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

[MoneyMarathon]

Understood. This entire thread relies on the efficiency boost of uncorrelated assets hence why I find it ironic you’re spinning it as a positive. Your plan sounds great though so let’r rip.

Cheers!

Leverage can juice returns for traditional equity-risk-heavy portfolios too.

[MotoTrojan]

Understood. This entire thread relies on the efficiency boost of uncorrelated assets hence why I find it ironic you’re spinning it as a positive. Your plan sounds great though so let’r rip.

Cheers!

Leverage can juice returns for traditional equity-risk-heavy portfolios too.

What is the equity leverage of the PIMCO fund? Thought it was effectively 1x exposure plus bonds?

[HEDGEFUNDIE]

Understood. This entire thread relies on the efficiency boost of uncorrelated assets hence why I find it ironic you’re spinning it as a positive. Your plan sounds great though so let’r rip.

Cheers!

Leverage can juice returns for traditional equity-risk-heavy portfolios too.

What is the equity leverage of the PIMCO fund? Thought it was effectively 1x exposure plus bonds?

It’s 2x exposure.

[no simpler]

Just discovered saved benchmarks in PV, whichs makes it a lot easier to compare conventional portfolios with market timing strategies.

Portfolio 1: 50/50 VEDTX (EDV) & TMF as combined bond component, rebalanced monthly for risk-parity w/ UPRO (20 day look-back)
Portfolio 2: TMF & UPRO rebalanced monthly for risk-parity
Portfolio 3: Original 40/60 UPRO/TMF quarterly rebalance
S&P500

The TMF/UPRO risk-parity is a significant boost in returns, but you can still get a significant leg-up on the original strategy with significantly less risk/drawdown via the risk-parity variant dosed with EDV. Shown here are the above 4 portfolios from January 2008 to January 2019:



Here is the original strategy compared to the 1-month lookback risk-parity with TMF/UPRO from 1987 to 2019:

This is really really cool. There's strong enough theoretical grounds for volatility targeting that this is likely not just the result of back-fitting. Would be cool if total leverage could also be scaled up and down based on total predicted portfolio volatility.

Also, when time allows, I'd love to try predicting vol using slightly more advanced methods which are more accurate than historical lookback. It's long been known that vol is predictable (unlike returns), but hard to tie better vol prediction to better long only portfolio returns. This seems like a way to make the link between the two.

Here's an interesting paper on predicting vol from Lazard: https://www.lazardassetmanagement.com/d ... rch_en.pdf

Lastly, I tried using the search and couldn't find anything - but has anyone tried swapping out UPRO with a simulated 3X min vol global equity? This is basically vol targeting at the individual equity level, rebalanced twice a year: https://www.msci.com/documents/10199/24 ... 5a2ec105fe

[MoneyMarathon]

What is the equity leverage of the PIMCO fund? Thought it was effectively 1x exposure plus bonds?

Yes, I think that's their target. To get there they need to use financial instruments (futures, swaps, options) that mean they are roughly 2x levered overall, relative to net assets, split about equally between the S&P 500 index and their bond holdings.

From the prospectus:

"The Fund typically will seek to gain long exposure to the S&P 500 Index in an amount, under normal circumstances, approximately equal to the Fund's net assets. The value of S&P 500 Index derivatives should closely track changes in the value of the S&P 500 Index. However, S&P 500 Index derivatives may be purchased with a small fraction of the assets that would be needed to purchase the equity securities directly, so that the remainder of the assets may be invested in Fixed Income Instruments."

Derivatives are used on the stock side unless "derivatives appear to be overvalued relative to the S&P 500 Index." They apparently give themselves a bit of license, overall:

Though the Fund does not normally invest directly in S&P 500 Index securities, when S&P 500 Index derivatives appear to be overvalued relative to the S&P 500 Index, the Fund may invest all of its assets in a "basket" of S&P 500 Index stocks. The Fund also may invest in exchange-traded funds based on the S&P 500 Index, such as Standard & Poor's Depositary Receipts.

The Fund may invest, without limitation, in derivative instruments, such as options, futures contracts or swap agreements, or in mortgage- or asset-backed securities, subject to applicable law and any other restrictions described in the Fund's prospectus or Statement of Additional Information. The Fund may purchase or sell securities on a when-issued, delayed delivery or forward commitment basis and may engage in short sales. Assets not invested in equity securities or derivatives may be invested in Fixed Income Instruments. The Fund may invest up to 10% of its total assets in high yield securities ("junk bonds") rated B or higher by Moody's Investors Service, Inc. ("Moody's"), or equivalently rated by Standard & Poor's Ratings Services ("S&P") or Fitch, Inc. ("Fitch"), or if unrated, determined by PIMCO to be of comparable quality. The Fund may invest up to 30% of its total assets in securities denominated in foreign currencies and may invest beyond this limit in U.S. dollar-denominated securities of foreign issuers. The Fund may invest up to 15% of its total assets in securities and instruments that are economically tied to emerging market countries (this limitation does not apply to investment grade sovereign debt denominated in the local currency with less than 1 year remaining to maturity, which means the Fund may invest, together with any other investments denominated in foreign currencies, up to 30% of its total assets in such instruments). The Fund will normally limit its foreign currency exposure (from non-U.S. dollar-denominated securities or currencies) to 20% of its total assets. The Fund may also invest up to 10% of its total assets in preferred stocks.

One nice thing is that this fund goes back to 2007 - "We launched the StocksPLUS Long Duration Fund in 2007" - and PIMCO has survived some adverse conditions: "Our success in managing long duration bond portfolios since 1988 (a period that included rising and falling interest rate environments) speaks to the strength of our disciplined investment process and portfolio construction approach."

[MotoTrojan]

One nice thing is that this fund goes back to 2007 - "We launched the StocksPLUS Long Duration Fund in 2007" - and PIMCO has survived some adverse conditions: "Our success in managing long duration bond portfolios since 1988 (a period that included rising and falling interest rate environments) speaks to the strength of our disciplined investment process and portfolio construction approach."

