to a point. But if I could see whether an 80/20 upper limit and 20/80 lower bound didn’t do much better, or worse, worse than a 40/60 or even up to a 60/40 quarterly rebalancing then I’d dismiss it as just an exercise in “feeding the birds” (paying monthly commissions over at Fidelity) for no real benefit. Right now, I’ve modeled an 80/20 UPRO/TMF asset as “in market” with a 20/80 monthly rebalanced “out of market” asset and while it beat the OP strategy and 60/40 UPRO/TMF it wasn’t till 2012ish that it pulled away (when starting in 1987).

In other words, I’m worried that the self-imposed caps that I’d implement might suck out any of the perceived benefit over the quarterly-rebalancing path I’m currently on.

I PMed PVguy about the issue a few days ago but haven’t heard back as to whether my methodology was a correct implementation of the TV tool or not. If you recall, you said it wasn’t, so that’s why I’m currently “vacillating on volatility targeting.” <—say that 10 times fast :o

Hydromod
Posts: 232
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod » Fri Aug 02, 2019 10:03 pm

MotoTrojan wrote:
Fri Aug 02, 2019 9:22 pm
privatefarmer wrote:
Fri Aug 02, 2019 9:06 pm
snoopdoug1 wrote:
Fri Aug 02, 2019 7:26 pm
Does M1 automatically re balance for you? The risk / volatility method is confusing to me.
Seriously. I’ve been following this thread since inception and still can’t wrap my head around this “volatility targeting” strategy. So you basically rebalance at the end of each month, using the 20-day volatility of UPRO/TMF with the aim of having an equal amount of volatility from the two? So if UPRO had a volatility of 10% and TMF 5%, then you’d have essentially a 33/66 UPRO/TMF allocation?
To be clear, there are two methods:

Risk-parity w/ a look-back period (1 month often) would be 33/66 if UPRO was 10% and TMF 5%, for the past month, correct. Often it would be a much higher allocation to UPRO as well. This method backtests well, but most of the research actually favors specifically looking at volatility on the equity side. I have seen some research showing that bonds actually perform better during high volatility periods.

Volatility targeting is picking a target volatility for your equity holding, and resetting your allocation with a look-back period (usually equal to the rebalance period, and frequently discussed here as 1 month) to achieve that target volatility based on the past month. If your target was 9% and UPRO was 10%, then you'd just be 100% UPRO. If your target was 5%, then you'd be 50/50 UPRO/TMF. TMF has no impact on the decision. Numbers ranging from about 15-25% for targets have been explored here but I have found 20% to be the sweetspot.

This thread discusses it further. It works well for people who want 100/0 equity type returns, but with less drawdowns. It also historically worked well with this strategy, but depending on your target allocation you better be prepared for some 100% UPRO months; because of this some of us are implementing a cap such as 80% on UPRO for those months where it would be higher.

viewtopic.php?f=10&t=281691
I tried the volatility targeting approach as you described here. I didn't have much luck with getting reproducibly good results for my back test from 1987 to present. I was surprised by that. I got some improvement when I accounted for tmf volatility as well, and matched total variance. Intuitively that made sense, because the volatility targeting approach is designed for a riskless asset.

YoungSisyphus
Posts: 29
Joined: Mon Sep 24, 2018 7:35 am

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

Post by YoungSisyphus » Sat Aug 03, 2019 10:39 am

I'm sorry, I may be dense, but I'm having a hard time on the mechanics of the strategy that sounds interesting:

Taking the original 60/40 split and applying:
1. 20-day look back volatility
2. Re-balancing monthly
3. Capping UPRO to 80% regardless of #1

I am having difficult with a plug and play of actually doing this. I think the closest I saw was a manual excel download from Yahoo Finance of returns on UPRO and applying a STDEV / SQRT function to 20 days of adjusted close returns on UPRO?

Is the 20 day look back dependent on #2 - re-balancing monthly? E.g. If you re-balance on the 1st of every month, take the 20 days prior to the first of the current in-month?

Which means you would have to continually re-download the adjusted close data and adjust formula for STDEV(20-days) -> SQRT of STDEV(20-days)*annual number of trading days (252) = % annualized volatility * (your UPRO weighting adjustment)?

I feel rather ridiculous investing in something I can't even do the mechanics on correctly (much less fully understand the underlying principles) - but it is fun money and I'm having fun learning... :mrgreen:

Edit2: And as an output of all that - then it would be a simple SUM(UPRO + TMF) * new monthly UPRO rate where you would sell UPRO or TMF to reach desired ratio if you were not doing additional contributions?

Topic Author
HEDGEFUNDIE
Posts: 3527
Joined: Sun Oct 22, 2017 2:06 pm

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

Post by HEDGEFUNDIE » Sat Aug 03, 2019 11:01 am

Six months and 3000 posts later, the Excellent Adventure is still going strong!

At this rate we could probably launch our own annual Conference. Speakers from Direxion and ProShares?

schismal
Posts: 135
Joined: Sat Apr 13, 2019 8:53 pm

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

Post by schismal » Sat Aug 03, 2019 11:51 am

GOOGLE SHEETS LINK

So I pulled the data for 20% absolute volatility (20 day lookback) and compared the returns with (1) no asset capping, (2) a 60%, 70%, and 80% cap of either asset, and (3) the quarterly rebalanced 40/60 OP strategy. For $10k starting balance:

Code: Select all

OP strategy:      $108,841.98 (SD 21.09%)

20% vol (no cap): $172,704.57 (SD 23.72%)
60% cap:          $129,126.64 (SD 19.32%)
70% cap:          $146,463.61 (SD 20.26%)
80% cap:          $172,559.92 (SD 22.19%)
Image

User avatar
privatefarmer
Posts: 598
Joined: Mon Sep 08, 2014 2:45 pm

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

Post by privatefarmer » Sat Aug 03, 2019 2:32 pm

schismal wrote:
Sat Aug 03, 2019 11:51 am
GOOGLE SHEETS LINK

So I pulled the data for 20% absolute volatility (20 day lookback) and compared the returns with (1) no asset capping, (2) a 60%, 70%, and 80% cap of either asset, and (3) the quarterly rebalanced 40/60 OP strategy. For $10k starting balance:

Code: Select all

OP strategy:      $108,841.98 (SD 21.09%)

20% vol (no cap): $172,704.57 (SD 23.72%)
60% cap:          $129,126.64 (SD 19.32%)
70% cap:          $146,463.61 (SD 20.26%)
80% cap:          $172,559.92 (SD 22.19%)
Image
Okay, but this only got back to 2010 which is fairly meaningless. Obviously, any strategy that over weighted UPRO the last 10 years would’ve done fantastic.

