20% annual returns over 40 years...interested?

[MediumTex]

I can't imagine being around in 1981 to see both P/E ratios and bond yields at those rates.

Pension plans could have totally neutralized their funding issues back then by simly buying 100% 30 year treasuries (projected pension plan asset returns were 8-10% and LT treasuries were yielding way above that) and holding them to maturity.

Even though this would have been simple and easy (and would have led to dramatic capital gains in addition to a very nice dividend stream in coming decades), as I understand it virtually no one took advantage of this opportunity because bonds were perceived to be so risky.

[clacy]

Clive,

For a leveraged PP, would it make sense to use tighter rebalance bands for more volatility capture?  Instead of 40% bands, maybe 30% would be of more benefit because of leveraged nature of a 2x or 3x?

[D1984]

If you look at the PP gains 1981 onwards, it lagged STT - by more than -25% at one point based on yearly values (perhaps more on a more frequent review basis) i.e. the gains from the leveraged PP levels wouldn't have been large enough to even compensate for the cost of borrowing to create that leverage.

Perhaps not exceptionally bad however as the cost of borrowing (cash rates) were very high in the 1980's.

Clive, perhaps the "cost of borrowing" wouldn't have effected a leveraged PP as much as you think. The PP lagged STT total returns from 1981 to 2005 but since we are using 1-3 year Treasurys and the total return includes capital gain/loss as well as interest then the results would be different if you assume borrowing costs were just based on interest rates alone. For instance, average 3-month T-bill rates + 1.0% (used as a proxy for money market borrowing costs) in these years were as follows:

1981 = 15.08%

1982 = 11.71%

1983 = 9.63%

1984 = 10.56%

1985 = 8.48%

1986 = 6.98%

1987 = 6.84%

1988 = 7.67%

1989 = 9.12%

1990 = 8.51%

1991 = 6.42%

1992 = 4.46%

1993 = 4.02%

1994 = 5.29%

1995 = 6.51%

1996 = 6.02%

1997 = 6.07%

1998 = 5.81%

1999 = 5.66%

2000 = 6.82%

2001 = 4.40%

2002 = 2.60%

2003 = 2.01%

2004 = 2.38%

2005 = 4.15%

All but about seven or eight of these years had lower costs than the total return on STTs (the five with significantly higher costs than STT returns were years like 1983, 1987-1989, 1994, 1996, and 1999, and 2005...all rising rate years where STT total returns were hurt by capital losses on 1-3 yr T-bills) but almost all the other years showed significantly lower numbers than the total return figures (in some cases hugely lower numbers during falling STT rate years like 1985-86, 1991-1993, 1995, and 2001-2002)

[amp]

This post and the one Clive made above it make sense to me. In a low interest rate environment, where is the downside?  Are there other factors in the decay besides borrowing costs for leverage?

Don't underestimate the power of tracking error to ruin you.  Check out this chart:



Green is spot.  Magenta 2x.  Turquoise 3x.

Unfortunately, the charting period is only limited to 3 months, but longer term the effects of decay and tracking error is absolutely devestating.

MG

Help me out a bit here, I'm not seeing the issue.  If spot (green) is down 39% and 2x (magenta) is only down 65% (less than 39% x 2), isn't that a good thing?  What am I missing?

Thanks,
amp

[Wonk]

Watching the 1x, 2x & 3x PP today was pretty cool.  The 3x portfolio had some pretty righteous moves. UGLD at -15.67% was the big story.  Overall the 3x was down 5% today compared to the 1x at 1.65%.  A little bit of tracking error, but pretty close.  Hopefully we see a major meltdown of some sort to see how they react.

[systemskeptic]

Anyone actually end up putting some money into a 2x PP? 

So far 33% each of SSO/UBT/UGL is up 25.1% YTD with a 1.53 Sharpe ratio.

[MediumTex]

Anyone actually end up putting some money into a 2x PP? 

So far 33% each of SSO/UBT/UGL is up 25.1% YTD with a 1.53 Sharpe ratio.

We will know when someone actually tries this because it will promptly stop working.

[clacy]

When the central banks stop with QE, the 2xPP strategy will likely turn into a loser, IMO.