I like this last part. Sure if you zoom in enough they went up and down .

[MotoTrojan]

Understood. This entire thread relies on the efficiency boost of uncorrelated assets hence why I find it ironic you’re spinning it as a positive. Your plan sounds great though so let’r rip.

Cheers!

Leverage can juice returns for traditional equity-risk-heavy portfolios too.

What is the equity leverage of the PIMCO fund? Thought it was effectively 1x exposure plus bonds?

It’s 2x exposure.

MoneyMarathon's quote seems to say otherwise. 2x leverage to achieve 1x exposure to S&P500 plus some bonds perhaps?

[MoneyMarathon]

Understood. This entire thread relies on the efficiency boost of uncorrelated assets hence why I find it ironic you’re spinning it as a positive. Your plan sounds great though so let’r rip.

Cheers!

Leverage can juice returns for traditional equity-risk-heavy portfolios too.

What is the equity leverage of the PIMCO fund? Thought it was effectively 1x exposure plus bonds?

It’s 2x exposure.

MoneyMarathon's quote seems to say otherwise. 2x leverage to achieve 1x exposure to S&P500 plus some bonds perhaps?

I think HEDGEFUNDIE, you, and I are all saying the same thing (in slightly different ways). It's about 1x stocks and 2x overall.

40% UPRO / 60% TMF for example is about 1.2x exposure to the S&P 500 but 3x leverage overall.

[MotoTrojan]

Just discovered saved benchmarks in PV, whichs makes it a lot easier to compare conventional portfolios with market timing strategies.

Portfolio 1: 50/50 VEDTX (EDV) & TMF as combined bond component, rebalanced monthly for risk-parity w/ UPRO (20 day look-back)
Portfolio 2: TMF & UPRO rebalanced monthly for risk-parity
Portfolio 3: Original 40/60 UPRO/TMF quarterly rebalance
S&P500

The TMF/UPRO risk-parity is a significant boost in returns, but you can still get a significant leg-up on the original strategy with significantly less risk/drawdown via the risk-parity variant dosed with EDV. Shown here are the above 4 portfolios from January 2008 to January 2019:


Here is the original strategy compared to the 1-month lookback risk-parity with TMF/UPRO from 1987 to 2019:

This is really really cool. There's strong enough theoretical grounds for volatility targeting that this is likely not just the result of back-fitting. Would be cool if total leverage could also be scaled up and down based on total predicted portfolio volatility.

You could maintain risk-parity between equity & bond holdings whenever you rebalance, and also tune overall leverage based on target volatility by adjusting the bond side between 100% EDV, a blend, and 100% TMF (the more EDV you hold, the less UPRO you'd hold, and less overall leverage). Too many moving pieces for me though. I feel good with my 50/50 bond compromise and performance historically has been very close to the OPs quarterly method, with far less volatility/drawdown.

[MotoTrojan]

Rerunning numbers

[MoneyMarathon]

Investors can be funny creatures. When I looked at the outflows for PSLDX, the biggest came in 2013. That year PSLDX gained 18.86% but the SPY index gained 32.31% ... so, both were good, but the increase in bond yields hurt bond prices and PSLDX had a smaller double digit gain. Just feeling that they had "lost" against the index is apparently enough to get people to flood out of a strategy. Then, in 2014 and later, there has been a gain of 147% for PSLDX and a gain of 77% for SPY.

I guess I am just echoing those saying that "staying the course" is key to this strategy, like any other, and the hardest part.

[rascott]

Here's the prospectus:

https://www.pimco.com/handlers/displayd ... BrQp1k4%3D

Here's a white paper from PIMCO:

https://www.pimco.com/en-us/insights/in ... fficiency/

Here's a key line: "PIMCO actively manages the Fixed Income Instruments held by the Fund with a view toward enhancing the Fund’s total return, subject to an overall portfolio duration which normally varies within two years (plus or minus) of the portfolio duration of the securities comprising the Bloomberg Barclays Long-Term Government/Credit Index, as calculated by PIMCO, which as of May 31, 2018 was 14.57 years."

The equity side is passive. I'm more comfortable with active management of bonds since they suffer less from the needle/haystack problem of excluding the stocks with the biggest gains, since bonds can't really deliver much more than they promise, just duration exposure and credit risk. I also think the duration risk on the bond side is reasonable and they do seem to have enough in treasuries. They had a reasonable 2008 drawdown (-50% and pretty much matching the S&P 500 index).

I agree with HEDGEFUNDIE and think I will use this for roughly 2x leverage. If it loses a couple basis points, against something I could have done manually, at least they pushed the buttons for me.

So where are you buying it....and what's the min?

[MoneyMarathon]

So where are you buying it....and what's the min?

Vanguard, in a Roth IRA, minimum $25,000.

[dave_k]

I just checked Vanguard and they also let you buy it, with a $25,000 minimum.

What are the chances PSLDX would be allowed in a Vanguard 401k plan? I could request that it be added, but not sure if Vanguard would do it.

Apparently Fidelity imposes the $1M minimum, so I couldn't do it in the IRA that has some money dedicated to UPRO/TMF.

[MoneyMarathon]

I just checked Vanguard and they also let you buy it, with a $25,000 minimum.

What are the chances PSLDX would be allowed in a Vanguard 401k plan? I could request that it be added, but not sure if Vanguard would do it.

I don't know. I hope Vanguard doesn't get a bunch of inquiries into this and maybe, possibly decide to close it to new investments or raise minimum in IRA.

Apparently Fidelity imposes the $1M minimum, so I couldn't do it in the IRA that has some money dedicated to UPRO/TMF.

I think the simplest thing would just be to transfer $25,000+ to a Vanguard IRA.

[firebirdparts]

What's the deal with this fund? Can individuals buy? Is it $100k minimum?

You know, I hold PSLDX in a Fidelity 401k with brokeragelink, below $25,000, but at least in my account, they no longer offer it. I guess I can't buy more. I am not complaining. I bought it and also Absolute return PSPTX before I read this thread. Thinking now I wouldn't mind having some more.