My question because I am too dumb to figure out to how to run the backtests : what would your max DD have been (specifically during the GFC) using just UPRO/TMF “20-day volatility” look back

1) as a risk parity approach (having equal amounts of volatility each month from the two assets)

2) as a vol-targeting approach, let’s say 20% since that is roughly what a 100% SCV portfolio would have.

I might caveat this by saying that an UNleveraged risk parity portfolio (40/50/10 stocks/LTT/gold) would have had a max Dd of about 20% with roughly a 10% CAGR, to which you could lever up to your hearts content.

schismal
Posts: 135
Joined: Sat Apr 13, 2019 8:53 pm

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

Post by schismal » Sat Aug 03, 2019 4:26 pm

privatefarmer wrote:
Sat Aug 03, 2019 2:32 pm
Okay, but this only got back to 2010 which is fairly meaningless. Obviously, any strategy that over weighted UPRO the last 10 years would’ve done fantastic.
I think you missed the point a tad.

If someone wants to attempt a targeted volatility approach (given superior historical performance compared to the risk parity approach in the OP), then they must recognize that there is increasing risk as either asset moves towards 0%. A 100% UPRO month, for instance, is very risky if UPRO tanks while you're not holding an anti-correlated asset. That's why some folks are interested in capping UPRO and/or TMF at a certain percentage. I was interested in measuring the risk/return tradeoff of those capped portfolios compared to the uncapped volatility method (and for comparison, the OP's risk parity scheme).

I'd love to look at the simulated data that goes back farther, but I'm not sure where folks have gotten that.

samsdad
Posts: 744
Joined: Sat Jan 02, 2016 6:20 pm

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

Post by samsdad » Sat Aug 03, 2019 4:32 pm

schismal wrote:
Sat Aug 03, 2019 4:26 pm
I'd love to look at the simulated data that goes back farther, but I'm not sure where folks have gotten that.
It’s in the original post. See the blue hyperlink about half-way through the post that says “here.”

Also, I haven’t looked at your data yet. If you could include the equations that you’re using, that would be appreciated.

Hydromod
Posts: 232
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod » Sat Aug 03, 2019 4:44 pm

privatefarmer wrote:
Sat Aug 03, 2019 2:32 pm

Okay, but this only got back to 2010 which is fairly meaningless. Obviously, any strategy that over weighted UPRO the last 10 years would’ve done fantastic.

My question because I am too dumb to figure out to how to run the backtests : what would your max DD have been (specifically during the GFC) using just UPRO/TMF “20-day volatility” look back

1) as a risk parity approach (having equal amounts of volatility each month from the two assets)

2) as a vol-targeting approach, let’s say 20% since that is roughly what a 100% SCV portfolio would have.

I might caveat this by saying that an UNleveraged risk parity portfolio (40/50/10 stocks/LTT/gold) would have had a max Dd of about 20% with roughly a 10% CAGR, to which you could lever up to your hearts content.
I've been playing with the UPROSIM/TMFSIM dataset with some volatility weighting strategies for the last couple of days, so I'll try to address this question. This goes back to the start of 1987 through the start of 2019. I'm not sure if the expense ratios are completely priced in with this, but it gives an idea.

For 20-day rebalance frequency:

40/60 scheme

maximum drawdown of 47 to 57% and overall CAGR of 14.6 to 15.4%.

Target volatility of 20%

20-day lookback: maximum drawdown of 32 to 39% and overall CAGR of 16 to 19%.
60-day lookback: maximum drawdown of 32 to 40% and overall CAGR of 16.8 to 17.9%.

Straight risk parity (target volatility not used)

20-day lookback: maximum drawdown of 32 to 41% and overall CAGR of 16.3 to 17.6%.
60-day lookback: maximum drawdown of 31 to 41% and overall CAGR of 16.6 to 17.5%.

Variance parity (target volatility not used)

20-day lookback: maximum drawdown of 23 to 37% and overall CAGR of 15.8 to 17.8%.
60-day lookback: maximum drawdown of 32 to 46% (oops finger fumble should be 22 to 36) and overall CAGR of 17.1 to 18.1%.

Overall CAGR is for the entire 1987 to 2019 sequence.

The range of values depends on the starting day in the sequence. For a 20-day rebalance, there are 20 possible starting days with respect to calculating maximum drawdown. I did the calculations starting on the first 20 days of the UPROSIM/TMFSIM sequence. There is a significant sensitivity to starting date when dealing with large sudden events.

The variance parity case is like the risk parity, except using variance instead of standard deviation. The average UPRO weight is ~51% for the volatility target scheme, ~40% for the straight risk parity scheme, and ~33% for the variance parity scheme.

All bets are off for these schemes prior to circa 1982.
Last edited by Hydromod on Sat Aug 03, 2019 5:20 pm, edited 1 time in total.

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

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

Post by perplexed » Sat Aug 03, 2019 5:04 pm

This is an excellent analysis. So one of the takeaway points I see is that all of these strategies (target vol, ….) tend to reduce the max drawdown substantially from the straight up 40-60.