[schismal]

You know, I hold PSLDX in a Fidelity 401k with brokeragelink, below $25,000, but at least in my account, they no longer offer it.

Same. It's no longer offered to us plebeian investors in my Fidelity brokeragelink or Roth IRA.

[PluckyDucky]

Are dividends auto reinvested in PSLDX? Or is the $25k min just the initial investment, but then you can add in any increments?

[columbia]

Are dividends auto reinvested in PSLDX? Or is the $25k min just the initial investment, but then you can add in any increments?

Vanguard site says auto investing of dividends and subsequent purchases of >= $1000 (plus $20 purchase free).

[rascott]

You know, I hold PSLDX in a Fidelity 401k with brokeragelink, below $25,000, but at least in my account, they no longer offer it.

Same. It's no longer offered to us plebeian investors in my Fidelity brokeragelink or Roth IRA.

Ally Invest seems to allow the purchase in an IRA....$100 min, $9.95 fee. Though I haven't tried actually executing an order.

[FIProfessor]

If a 3x ETF has 0.9% ER, the finance cost (above Libor) per leverage unit as a multiple of equity is ca. 0.45%. (I attribute the ER of LETFs to the financing cost, as unleveraged ETFs have ERs near zero or zero.) Futures unfortunately had similar spreads as of late. Futures would still lack the volatility drag, but I hear it is also very small for LETFs with monthly rebalancing.

Unfortunately, I'm not sure the financing cost is actually accounted for in the ER. According to the prospectus for UPRO, the ER is split into .75% for advisory management fees and .17% for other expenses. The Direxion monthly ETF have a similar advisory fee. Siamond has put a lot of effort into trying to figure out what the actual financing costs are in this thread (viewtopic.php?f=10&t=272640), but it seems really tricky to figure out.

[HEDGEFUNDIE]

If a 3x ETF has 0.9% ER, the finance cost (above Libor) per leverage unit as a multiple of equity is ca. 0.45%. (I attribute the ER of LETFs to the financing cost, as unleveraged ETFs have ERs near zero or zero.) Futures unfortunately had similar spreads as of late. Futures would still lack the volatility drag, but I hear it is also very small for LETFs with monthly rebalancing.

Unfortunately, I'm not sure the financing cost is actually accounted for in the ER. According to the prospectus for UPRO, the ER is split into .75% for advisory management fees and .17% for other expenses. The Direxion monthly ETF have a similar advisory fee. Siamond has put a lot of effort into trying to figure out what the actual financing costs are in this thread (viewtopic.php?f=10&t=272640), but it seems really tricky to figure out.

The financing cost is definitely not included in the ER, but it’s also not difficult to figure out. It is essentially the short term interest rate on the amount that is being borrowed.

So for a 3x fund that borrows 2x, you pay the ER of 1% and 2x the short term interest rate (let’s call it 2%). So 5% total annual costs.

[MotoTrojan]

Just discovered saved benchmarks in PV, whichs makes it a lot easier to compare conventional portfolios with market timing strategies.

Portfolio 1: 50/50 VEDTX (EDV) & TMF as combined bond component, rebalanced monthly for risk-parity w/ UPRO (20 day look-back)
Portfolio 2: TMF & UPRO rebalanced monthly for risk-parity
Portfolio 3: Original 40/60 UPRO/TMF quarterly rebalance
S&P500

The TMF/UPRO risk-parity is a significant boost in returns, but you can still get a significant leg-up on the original strategy with significantly less risk/drawdown via the risk-parity variant dosed with EDV. Shown here are the above 4 portfolios from January 2008 to January 2019:



Here is the original strategy compared to the 1-month lookback risk-parity with TMF/UPRO from 1987 to 2019:

Updated the 1st image in the above quote; I had incorrectly modeled portfolio 2. Returns are slightly boosted now due to a moderately larger exposure to the TMF parity portfolio. Now essentially a 1-1 with the OP but with way less volatility/drawdown. This will be my long-term portfolio starting end of this month.

[butricksaid]

Just discovered saved benchmarks in PV, whichs makes it a lot easier to compare conventional portfolios with market timing strategies.

Portfolio 1: 50/50 VEDTX (EDV) & TMF as combined bond component, rebalanced monthly for risk-parity w/ UPRO (20 day look-back)
Portfolio 2: TMF & UPRO rebalanced monthly for risk-parity
Portfolio 3: Original 40/60 UPRO/TMF quarterly rebalance
S&P500

The TMF/UPRO risk-parity is a significant boost in returns, but you can still get a significant leg-up on the original strategy with significantly less risk/drawdown via the risk-parity variant dosed with EDV. Shown here are the above 4 portfolios from January 2008 to January 2019:



Here is the original strategy compared to the 1-month lookback risk-parity with TMF/UPRO from 1987 to 2019:

Updated the 1st image in the above quote; I had incorrectly modeled portfolio 2. Returns are slightly boosted now due to a moderately larger exposure to the TMF parity portfolio. Now essentially a 1-1 with the OP but with way less volatility/drawdown. This will be my long-term portfolio starting end of this month.

What's the difference between Portfolio 2 and 3? Are they both 40/60 UPRO/TMF and #2 is rebalanced monthly while #3 is quarterly or is Portfolio 2 (50/50)?

If they aren't both the same, how does frequency of rebalancing affect a 40/60 UPRO/TMF portfolio?
Is there an optimal rebalancing frequency when making additional contributions?
Is there an optimal rebalancing frequency when not making additional contributions?

[FIProfessor]

If a 3x ETF has 0.9% ER, the finance cost (above Libor) per leverage unit as a multiple of equity is ca. 0.45%. (I attribute the ER of LETFs to the financing cost, as unleveraged ETFs have ERs near zero or zero.) Futures unfortunately had similar spreads as of late. Futures would still lack the volatility drag, but I hear it is also very small for LETFs with monthly rebalancing.

Unfortunately, I'm not sure the financing cost is actually accounted for in the ER. According to the prospectus for UPRO, the ER is split into .75% for advisory management fees and .17% for other expenses. The Direxion monthly ETF have a similar advisory fee. Siamond has put a lot of effort into trying to figure out what the actual financing costs are in this thread (viewtopic.php?f=10&t=272640), but it seems really tricky to figure out.

The financing cost is definitely not included in the ER, but it’s also not difficult to figure out. It is essentially the short term interest rate on the amount that is being borrowed.

So for a 3x fund that borrows 2x, you pay the ER of 1% and 2x the short term interest rate (let’s call it 2%). So 5% total annual costs.

I agree with you that to first order, that is essentially it. But it's also essentially the cost of financing on futures, right? So in the context of trying to figure out whether futures are cheaper and by how much (setting aside ER), which is what I think comeinvest was discussing, it seems hard to get a precise enough analysis of the cost of carrying and rebalancing.

[schismal]

What's the difference between Portfolio 2 and 3? Are they both 40/60 UPRO/TMF and #2 is rebalanced monthly while #3 is quarterly or is Portfolio 2 (50/50)?

If they aren't both the same, how does frequency of rebalancing affect a 40/60 UPRO/TMF portfolio?
Is there an optimal rebalancing frequency when making additional contributions?
Is there an optimal rebalancing frequency when not making additional contributions?

MJ can clarify, but if I'm reading correctly, P2 is adjusted monthly for risk parity with regards to asset volatility over the preceding month, while P3 is rebalanced quarterly to the original 60/40 split. So instead of being a static allocation like P3, the ratio of UPRO to TMF changes month to month in P2 -- for instance, it'd currently be divided 47/53 UPRO/TMF based on June volatility data (if I'm calculating this correctly).

[Kbg]

EffInv,

Glad you did your homework. May want to check to see if you can designate a liquidate first asset or liquidation priority somehow. I’m with IB and you can do this.

For the previous page of the thread...for me I’m kinda surprised that the bond choice isn’t obvious. I would think IEF hands down. What your analysis is demonstrating to me is that the risk (rising rates) to reward (falling) is significantly more favorable to IEF vs. TMF.

Seriously, who is going to sit through 25 years of a 1.something Cagr? You could probably sit through 9.something. None of us know the future, we do know the results of two actual scenarios...prudence suggests one chooses an option that is acceptable in either scenario.

In the thread some are getting to a point where they are seeing the value of diversifying beyond stocks and bonds. I’m now on year 5 of such a 3x portfolio (live). Repeat: What your analysis is demonstrating to me is that the risk (rising rates) to reward (falling) is significantly more favorable to IEF vs. TMF. I’m very seriously looking at closing out my TMF. Lastly, if one is going to put real money is this you should FEAR correlation greatly. Whatever assets you put into this and if history is any guide, will go to zero individually. Stocks and bonds alternate between positive and negative correlation for long periods of time historically. Seems pretty stupid to me to have a two asset portfolio given that fact.

Side note: A 3xETF portfolio is quite entertaining to watch. I got lucky and rebalanced at the December low and my TQQQ position is up ~100%. Meanwhile, it was just short of torture having UGLD in my port. It is finally in the green for the first time ever.

Research hint...look at objective rebalancing strategies beyond just time based.

[HEDGEFUNDIE]

If a 3x ETF has 0.9% ER, the finance cost (above Libor) per leverage unit as a multiple of equity is ca. 0.45%. (I attribute the ER of LETFs to the financing cost, as unleveraged ETFs have ERs near zero or zero.) Futures unfortunately had similar spreads as of late. Futures would still lack the volatility drag, but I hear it is also very small for LETFs with monthly rebalancing.

Unfortunately, I'm not sure the financing cost is actually accounted for in the ER. According to the prospectus for UPRO, the ER is split into .75% for advisory management fees and .17% for other expenses. The Direxion monthly ETF have a similar advisory fee. Siamond has put a lot of effort into trying to figure out what the actual financing costs are in this thread (viewtopic.php?f=10&t=272640), but it seems really tricky to figure out.

The financing cost is definitely not included in the ER, but it’s also not difficult to figure out. It is essentially the short term interest rate on the amount that is being borrowed.

So for a 3x fund that borrows 2x, you pay the ER of 1% and 2x the short term interest rate (let’s call it 2%). So 5% total annual costs.

I agree with you that to first order, that is essentially it. But it's also essentially the cost of financing on futures, right? So in the context of trying to figure out whether futures are cheaper and by how much (setting aside ER), which is what I think comeinvest was discussing, it seems hard to get a precise enough analysis of the cost of carrying and rebalancing.

The leverage is predominantly executed via total return swaps, not futures. So the cost of borrowing is explicitly laid out in the financial reports. And it approximates the 1-month LIBOR rate.

[MotoTrojan]

What's the difference between Portfolio 2 and 3? Are they both 40/60 UPRO/TMF and #2 is rebalanced monthly while #3 is quarterly or is Portfolio 2 (50/50)?

If they aren't both the same, how does frequency of rebalancing affect a 40/60 UPRO/TMF portfolio?
Is there an optimal rebalancing frequency when making additional contributions?
Is there an optimal rebalancing frequency when not making additional contributions?

MJ can clarify, but if I'm reading correctly, P2 is adjusted monthly for risk parity with regards to asset volatility over the preceding month, while P3 is rebalanced quarterly to the original 60/40 split. So instead of being a static allocation like P3, the ratio of UPRO to TMF changes month to month in P2 -- for instance, it'd currently be divided 47/53 UPRO/TMF based on June volatility data (if I'm calculating this correctly).

That is correct. Also I misspoke when I said I corrected portfolio #2, it was #1 that I updated.

P2 is the OPs strategy of rebalancing quarterly. If rebalanced monthly the return came down to 15.29%; this has been one of my main frustrations with the originally strategy, there is no logical reason quarterly will be best in the future. My gut says quarterly was long enough to get some momentum bonus where-as monthly reset too often to achieve this (rebalanced into equities on the way down and out of them on the way up). Historically there was an optimal frequency, but it is hard to say what is best for the future. Hydromod did some interesting analysis in this thread where he shifted the starting date to see how sensitive rebalancing period was in a more rigorous way (PV is fixed).