I am curious if by any chance you have the data to play with TQQQ. Also, is this TMFSIM/UPROSIM data already uploaded in PV?

many thanks.

Hydromod
Posts: 232
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod » Sat Aug 03, 2019 5:14 pm

perplexed wrote:
Sat Aug 03, 2019 5:04 pm
This is an excellent analysis. So one of the takeaway points I see is that all of these strategies (target vol, ….) tend to reduce the max drawdown substantially from the straight up 40-60.

I am curious if by any chance you have the data to play with TQQQ. Also, is this TMFSIM/UPROSIM data already uploaded in PV?

many thanks.
This is the dataset that Hedgefundie links to in the original post, about midway down. Various folks have uploaded their own copy to PV. You need to have an account first, then you can import the data as a benchmark.

I don't have TQQQ, unfortunately. Some of the other backtesting gurus may have developed a synthetic series, but I'm not aware of it.

I'll mention that I was doing this analysis outside of PV, just so I know exactly what my assumptions are.

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

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

Post by perplexed » Sat Aug 03, 2019 6:57 pm

I am trying to generate a TQQQSIM by importing historical price NASDAQ data from yahoo, calculating daily returns (%), and multiplying it by 3.

Comparing it with TQQQ, for the period TQQQ data is available, looks like PV is showing significantly lower returns in TQQQSIM. I am wondering what may be the source of errors. FEW issues come to mind, and there may easily be more.

- I don't know how to factor in dividends. Is there a clean way to do this? For example, how was it done for TMFSIM or UPROSIM.
- Is there a way to factor in leverage and ETF costs in the returns?

Again, many thanks folks.

Hydromod
Posts: 232
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod » Sat Aug 03, 2019 8:37 pm

perplexed wrote:
Sat Aug 03, 2019 6:57 pm
I am trying to generate a TQQQSIM by importing historical price NASDAQ data from yahoo, calculating daily returns (%), and multiplying it by 3.

Comparing it with TQQQ, for the period TQQQ data is available, looks like PV is showing significantly lower returns in TQQQSIM. I am wondering what may be the source of errors. FEW issues come to mind, and there may easily be more.

- I don't know how to factor in dividends. Is there a clean way to do this? For example, how was it done for TMFSIM or UPROSIM.
- Is there a way to factor in leverage and ETF costs in the returns?

Again, many thanks folks.
The dividends are incorporated in the adjusted close price, just use adjusted close price to calculate total returns.

There's an extensive thread started by EfficientInvestor on simulating leveraged ETFs. It's a few pages back in the forum by now. This talks about such issues in detail.

schismal
Posts: 135
Joined: Sat Apr 13, 2019 8:53 pm

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

Post by schismal » Sun Aug 04, 2019 11:33 am

samsdad wrote:
Sat Aug 03, 2019 4:32 pm
It’s in the original post. See the blue hyperlink about half-way through the post that says “here.”

Also, I haven’t looked at your data yet. If you could include the equations that you’re using, that would be appreciated.
Thanks! Here is the updated chart from 1987 through 2019: EXCEL LINK

I did the selection manually, so there are no real equations (beyond calculating StDev). I just looked at the volatility numbers and decided the allocation monthly based on the asset cap rule (80%, 70%, 60%). If you download it, you can unhide additional columns where I pulled the the return data (for 80/20 & 20/80, 70/30 & 30/70, etc. rebalanced monthly) and selected the allocation for each rule based on the prior month's volatility. Feel free to look for errors, since this was done by hand.

Image

User avatar
physixfan
Posts: 24
Joined: Wed Apr 03, 2019 4:09 pm

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

Post by physixfan » Sun Aug 04, 2019 1:20 pm

MotoTrojan wrote:
Thu Jul 25, 2019 1:43 pm
caklim00 wrote:
Thu Jul 25, 2019 12:29 pm
MotoTrojan wrote:
Thu Jul 25, 2019 11:24 am
PV will do this for you with one bond fund but it’s simple math. Allocation of UPRO = (1/upro_vol)/((1/upro_vol)+(1/bond_vol)). For me the bond vol is simply the average of both funds since their correlation is essentially 1.

I trade last day of month but that’s arbitrary. Vol comes from past 20 trading days.
Sorry for being such a pain. Where do you find the volume in PV? Or ar you just clicking into one of the sections? I was trying to use the Portfolio Optimization section using Risk Parity as Optimization Goal, but realize I am way off.

I'm trying to setup an easy way that I can calculate every month else I'm better off trying a static allocation such as 50/50 and then just doing quarterly rebalancing.
Step by step:
1: PV website, click on Adaptive Allocation under Timing Models (bottom right)
2: Add your tickers (only use 2 for now)
3: Set Volatility Period to Specify and then manually enter 20 days
4: Monthly rebalance
5: Run the test
6: Under results, click Timing Periods tab
7: Scroll to the very bottom and it'll show you the allocation based on the current signal date, as well as the raw volatility of each fund

If you wanted to adapt my 50/50 bond strategy or anything similar then you could add all three assets, turn the number of assets to 3, and then use the some process but you would ignore the allocation recommendation and just use PV as a tool to give you the raw volatility of each holding, then do the math yourself.

Equity markets are getting awfully calm right now. Will be an interesting rebalance on Wednesday!
Thank you for the steps! It seems Portfolio Visualizer has a very convenient sharing feature, am I doing this correctly?

https://www.portfoliovisualizer.com/tes ... 0&total1=0

MotoTrojan
Posts: 6474
Joined: Wed Feb 01, 2017 8:39 pm

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

Post by MotoTrojan » Sun Aug 04, 2019 2:46 pm

physixfan wrote:
Sun Aug 04, 2019 1:20 pm
MotoTrojan wrote:
Thu Jul 25, 2019 1:43 pm
caklim00 wrote:
Thu Jul 25, 2019 12:29 pm
MotoTrojan wrote:
Thu Jul 25, 2019 11:24 am
PV will do this for you with one bond fund but it’s simple math. Allocation of UPRO = (1/upro_vol)/((1/upro_vol)+(1/bond_vol)). For me the bond vol is simply the average of both funds since their correlation is essentially 1.