P1 & P3 are both rebalanced monthly but the allocation resets based on the past 20 day's (1 month) volatility as schismal stated. For P1 the bond allocation is first fixed to be 50/50 TMF/EDV then the allocation adjustment is made. This strategy can have months with significant exposure to UPRO, I believe as high as 80%+ in P3's case.

[MotoTrojan]

EffInv,

For the previous page of the thread...for me I’m kinda surprised that the bond choice isn’t obvious. I would think IEF hands down. What your analysis is demonstrating to me is that the risk (rising rates) to reward (falling) is significantly more favorable to IEF vs. TMF.

Seriously, who is going to sit through 25 years of a 1.something Cagr? You could probably sit through 9.something. None of us know the future, we do know the results of two actual scenarios...prudence suggests one chooses an option that is acceptable in either scenario.

In order to maintain risk-parity with IEF you must hold significantly less equity. I am not sure how you can say it is a hands down choice, perhaps it was from 1955-1982 but today that is a bold statement. Awesome job on being in this for the past 5 years though. What is your CAGR?

[MoneyMarathon]

In order to maintain risk-parity with IEF you must hold significantly less equity.

True. But there is no "must" about maintaining risk parity (defined here as equal risk in each asset class).

So there's no must about less equity.

[Kbg]

EffInv,

For the previous page of the thread...for me I’m kinda surprised that the bond choice isn’t obvious. I would think IEF hands down. What your analysis is demonstrating to me is that the risk (rising rates) to reward (falling) is significantly more favorable to IEF vs. TMF.

Seriously, who is going to sit through 25 years of a 1.something Cagr? You could probably sit through 9.something. None of us know the future, we do know the results of two actual scenarios...prudence suggests one chooses an option that is acceptable in either scenario.

In order to maintain risk-parity with IEF you must hold significantly less equity. I am not sure how you can say it is a hands down choice, perhaps it was from 1955-1982 but today that is a bold statement. Awesome job on being in this for the past 5 years though. What is your CAGR?

For me, I look at the portfolio as a whole and I’ve gotten pretty good at ignoring individual assets in it...you have to with something like this. My rationale was quite simple when looking at all the stock/bond combos posted on thread page 51. Places eyes on the top line and bottom line for each section.

1955-81...good heavens 40/60 3x. Not even worth considering it’s so bad. And in real terms by very rough estimates I think your returns were between $40 and 25 per $100 invested. In other words you lost 60-75 real. An 8 CAGR delta between high and low.

82-fwd the top line, not great but ok and a 6 CAGR high/low delta.

Full period...pretty much the same performance, best sharpe, best dd and you have respectable returns throughout. No hoping for a repeat of 82 on...which OBTW is mathematically impossible.

My CAGR...good question. It’s been a while since I looked. Low double digits 10% ish I think. Gold has totally stunk it up since pretty much day one.

[schismal]

In order to maintain risk-parity with IEF you must hold significantly less equity. I am not sure how you can say it is a hands down choice, perhaps it was from 1955-1982 but today that is a bold statement. Awesome job on being in this for the past 5 years though. What is your CAGR?

This is true if you go UPRO/IEF. But, if you decide that the anti-correlation segment is going to be a 50/50 split between TMF and IEF (which has historically shown a decent compromise between volatility and return) and group them when calculating risk parity (meaning average the standard deviations), the split based on June movements would be 34/33/33 UPRO/TMF/IEF, which is pretty close to the 40/30/30 simulation MM ran on the previous page. So that's actually pretty close to risk parity between the anti-correlated parts.

There is no hand-down correct answer, of course, but I think it's a decent compromise.

[MotoTrojan]

In order to maintain risk-parity with IEF you must hold significantly less equity. I am not sure how you can say it is a hands down choice, perhaps it was from 1955-1982 but today that is a bold statement. Awesome job on being in this for the past 5 years though. What is your CAGR?

This is true if you go UPRO/IEF. But, if you decide that the anti-correlation segment is going to be a 50/50 split between TMF and IEF (which has historically shown a decent compromise between volatility and return) and group them when calculating risk parity (meaning average the standard deviations), the split based on June movements would be 34/33/33 UPRO/TMF/IEF, which is pretty close to the 40/30/30 simulation MM ran on the previous page. So that's actually pretty close to risk parity between the anti-correlated parts.

There is no hand-down correct answer, of course, but I think it's a decent compromise.

Fair, although June was a fairly high equity allocation month (47% with just TMF on the other side). When I get some time I’ll run a 1-month look-back parity using a 50/50 between TMF and unleveraged 10-year; can only go back to 1987 but it’ll be interesting to see how it performs. Ill also see about simulating a pure EDV case which would probably be similar exposure. I did run a pure unleveraged LTT 1-month look-back and the results since 1987 were similar to 100% S&P but much less risk.

[MotoTrojan]

In order to maintain risk-parity with IEF you must hold significantly less equity.

True. But there is no "must" about maintaining risk parity (defined here as equal risk in each asset class).

So there's no must about less equity.

Totally agreed but at some point the bond side is no longer offsetting drawdowns in UPRO which is the point.

[schismal]

Fair, although June was a fairly high equity allocation month (47% with just TMF on the other side). When I get some time I’ll run a 1-month look-back parity using a 50/50 between TMF and unleveraged 10-year; can only go back to 1987 but it’ll be interesting to see how it performs. Ill also see about simulating a pure EDV case which would probably be similar exposure. I did run a pure unleveraged LTT 1-month look-back and the results since 1987 were similar to 100% S&P but much less risk.

Looking forward to the results!

[MoneyMarathon]

In order to maintain risk-parity with IEF you must hold significantly less equity.

True. But there is no "must" about maintaining risk parity (defined here as equal risk in each asset class).

So there's no must about less equity.

Totally agreed but at some point the bond side is no longer offsetting drawdowns in UPRO which is the point.

TMF on its own had drawdowns of around -50% in its short history. It's not offsetting drawdowns if it doesn't pop upwards enough at the right time. It would be more accurate to say that it has a chance of offsetting drawdowns in UPRO. Going for that chance is a very different strategy than going for the near-certainty that a less-volatile asset like IEF will hold its value. Both are valid, and one is shooting for higher risk / possibly higher reward than the other. Two-asset 3x leveraged UPRO/TMF isn't premised on managing risk more prudently and conservatively than the alternatives. It's about hoping for bigger gains and having a willingness to take greater risk for that potential.

Using 60% TMF is a riskier way (more dispersion of results / potential for both bigger upside and bigger downside) to try to preserve capital when UPRO has a drawdown. There seems to be a somewhat unjustified confidence here that it will do that job in the future, over the next 20-30 years. It did do very well in the two most recent bear markets (dot com, great financial crisis).

[MotoTrojan]

In order to maintain risk-parity with IEF you must hold significantly less equity.

True. But there is no "must" about maintaining risk parity (defined here as equal risk in each asset class).

So there's no must about less equity.

Totally agreed but at some point the bond side is no longer offsetting drawdowns in UPRO which is the point.

TMF on its own had drawdowns of around -50% in its short history. It's not offsetting drawdowns if it doesn't pop upwards enough at the right time. It would be more accurate to say that it has a chance of offsetting drawdowns in UPRO. Going for that chance is a very different strategy than going for the near-certainty that a less-volatile asset like IEF will hold its value. Both are valid, and one is shooting for higher risk / possibly higher reward than the other. Two-asset 3x leveraged UPRO/TMF isn't premised on managing risk more prudently and conservatively than the alternatives. It's about hoping for bigger gains and having a willingness to take greater risk for that potential.

Using 60% TMF is a riskier way (more dispersion of results / potential for both bigger upside and bigger downside) to try to preserve capital when UPRO has a drawdown. There seems to be a somewhat unjustified confidence here that it will do that job in the future, over the next 20-30 years. It did do very well in the two most recent bear markets (dot com, great financial crisis).

While UPRO/TMF is certainly more risky, there is something to be said for the concept of risk-parity and holding 40% UPRO w/ 30% IEF and 30% TMF isn't achieving that. I do agree that TMF increases overall risk when held in parity, thus why I am looking to dilute it with EDV to reduce overall leverage.

[MoneyMarathon]

In order to maintain risk-parity with IEF you must hold significantly less equity.

True. But there is no "must" about maintaining risk parity (defined here as equal risk in each asset class).

So there's no must about less equity.

Totally agreed but at some point the bond side is no longer offsetting drawdowns in UPRO which is the point.

TMF on its own had drawdowns of around -50% in its short history. It's not offsetting drawdowns if it doesn't pop upwards enough at the right time. It would be more accurate to say that it has a chance of offsetting drawdowns in UPRO. Going for that chance is a very different strategy than going for the near-certainty that a less-volatile asset like IEF will hold its value. Both are valid, and one is shooting for higher risk / possibly higher reward than the other. Two-asset 3x leveraged UPRO/TMF isn't premised on managing risk more prudently and conservatively than the alternatives. It's about hoping for bigger gains and having a willingness to take greater risk for that potential.

Using 60% TMF is a riskier way (more dispersion of results / potential for both bigger upside and bigger downside) to try to preserve capital when UPRO has a drawdown. There seems to be a somewhat unjustified confidence here that it will do that job in the future, over the next 20-30 years. It did do very well in the two most recent bear markets (dot com, great financial crisis).

While UPRO/TMF is certainly more risky, there is something to be said for the concept of risk-parity and holding 40% UPRO w/ 30% IEF and 30% TMF isn't achieving that. I do agree that TMF increases overall risk when held in parity, thus why I am looking to dilute it with EDV to reduce overall leverage.

Risk parity (going for equal or near-equal risk in each asset) strategies are more effective at reducing the risk of the portfolio when there are more uncorrelated assets. TMF and EDV are 99%-100% correlated (with both having high volatility) and thus the mix does almost nothing to reduce the level of risk, compared to just using TMF by itself. It's just moving deck chairs on the Titanic -- if the Titanic is fine, then you're fine, but if it's not you're sunk.

TMF, UPRO, and gold are uncorrelated (and other commodities are also less correlated), so adding gold or other commodities would actually reduce the overall level of risk of the portfolio.

[MotoTrojan]

Fair, although June was a fairly high equity allocation month (47% with just TMF on the other side). When I get some time I’ll run a 1-month look-back parity using a 50/50 between TMF and unleveraged 10-year; can only go back to 1987 but it’ll be interesting to see how it performs. Ill also see about simulating a pure EDV case which would probably be similar exposure. I did run a pure unleveraged LTT 1-month look-back and the results since 1987 were similar to 100% S&P but much less risk.

Looking forward to the results!

Anybody know any intermediate funds that have been around longer than 1992? VFITX has been around since then and actually returned a bit less than IEF during IEF's time, so let's call this a conservative approach.

Also for an overview of my methodology:
1: Run 1-month risk-parity of UPRO/TMF & UPRO/VFITX separately and save as a benchmark.
2: Use the volatility of UPRO/TMF/VFITX separately since inception of VFITX (1992) to determine the average allocation of UPRO to each.
3: Given these weights, determine what allocation between the two 1-month portfolios would on average hold equal amounts (in dollars) of each bond fund.
4: Run a 1-month rebalance of the two separate risk-parity funds using that allocation (I got 60.4% UPRO/TMF and 39.6% UPRO/VFITX)

Portfolio 1: UPRO balanced monthly w/ 50/50 TMF/VFITX (representing TMF/IEF; average equity exposure is 29% UPRO)
Portfolio 2: UPRO balanced monthly w/ 50/50 TMF/EDV (this uses the volatilities since inception of VEDTX so bit of error there but quite small)
Portfolio 3: OP's 40/60 UPRO/TMF rebalanced quarterly
Portfolio 4: S&P500

As expected it knocked down returns some but overall still a very efficient portfolio. Curious to know what siamonds spreadsheet (really need to download that...) would return from 1955-present w/ 29% UPRO, and 35.5% of TMF & IEF.