I trade last day of month but that’s arbitrary. Vol comes from past 20 trading days.
Sorry for being such a pain. Where do you find the volume in PV? Or ar you just clicking into one of the sections? I was trying to use the Portfolio Optimization section using Risk Parity as Optimization Goal, but realize I am way off.

I'm trying to setup an easy way that I can calculate every month else I'm better off trying a static allocation such as 50/50 and then just doing quarterly rebalancing.
Step by step:
1: PV website, click on Adaptive Allocation under Timing Models (bottom right)
2: Add your tickers (only use 2 for now)
3: Set Volatility Period to Specify and then manually enter 20 days
4: Monthly rebalance
5: Run the test
6: Under results, click Timing Periods tab
7: Scroll to the very bottom and it'll show you the allocation based on the current signal date, as well as the raw volatility of each fund

If you wanted to adapt my 50/50 bond strategy or anything similar then you could add all three assets, turn the number of assets to 3, and then use the some process but you would ignore the allocation recommendation and just use PV as a tool to give you the raw volatility of each holding, then do the math yourself.

Equity markets are getting awfully calm right now. Will be an interesting rebalance on Wednesday!
Thank you for the steps! It seems Portfolio Visualizer has a very convenient sharing feature, am I doing this correctly?

https://www.portfoliovisualizer.com/tes ... 0&total1=0
Looks good to me! From there you'd simply skip to Step 6 every month to see your upcoming allocation.

caklim00
Posts: 1914
Joined: Mon May 26, 2008 10:09 am

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

Post by caklim00 » Sun Aug 04, 2019 4:12 pm

Setup everything in M1, just waiting on the money to trandsfer over. Doing 34% in 20% volatility, 33% in 20 day lookback for risk parity, and 33% in OP original strategy. I slightly tweeked the 20% volatity putting a little more focus on Days 1-10, less on 11-15, and even less on days 16-20. I couldn't figure out how to alter the 20 day lookback to account for more recent days so I wasn't able to do that unfortunately.

Here are the links:
1)20 Day lookback for Risk Parity:
https://www.portfoliovisualizer.com/tes ... 0&total1=0
2) 20 Day volatility (10-30%,15-30%,20-40%):
https://www.portfoliovisualizer.com/tes ... total1=100

User avatar
privatefarmer
Posts: 598
Joined: Mon Sep 08, 2014 2:45 pm

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

Post by privatefarmer » Sun Aug 04, 2019 6:16 pm

MotoTrojan wrote:
Sun Aug 04, 2019 2:46 pm
physixfan wrote:
Sun Aug 04, 2019 1:20 pm
MotoTrojan wrote:
Thu Jul 25, 2019 1:43 pm
caklim00 wrote:
Thu Jul 25, 2019 12:29 pm
MotoTrojan wrote:
Thu Jul 25, 2019 11:24 am
PV will do this for you with one bond fund but it’s simple math. Allocation of UPRO = (1/upro_vol)/((1/upro_vol)+(1/bond_vol)). For me the bond vol is simply the average of both funds since their correlation is essentially 1.

I trade last day of month but that’s arbitrary. Vol comes from past 20 trading days.
Sorry for being such a pain. Where do you find the volume in PV? Or ar you just clicking into one of the sections? I was trying to use the Portfolio Optimization section using Risk Parity as Optimization Goal, but realize I am way off.

I'm trying to setup an easy way that I can calculate every month else I'm better off trying a static allocation such as 50/50 and then just doing quarterly rebalancing.
Step by step:
1: PV website, click on Adaptive Allocation under Timing Models (bottom right)
2: Add your tickers (only use 2 for now)
3: Set Volatility Period to Specify and then manually enter 20 days
4: Monthly rebalance
5: Run the test
6: Under results, click Timing Periods tab
7: Scroll to the very bottom and it'll show you the allocation based on the current signal date, as well as the raw volatility of each fund

If you wanted to adapt my 50/50 bond strategy or anything similar then you could add all three assets, turn the number of assets to 3, and then use the some process but you would ignore the allocation recommendation and just use PV as a tool to give you the raw volatility of each holding, then do the math yourself.

Equity markets are getting awfully calm right now. Will be an interesting rebalance on Wednesday!
Thank you for the steps! It seems Portfolio Visualizer has a very convenient sharing feature, am I doing this correctly?

https://www.portfoliovisualizer.com/tes ... 0&total1=0
Looks good to me! From there you'd simply skip to Step 6 every month to see your upcoming allocation.
Okay now I am very intrigued. Can someone explain “adaptive allocation” to me in “idiot” terms?? Is it basically rebalancing the portfolio monthly using a 20day look back with the goal of obtaining risk parity between the two assets?? And we would do this bc, historically, volatility has been easier to predict than returns of asset classes? Thanks!!

User avatar
privatefarmer
Posts: 598
Joined: Mon Sep 08, 2014 2:45 pm

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

Post by privatefarmer » Sun Aug 04, 2019 6:22 pm

And what does “timing period” on that PV test mean? It defaults to 3 months

rascott
Posts: 982
Joined: Wed Apr 15, 2015 10:53 am

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

Post by rascott » Sun Aug 04, 2019 8:28 pm

55% UPRO (risk parity)
80% UPRO (target 20% vol)


This sound correct?

caklim00
Posts: 1914
Joined: Mon May 26, 2008 10:09 am

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

Post by caklim00 » Sun Aug 04, 2019 8:52 pm

rascott wrote:
Sun Aug 04, 2019 8:28 pm
55% UPRO (risk parity)
80% UPRO (target 20% vol)