[MotoTrojan]

Risk parity (going for equal or near-equal risk in each asset) strategies are more effective at reducing the risk of the portfolio when there are more uncorrelated assets. TMF and EDV are 99%-100% correlated (with both having high volatility) and thus the mix does almost nothing to reduce the level of risk, compared to just using TMF by itself. It's just moving deck chairs on the Titanic -- if the Titanic is fine, then you're fine, but if it's not you're sunk.

TMF, UPRO, and gold are uncorrelated (and other commodities are also less correlated), so adding gold or other commodities would actually reduce the overall level of risk of the portfolio.

I don't think you understood my strategy; I am not holding TMF & EDV in parity with each-other, I am blending them into one asset and holding them in parity with UPRO. 50/50 TMF/EDV has less volatility than 100% TMF and thus when balanced monthly with UPRO you'll hold less UPRO and have less overall leverage (and risk). The fact that TMF and EDV are so closely correlated works in my favor since I am blindly averaging their volatility and assuming they don't offset eachother at all, just a way to target a specific amount of volatility on the bond side, and thus a balancing one on the equity side.

[MoneyMarathon]

Risk parity (going for equal or near-equal risk in each asset) strategies are more effective at reducing the risk of the portfolio when there are more uncorrelated assets. TMF and EDV are 99%-100% correlated (with both having high volatility) and thus the mix does almost nothing to reduce the level of risk, compared to just using TMF by itself. It's just moving deck chairs on the Titanic -- if the Titanic is fine, then you're fine, but if it's not you're sunk.

TMF, UPRO, and gold are uncorrelated (and other commodities are also less correlated), so adding gold or other commodities would actually reduce the overall level of risk of the portfolio.

I don't think you understood my strategy; I am not holding TMF & EDV in parity with each-other, I am blending them into one asset and holding them in parity with UPRO. 50/50 TMF/EDV has less volatility than 100% TMF and thus when balanced monthly with UPRO you'll hold less UPRO and have less overall leverage (and risk). The fact that TMF and EDV are so closely correlated works in my favor since I am blindly averaging their volatility and assuming they don't offset eachother at all, just a way to target a specific amount of volatility on the bond side, and thus a balancing one on the equity side.

I understand the strategy. I'm saying it has a very, very, very similar risk profile to just going with 40% UPRO / 60% TMF.

[MotoTrojan]

Risk parity (going for equal or near-equal risk in each asset) strategies are more effective at reducing the risk of the portfolio when there are more uncorrelated assets. TMF and EDV are 99%-100% correlated (with both having high volatility) and thus the mix does almost nothing to reduce the level of risk, compared to just using TMF by itself. It's just moving deck chairs on the Titanic -- if the Titanic is fine, then you're fine, but if it's not you're sunk.

TMF, UPRO, and gold are uncorrelated (and other commodities are also less correlated), so adding gold or other commodities would actually reduce the overall level of risk of the portfolio.

I don't think you understood my strategy; I am not holding TMF & EDV in parity with each-other, I am blending them into one asset and holding them in parity with UPRO. 50/50 TMF/EDV has less volatility than 100% TMF and thus when balanced monthly with UPRO you'll hold less UPRO and have less overall leverage (and risk). The fact that TMF and EDV are so closely correlated works in my favor since I am blindly averaging their volatility and assuming they don't offset eachother at all, just a way to target a specific amount of volatility on the bond side, and thus a balancing one on the equity side.

I understand the strategy. I'm saying it has a very, very, very similar risk profile to just going with 40% UPRO / 60% TMF.

I think there is a clear mathematical answer to the theoretical risk reduction. Parity with only UPRO/EDV is lower, parity with only UPRO/IEF would be further lower.

[MoneyMarathon]

Risk parity (going for equal or near-equal risk in each asset) strategies are more effective at reducing the risk of the portfolio when there are more uncorrelated assets. TMF and EDV are 99%-100% correlated (with both having high volatility) and thus the mix does almost nothing to reduce the level of risk, compared to just using TMF by itself. It's just moving deck chairs on the Titanic -- if the Titanic is fine, then you're fine, but if it's not you're sunk.

TMF, UPRO, and gold are uncorrelated (and other commodities are also less correlated), so adding gold or other commodities would actually reduce the overall level of risk of the portfolio.

I don't think you understood my strategy; I am not holding TMF & EDV in parity with each-other, I am blending them into one asset and holding them in parity with UPRO. 50/50 TMF/EDV has less volatility than 100% TMF and thus when balanced monthly with UPRO you'll hold less UPRO and have less overall leverage (and risk). The fact that TMF and EDV are so closely correlated works in my favor since I am blindly averaging their volatility and assuming they don't offset eachother at all, just a way to target a specific amount of volatility on the bond side, and thus a balancing one on the equity side.

I understand the strategy. I'm saying it has a very, very, very similar risk profile to just going with 40% UPRO / 60% TMF.

I think there is a clear mathematical answer to the theoretical risk reduction. Parity with only UPRO/EDV is lower, parity with only UPRO/IEF would be further lower.

Risk parity with UPRO / TMF / UGLD or some other uncorrelated third asset class also reduces the risk of the portfolio.

Not adhering to risk parity also has its merits. Not keeping "pure" exact-risk-matching risk parity has its merits.

There are three things here:

(1) Something like 40% UPRO / 40% TMF / 10% UGLD / 10% something else -- deviating from "pure" risk parity, reduced risks

(2) Something like 40% UPRO / 60% IEF -- abandoning the paradigm of risk parity, greatly reduced risks

(3) Something with lower UPRO / greater IEF -- keeping risk parity, greatly reduced risks

I keep bringing up something like (1) or (2), and you keep bringing up "but (3)" in response like there's only one right way to invest.