This sound correct?
56% right now for risk parity. 80% sounds about right for 20% volatility. Mine shows as 74% but it's due to weighting most recent days more than days > 10

sparksfly
Posts: 29
Joined: Mon Oct 16, 2017 9:57 pm

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

Post by sparksfly » Sun Aug 04, 2019 8:55 pm

I was trying to update the sim data files for TMF and UPRO to make them current by adding the missing daily data after 02/21/19 (the last date in the sim files). However, I noticed something that I wanted to check with folks here. I downloaded the daily data from Feb to now from yahoo finance and calculated daily % returns. As I started adding, I noticed that the daily returns for existing data in the file were off compared to what I calculated from yahoo download. Few dates in Feb (column yahoo is the % return calculated from yahoo download, UPROSIM column is the return in the UPROSIM file and the last column is the diff between the two

Code: Select all

Date	Close . 	Yahoo	UPROSIM	SIM - YAHOO DIFF
2/1/19	43.150002	-	-	-
2/4/19	44.07 .  	2.13%	2.00%	-0.13%
2/5/19	44.639999	1.29%	1.40%	0.11%
2/6/19	44.419998	-0.49%	-0.70%	-0.21%
2/7/19	43.139999	-2.88%	-2.80%	0.08%
2/8/19	43.299999	0.37%	0.30%	-0.07%
2/11/19	43.369999	0.16%	0.20%	0.04%
2/12/19	45.029999	3.83%	3.90%	0.07%
2/13/19	45.43 .   	0.89%	0.90%	0.01%
2/14/19	45.119999	-0.68%	-0.70%	-0.02%
2/15/19	46.57 .  	3.21%	3.30%	0.09%
2/19/19	46.799999	0.49%	0.40%	-0.09%
2/20/19	47.080002	0.60%	0.60%	0.00%
2/21/19	46.560001	-1.10%	-1.00%	0.10%
can someone help explain where the gap is?

Topic Author
HEDGEFUNDIE
Posts: 3527
Joined: Sun Oct 22, 2017 2:06 pm

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

Post by HEDGEFUNDIE » Sun Aug 04, 2019 8:57 pm

sparksfly wrote:
Sun Aug 04, 2019 8:55 pm
I was trying to update the sim data files for TMF and UPRO to make them current by adding the missing daily data after 02/21/19 (the last date in the sim files). However, I noticed something that I wanted to check with folks here. I downloaded the daily data from Feb to now from yahoo finance and calculated daily % returns. As I started adding, I noticed that the daily returns for existing data in the file were off compared to what I calculated from yahoo download. Few dates in Feb (column yahoo is the % return calculated from yahoo download, UPROSIM column is the return in the UPROSIM file and the last column is the diff between the two

Code: Select all

Date	Close . 	Yahoo	UPROSIM	SIM - YAHOO DIFF
2/1/19	43.150002	-	-	-
2/4/19	44.07 .  	2.13%	2.00%	-0.13%
2/5/19	44.639999	1.29%	1.40%	0.11%
2/6/19	44.419998	-0.49%	-0.70%	-0.21%
2/7/19	43.139999	-2.88%	-2.80%	0.08%
2/8/19	43.299999	0.37%	0.30%	-0.07%
2/11/19	43.369999	0.16%	0.20%	0.04%
2/12/19	45.029999	3.83%	3.90%	0.07%
2/13/19	45.43 .   	0.89%	0.90%	0.01%
2/14/19	45.119999	-0.68%	-0.70%	-0.02%
2/15/19	46.57 .  	3.21%	3.30%	0.09%
2/19/19	46.799999	0.49%	0.40%	-0.09%
2/20/19	47.080002	0.60%	0.60%	0.00%
2/21/19	46.560001	-1.10%	-1.00%	0.10%
can someone help explain where the gap is?
Of course they will be different. UPROSIM is a model while UPRO is the real thing.

sparksfly
Posts: 29
Joined: Mon Oct 16, 2017 9:57 pm

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

Post by sparksfly » Sun Aug 04, 2019 9:04 pm

HEDGEFUNDIE wrote:
Sun Aug 04, 2019 8:57 pm
sparksfly wrote:
Sun Aug 04, 2019 8:55 pm
I was trying to update the sim data files for TMF and UPRO to make them current by adding the missing daily data after 02/21/19 (the last date in the sim files). However, I noticed something that I wanted to check with folks here. I downloaded the daily data from Feb to now from yahoo finance and calculated daily % returns. As I started adding, I noticed that the daily returns for existing data in the file were off compared to what I calculated from yahoo download. Few dates in Feb (column yahoo is the % return calculated from yahoo download, UPROSIM column is the return in the UPROSIM file and the last column is the diff between the two

Code: Select all

Date	Close . 	Yahoo	UPROSIM	SIM - YAHOO DIFF
2/1/19	43.150002	-	-	-
2/4/19	44.07 .  	2.13%	2.00%	-0.13%
2/5/19	44.639999	1.29%	1.40%	0.11%
2/6/19	44.419998	-0.49%	-0.70%	-0.21%
2/7/19	43.139999	-2.88%	-2.80%	0.08%
2/8/19	43.299999	0.37%	0.30%	-0.07%
2/11/19	43.369999	0.16%	0.20%	0.04%
2/12/19	45.029999	3.83%	3.90%	0.07%
2/13/19	45.43 .   	0.89%	0.90%	0.01%
2/14/19	45.119999	-0.68%	-0.70%	-0.02%
2/15/19	46.57 .  	3.21%	3.30%	0.09%
2/19/19	46.799999	0.49%	0.40%	-0.09%
2/20/19	47.080002	0.60%	0.60%	0.00%
2/21/19	46.560001	-1.10%	-1.00%	0.10%
can someone help explain where the gap is?
Of course they will be different. UPROSIM is a model while UPRO is the real thing.
That explains it...I was under impression that the data was simulated upto June 2009 (inception date) and real after that.

comeinvest
Posts: 290
Joined: Mon Mar 12, 2012 6:57 pm

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

Post by comeinvest » Sun Aug 04, 2019 9:06 pm

schismal wrote:
Fri Aug 02, 2019 11:09 am
perplexed wrote:
Fri Aug 02, 2019 10:48 am
Hello:

This has been an excellent thread and a solid learning.
I am curious what would be best options to implement this strategy using mutual funds. I have not found anything beyond 1.25X leverage using long-term treasuries. I would love to know if others have ventured using some products in the mutual funds world.

Many thanks.
They exist, (ULPIX and GVPIX, for instance). But if you're going to do a mutual fund, PSLDX seems like the best option.
Why would I buy PSLDX at 0.59% ER, when I can buy NTSX (WisdomTree 90/60 U.S. Balanced Fund) at 0.2% ER?
I realize that NTSX is 1.5% leveraged, while PSLDX is 2x leveraged. However, the ER/assets ratio is still more than twice for the Pimco fund. Since either fund can be used in combination with other ways to obtain leverage as discussed in this thread, I think ER/assets is a good measure of fees.
I also realize that PSLDX uses an actively managed bond portfolio, while NTSX uses treasury futures. The question is what is the benefit of the actively managed bond portfolio, and whether it justifies the higher ER. I guess they might achieve higher returns with probably higher tail risk, but less rebalancing bonus with stocks.
I also realize that PSLDX uses equity index futures, while NTSX uses treasury futures. I don't know which one is more efficient. I know that stock index futures had ca. 0.3%-0.4% penalty above LIBOR in the last few years due to rich rolls.

Topic Author
HEDGEFUNDIE
Posts: 3527
Joined: Sun Oct 22, 2017 2:06 pm

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

Post by HEDGEFUNDIE » Sun Aug 04, 2019 9:06 pm

sparksfly wrote:
Sun Aug 04, 2019 9:04 pm
HEDGEFUNDIE wrote:
Sun Aug 04, 2019 8:57 pm
sparksfly wrote:
Sun Aug 04, 2019 8:55 pm
I was trying to update the sim data files for TMF and UPRO to make them current by adding the missing daily data after 02/21/19 (the last date in the sim files). However, I noticed something that I wanted to check with folks here. I downloaded the daily data from Feb to now from yahoo finance and calculated daily % returns. As I started adding, I noticed that the daily returns for existing data in the file were off compared to what I calculated from yahoo download. Few dates in Feb (column yahoo is the % return calculated from yahoo download, UPROSIM column is the return in the UPROSIM file and the last column is the diff between the two

Code: Select all

Date	Close . 	Yahoo	UPROSIM	SIM - YAHOO DIFF
2/1/19	43.150002	-	-	-
2/4/19	44.07 .  	2.13%	2.00%	-0.13%
2/5/19	44.639999	1.29%	1.40%	0.11%
2/6/19	44.419998	-0.49%	-0.70%	-0.21%
2/7/19	43.139999	-2.88%	-2.80%	0.08%
2/8/19	43.299999	0.37%	0.30%	-0.07%
2/11/19	43.369999	0.16%	0.20%	0.04%
2/12/19	45.029999	3.83%	3.90%	0.07%
2/13/19	45.43 .   	0.89%	0.90%	0.01%
2/14/19	45.119999	-0.68%	-0.70%	-0.02%
2/15/19	46.57 .  	3.21%	3.30%	0.09%
2/19/19	46.799999	0.49%	0.40%	-0.09%
2/20/19	47.080002	0.60%	0.60%	0.00%
2/21/19	46.560001	-1.10%	-1.00%	0.10%
can someone help explain where the gap is?
Of course they will be different. UPROSIM is a model while UPRO is the real thing.
That explains it...I was under impression that the data was simulated upto June 2009 (inception date) and real after that.
If I had done that the thread would be at least a few hundred posts longer with people accusing me of pulling a fast one on them...

User avatar
privatefarmer
Posts: 598
Joined: Mon Sep 08, 2014 2:45 pm

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

Post by privatefarmer » Sun Aug 04, 2019 9:18 pm

Sorry for the novice question, but let’s say someday PV is down or no longer available. If we wanted to calculate the 20day look back risk parity ourselves, using just two assets, we would simply calculate the daily std dev over the last 20 trading days for each asset and then adjust the weights of the assets going forward so that they have equal weighted volatility in portfolio?

MoneyMarathon
Posts: 556
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon » Sun Aug 04, 2019 9:55 pm

comeinvest wrote:
Sun Aug 04, 2019 9:06 pm
Why would I buy PSLDX at 0.59% ER, when I can buy NTSX (WisdomTree 90/60 U.S. Balanced Fund) at 0.2% ER?
I realize that NTSX is 1.5% leveraged, while PSLDX is 2x leveraged. However, the ER/assets ratio is still more than twice for the Pimco fund. Since either fund can be used in combination with other ways to obtain leverage as discussed in this thread, I think ER/assets is a good measure of fees.
I also realize that PSLDX uses an actively managed bond portfolio, while NTSX uses treasury futures. The question is what is the benefit of the actively managed bond portfolio, and whether it justifies the higher ER. I guess they might achieve higher returns with probably higher tail risk, but less rebalancing bonus with stocks.
I also realize that PSLDX uses equity index futures, while NTSX uses treasury futures. I don't know which one is more efficient. I know that stock index futures had ca. 0.3%-0.4% penalty above LIBOR in the last few years due to rich rolls.
Thank you for bringing attention to the NTSX fund. Most of us haven't heard of it. Looks like it launched in 2018 and currently has about $5m in assets, while the PIMCO fund launched over ten years ago. Still, it looks look a very promising option.

These two funds are extremely different from each other. Both look like they could be among the best in their niche.

(1) PSLDX uses derivatives for equities that distribute lots of capital gains. It looks like one of the least tax-efficient investments out there. As a result in needs to be held in a tax-advantaged account. In a way, only the existence of tax-advantaged accounts makes PSLDX a viable investment.