[schismal]

Also for an overview of my methodology:
1: Run 1-month risk-parity of UPRO/TMF & UPRO/VFITX separately and save as a benchmark.
2: Use the volatility of UPRO/TMF/VFITX separately since inception of VFITX (1992) to determine the average allocation of UPRO to each.
3: Given these weights, determine what allocation between the two 1-month portfolios would on average hold equal amounts (in dollars) of each bond fund.
4: Run a 1-month rebalance of the two separate risk-parity funds using that allocation (I got 60.4% UPRO/TMF and 39.6% UPRO/VFITX)

Portfolio 1: UPRO balanced monthly w/ 50/50 TMF/VFITX (representing TMF/IEF; average equity exposure is 29% UPRO)
Portfolio 2: UPRO balanced monthly w/ 50/50 TMF/EDV (this uses the volatilities since inception of VEDTX so bit of error there but quite small)
Portfolio 3: OP's 40/60 UPRO/TMF rebalanced quarterly
Portfolio 4: S&P500

As expected it knocked down returns some but overall still a very efficient portfolio. Curious to know what siamonds spreadsheet (really need to download that...) would return from 1955-present w/ 29% UPRO, and 35.5% of TMF & IEF.

Very interesting -- thanks for running that simulation! I was limited to the data on PV, but it seemed like UPRO/TMF/IEF won the Sortino for almost every interval I tested. Looks the same in your data. Both IEF and EDV seem to have dampened the volatility of TMF and improved the risk profile.

[MotoTrojan]

Risk parity with UPRO / TMF / UGLD or some other uncorrelated third asset class also reduces the risk of the portfolio.

Not adhering to risk parity also has its merits. Not keeping "pure" exact-risk-matching risk parity has its merits.

There are three things here:

(1) Something like 40% UPRO / 40% TMF / 10% UGLD / 10% something else -- deviating from "pure" risk parity, reduced risks

(2) Something like 40% UPRO / 60% IEF -- abandoning the paradigm of risk parity, greatly reduced risks

(3) Something with lower UPRO / greater IEF -- keeping risk parity, greatly reduced risks

I keep bringing up something like (1) or (2), and you keep bringing up "but (3)" in response like there's only one right way to invest.

I think we are in violent agreement. Totally agree with what you are saying, just wanted some affirmation that replacing some TMF with EDV was still a sizable reduction in risk, even if not as much as going all IEF.

[MoneyMarathon]

I think we are in violent agreement. Totally agree with what you are saying, just wanted some affirmation that replacing some TMF with EDV was still a sizable reduction in risk, even if not as much as going all IEF.

OK, cheers.

[MotoTrojan]

Also for an overview of my methodology:
1: Run 1-month risk-parity of UPRO/TMF & UPRO/VFITX separately and save as a benchmark.
2: Use the volatility of UPRO/TMF/VFITX separately since inception of VFITX (1992) to determine the average allocation of UPRO to each.
3: Given these weights, determine what allocation between the two 1-month portfolios would on average hold equal amounts (in dollars) of each bond fund.
4: Run a 1-month rebalance of the two separate risk-parity funds using that allocation (I got 60.4% UPRO/TMF and 39.6% UPRO/VFITX)

Portfolio 1: UPRO balanced monthly w/ 50/50 TMF/VFITX (representing TMF/IEF; average equity exposure is 29% UPRO)
Portfolio 2: UPRO balanced monthly w/ 50/50 TMF/EDV (this uses the volatilities since inception of VEDTX so bit of error there but quite small)
Portfolio 3: OP's 40/60 UPRO/TMF rebalanced quarterly
Portfolio 4: S&P500

As expected it knocked down returns some but overall still a very efficient portfolio. Curious to know what siamonds spreadsheet (really need to download that...) would return from 1955-present w/ 29% UPRO, and 35.5% of TMF & IEF.

Very interesting -- thanks for running that simulation! I was limited to the data on PV, but it seemed like UPRO/TMF/IEF won the Sortino for almost every interval I tested. Looks the same in your data. Both IEF and EDV seem to have dampened the volatility of TMF and improved the risk profile.

You're welcome! I mentioned this before but I don't think there is a significant boost from the correlation differences between IEF & LTT/TMF so I actually do not see much value in using IEF unless you go 100%.

For example if you wanted to emulate the 40/30/30 UPRO/TMF/IEF using TMF & long-treasuries (VUSTX) you could do 40/22/38 UPRO/TMF/VUSTX; I arrived at these weights by simply using the volatility of the funds since 1992 (data available for all funds) and adjusting the weight of TMF/VUSTX to equal the volatility of 50/50 TMF/IEF.

The 40/30/30 w/ IEF had a CAGR of 11.83% from 1955-2018 w/ 24.70% Stdev.

The 40/22/38 w/ VUSTX and less TMF had a CAGR of 11.77% w/ 24.30% Stdev; the rest of the stats were quite similar as well.

I'd feel more comfortable with the 2nd portfolio as it holds less of TMF, which is more of a wildcard (issues tracking index, borrowing costs, volatility decay, etc...) and the returns/risk are quite similar between the two regardless of environment.

I think the real key is to pick a duration risk-exposure you are comfortable with and then use the funds to achieve that with as little TMF as possible (40/60 UPRO/EDV is close to the above portfolios for example w/ slightly more duration risk).

[MotoTrojan]

Very interesting -- thanks for running that simulation! I was limited to the data on PV, but it seemed like UPRO/TMF/IEF won the Sortino for almost every interval I tested. Looks the same in your data. Both IEF and EDV seem to have dampened the volatility of TMF and improved the risk profile.

I did want to make sure this was crystal clear though; the UPRO/TMF case shown is the OP's quarterly rebalance of 40/60 while the other two use the 1-month look-back. In my earlier post I showed the UPRO/TMF w/ 1-month look-back and it also has great efficiency and the best of any returns; the main thing that adding IEF or EDV does is reduce overall risk-exposure/leverage (and boost returns in 1955-1982 due to less drag).

Here is a comparison of the original quarterly 40/60 w/ the 1-month risk-parity again for ease (this goes back to 1987 though, above was 1992):

[mickens16]

Can one of you fine people explain how you go about doing the 20-day look back?

Thanks!