NTSX is designed for tax efficiency, making it suitable for taxable accounts. It's a passive investment in the S&P 500: "The fund invests 90% of its net assets in the 500 largest U.S. stocks by market capitalization." So it doesn't have much cause to distribute capital gains on the 0.9x exposure to the S&P 500. Big +1 for NTSX there.

NTSX also gives convenient access to the tax advantages of treasury futures, which now are of use without all the capital gains on stocks: "In instances where fixed income total returns are primarily driven by interest income and held in taxable accounts, any income distributions are subject to withholding tax rates of up to 39.6%. By comparison, capital gains on Treasury futures contracts are taxed at 60% long-term, 40% short-term capital gains rates." This could make it not just a more efficient use of cash but also more tax-efficient than buying an unlevered bond fund.

(2) NTSX has exposure to 0.9x the S&P 500. PSLDX has exposure to 1.0x the S&P 500 and roughly 0.7x exposure to bonds with credit risk (corporate, high yield, international). Credit risk has some correlation to beta.

Both also provide exposure to term risk. In the recent past, when term risk has had negative correlation with beta, this has given PSLDX a net beta loading of about 0.9 and given NTSX a net beta loading of about 0.8, if using a CAPM factor regression. At the same time, by taking on additional credit and term risk, beyond what NTSX takes on, they provided annualized alpha of 13.81% for PSLDX and 5.44% for NTSX.

https://www.portfoliovisualizer.com/fac ... sisResults

Basically, PSLDX is giving you a lot more risk for your money. For those who want to take that risk and have tax-advantaged space, the expense ratio does not do enough to take away from the additional exposure to market, credit, and term risks, when those risks are paying positive premiums. That's not to say more risk is always better, but those who want exposure to those risks and are willing to pay for it, may prefer PSLDX.

Expense ratio / leverage is not a meaningful measure for maximizing portfolio performance. If the gains from additional leverage are more than the expense ratio and other costs, the ER / leverage can be worse with the result still being better.

(3) NTSX may perform a bit better in a period of rising rates, due to its lower duration exposure of 3 to 8 years. This makes it an interesting holding for those who want to hedge the risks of rising rates.

(4) NTSX should perform a bit better in a market decline, due to its lower exposure to market and credit risks, using only treasuries on the bond side. This makes it interesting for those who want a more defensive holding.

In short, PSLDX is better for those who want to take on more market / credit / term risk (it's more loaded on all three risks) and have tax-advantaged space, while NTSX is better for those who want to take on less term risk, no credit risk, and slightly less market risk... or those who need a holding for a taxable account.

I would definitely consider buying NTSX in taxable. Thanks again for pointing it out.

HawkeyePierce
Posts: 708
Joined: Tue Mar 05, 2019 10:29 pm
Location: Colorado

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

Post by HawkeyePierce » Sun Aug 04, 2019 10:03 pm

My understanding of NTSX is that Wisdom tree intends for it to be held in something like a 60/40/33 where the last 33 could be different sources of risk or alternative asset classes (small cap value, gold, whatever).

I think it's an interesting fund and I could see using it but I'm wary of diving in with an ETF that small. IIRC it's on some ETF deathwatch lists.

MoneyMarathon
Posts: 556
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon » Sun Aug 04, 2019 10:16 pm

HawkeyePierce wrote:
Sun Aug 04, 2019 10:03 pm
I think it's an interesting fund and I could see using it but I'm wary of diving in with an ETF that small. IIRC it's on some ETF deathwatch lists.
$5m assets for a fund only a year old at WisdomTree isn't that bad. If WisdomTree killed every fund like that, there'd be no WisdomTree, lol. There's a chance of forced capital gains (the main risk of a killed ETF), but the juice could still be worth the squeeze, unless I can find a better way to do the same thing in taxable. There's always risks when investing, sometimes those risks are compensated, and this might be one of those cases.

MoneyMarathon
Posts: 556
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon » Sun Aug 04, 2019 10:31 pm

Here's the simple case for NTSX:

https://www.wisdomtree.com/blog/2019-03 ... ered-60-40
Image

It looks like a very well-designed product that does a simple job very well, is based on credible research, and is the first of its kind, with tax efficiency, a low expense ratio, and "just buy the fund" appeal. I think we'll see more funds flowing into it as awareness spreads.

Topic Author
HEDGEFUNDIE
Posts: 3527
Joined: Sun Oct 22, 2017 2:06 pm

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

Post by HEDGEFUNDIE » Sun Aug 04, 2019 10:40 pm

If only NTSX took theory seriously and invested in long term Treasuries, I would be all over it.

HawkeyePierce
Posts: 708
Joined: Tue Mar 05, 2019 10:29 pm
Location: Colorado

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

Post by HawkeyePierce » Sun Aug 04, 2019 10:50 pm

There was some discussion in azanon's all-weather thread about using NTSX as a source of leverage for an all-weather portfolio. Even as-is it could be an excellent source of cheap leverage for an investor looking to go all-weather without sacrificing returns (or learning how to trade futures).

MoneyMarathon
Posts: 556
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon » Sun Aug 04, 2019 10:56 pm

HEDGEFUNDIE wrote:
Sun Aug 04, 2019 10:40 pm
If only NTSX took theory seriously and invested in long term Treasuries, I would be all over it.
If you have room in taxable, it has been keeping pace with MTUM, with less beta than MTUM.

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

https://www.portfoliovisualizer.com/fac ... e&total1=0

6x leverage on intermediate average duration could have some diversification appeal... it will hold up a lot better than UPRO/TMF if both sides of that long/long trade suffer at once. There are also tax benefits (60% long-term, 40% short-term capital gains rates for gains on treasury futures). If municipal bonds have appeal, even though they are callable and don't have the same benefits of convexity (they have negative convexity), maybe this should too (6x leverage and positive convexity, even if the average duration is 3-8 years, will benefit a lot from falling yields).

Topic Author
HEDGEFUNDIE
Posts: 3527
Joined: Sun Oct 22, 2017 2:06 pm

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

Post by HEDGEFUNDIE » Sun Aug 04, 2019 11:39 pm

Here is a backtest of a synthetic NSTX (90% SPY + 60% ITT - 50% cash) against USMV:

https://www.portfoliovisualizer.com/bac ... total3=100

And according to the min vol index (https://www.msci.com/documents/10199/f5 ... 761d009094), here are the drawdowns during the Great Recession:

Max Drawdown
Min Vol: -47%
S&P 500: -55%
Synth NSTX: -42%

It appears the unleveraged min vol fund does the job just fine. Tax cost is around 0.8% annually, I suspect NTSX will come in at around the same.

Hydromod
Posts: 232
Joined: Tue Mar 26, 2019 10:21 pm

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

Post by Hydromod » Sun Aug 04, 2019 11:40 pm

For those interested, I posted a lengthy exercise in my thread on refinements to Hedgefundie's approach. The post examines potential improvements to the adaptive parity approach by including macroeconomic information in the form of the monthly unemployment index, which provides a well-known leading indicator of recessions. This basically adapts the ideas from willthrill81's market timing thread.

A few percentage points were gained on the 37-year CAGR by pushing the equity weighting higher when non-recessions are indicated and quickly retreating to treasuries when upcoming recessions were signaled. This worked well for "slow" recessions, but failed for fast events such as Black Monday.

MoneyMarathon
Posts: 556
Joined: Sun Sep 30, 2012 3:38 am

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

Post by MoneyMarathon » Mon Aug 05, 2019 12:36 am

HEDGEFUNDIE wrote:
Sun Aug 04, 2019 11:39 pm
Max Drawdown
Min Vol: -47%
S&P 500: -55%
Synth NSTX: -42%

It appears the unleveraged min vol fund does the job just fine.
Interesting, thanks. Hopefully it works as well in the next drawdown. I do wonder how the financialization of min vol and the big inflows (lots of AUM from what may be "weak hands" in the next bear market) account for its recent run up & also how it will react to a drawdown. IDK.

rascott
Posts: 982
Joined: Wed Apr 15, 2015 10:53 am

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

Post by rascott » Mon Aug 05, 2019 4:34 am

HEDGEFUNDIE wrote:
Sun Aug 04, 2019 11:39 pm
Here is a backtest of a synthetic NSTX (90% SPY + 60% ITT - 50% cash) against USMV:

https://www.portfoliovisualizer.com/bac ... total3=100

And according to the min vol index (https://www.msci.com/documents/10199/f5 ... 761d009094), here are the drawdowns during the Great Recession:

Max Drawdown
Min Vol: -47%
S&P 500: -55%
Synth NSTX: -42%

It appears the unleveraged min vol fund does the job just fine. Tax cost is around 0.8% annually, I suspect NTSX will come in at around the same.

So when is someone going to come out with a 2x min vol etf? (120/80)?

rascott
Posts: 982
Joined: Wed Apr 15, 2015 10:53 am

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

Post by rascott » Mon Aug 05, 2019 4:39 am

Kind of glad I haven't jumped in with both feet on the volatility lookback..... Things looking pretty ugly this morning if I'd loaded up to 80% UPRO :o :o :shock:

JBeck
Posts: 147
Joined: Fri Apr 15, 2016 4:54 am

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

Post by JBeck » Mon Aug 05, 2019 5:39 am

rascott wrote:
Mon Aug 05, 2019 4:39 am
Kind of glad I haven't jumped in with both feet on the volatility lookback..... Things looking pretty ugly this morning if I'd loaded up to 80% UPRO :o :o :shock:
What about PSLDX, did you end up buying it?

rascott
Posts: 982
Joined: Wed Apr 15, 2015 10:53 am

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

Post by rascott » Mon Aug 05, 2019 5:52 am

JBeck wrote:
Mon Aug 05, 2019 5:39 am
rascott wrote:
Mon Aug 05, 2019 4:39 am
Kind of glad I haven't jumped in with both feet on the volatility lookback..... Things looking pretty ugly this morning if I'd loaded up to 80% UPRO :o :o :shock:
What about PSLDX, did you end up buying it?

Haven't yet, but I want to...just can't get into it from anywhere I currently have any decent size accounts. Thinking of moving some things over to Ally to see if it will actually go through there.

Really sucks that Fidelity Brokerage link doesn't appear to offer it (per posts here, haven't checked myself.

jaj2276
Posts: 469
Joined: Sat Apr 16, 2011 5:13 pm

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

Post by jaj2276 » Mon Aug 05, 2019 6:35 am

rascott wrote:
Mon Aug 05, 2019 4:39 am
Kind of glad I haven't jumped in with both feet on the volatility lookback..... Things looking pretty ugly this morning if I'd loaded up to 80% UPRO :o :o :shock:
Ya not ideal, ha.

rascott
Posts: 982
Joined: Wed Apr 15, 2015 10:53 am

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

Post by rascott » Mon Aug 05, 2019 6:38 am

jaj2276 wrote:
Mon Aug 05, 2019 6:35 am
rascott wrote:
Mon Aug 05, 2019 4:39 am
Kind of glad I haven't jumped in with both feet on the volatility lookback..... Things looking pretty ugly this morning if I'd loaded up to 80% UPRO :o :o :shock:
Ya not ideal, ha.
I'll likely still do some for a small piece....because I like the action and the fiddling....and keeps me away from messing with my "normal" accounts. :P probably even use TQQQ :twisted:

rascott
Posts: 982
Joined: Wed Apr 15, 2015 10:53 am

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

Post by rascott » Mon Aug 05, 2019 9:17 am

Almost like this risk parity thing works. TMF+4%, UPRO -6%.

60/40 dead even. :greedy

Locked