Up until then, I'm guessing there is an edge.  The problem is if you pull the trigger just as the effect of the latest round of QE is wearing off, and another round is not announced, then you could be in for some pain.

[systemskeptic]

When the central banks stop with QE, the 2xPP strategy will likely turn into a loser, IMO.

Up until then, I'm guessing there is an edge.  The problem is if you pull the trigger just as the effect of the latest round of QE is wearing off, and another round is not announced, then you could be in for some pain.

Is this saying the PP will perform poorly if QE stops, or just that the borrowing rates will rise too high?  What could be used to hedge a 2x PP position?

[clacy]

When the central banks stop with QE, the 2xPP strategy will likely turn into a loser, IMO.

Up until then, I'm guessing there is an edge.  The problem is if you pull the trigger just as the effect of the latest round of QE is wearing off, and another round is not announced, then you could be in for some pain.

Is this saying the PP will perform poorly if QE stops, or just that the borrowing rates will rise too high?  What could be used to hedge a 2x PP position?

I guess I feel like if the PP goes sideways or down for some period of time, due to borrowing costs and decay from daily rebalancing, a 2x version would be hammered beyond 2x the draw down.

[clacy]

Right now, all assets are performing up or flat.  I think that's because of QE.

When QE stops, I believe at a minimum of 1 asset class, and more likely 2 asset classes, with a possibility that 3 asset classes will go down significantly.

Of course what I believe doesn't necessarily mean much, which is why I stay as hedged as possible with the PP.

[Gosso]

Anyone actually end up putting some money into a 2x PP? 

So far 33% each of SSO/UBT/UGL is up 25.1% YTD with a 1.53 Sharpe ratio.

I only have 17% YTD for the 2x PP, but I'm a little low on UBT.  For the 3x PP I have 9.5% since Feb 9, 2012 (this was when 3xTLT was introduced).  The standard 4x25 PP is up 6.7% YTD.

The only way a levered ETF PP will work over the long term is if you apply a trend following strategy which lowers volatility and drawdowns (otherwise you eventually fall back to the 1x PP).  This of course creates the hassle of monitoring the 200 day moving average, then having to trade and pay the commish, spreads, and taxes.  There is also the risk of excessive whipsaw, but this can be blunted through applying a buffer.  Then there is the mental stress of selling an asset after it has fallen for a few days/weeks, and buy an asset that has risen for a few days/weeks.  It helps to be somewhat emotionless.

Here's a fun spreadsheet that allows you to backtest moving average strategies:

http://www.financialwebring.org/gummy-s ... day-MA.htm

Overall I think it's possible to squeeze an extra few percentage points out of the PP using LETF's and a trend following strategy but with higher volatility. 

As an example I have backtested the following from 1972 to 2012 (rebalance annually):

-> 50% 1x Gold / 50% 1x S&P500 = 10.8% CAGR and 16% Stdev

-> 50% 3x Gold / 50% 3x S&P500 = 13.2% CAGR and 51% Stdev

-> 50% 3x Gold / 50% 3x S&P500 (with trend following of the 200 day SMA, fees/taxes are excluded) = 21.8% CAGR and 42.6% Stdev

-> 50% 3x S&P 500 / 25% 3x Gold / 25% Cash (with trend following of the 200 day SMA, fees/taxes are excluded) = 18.1% CAGR and 27.2% Stdev

For more info I'd recommend reading through Mebane Faber's white paper "A Quantitative Approach to Tactical Asset Allocation" (page 28 discusses using leverage).

[clacy]

I would be interested in implementing some sort of leverage scale based on draw down.  So if the standard 4x25 HBPP draws down 10%, you add 25-50% leverage (so using 1.25-1.5x leverage at that point).  Then if the standard portfolio experiences another 10% decline, you increase leverage to 1.5-1.75x).

It would be interesting to hear how that worked over the past 40 years.  I'm not equipped to backtest something that elaborate, but I'm sure many here are.

[escafandro]

Anyone actually end up putting some money into a 2x PP? 

A little amount since mid-December 2011.
25/25/25/25 - SSO / UBT / UGL / CASH
13.84% up to this day

[MediumTex]

Anyone actually end up putting some money into a 2x PP? 

A little amount since mid-December 2011.
25/25/25/25 - SSO / UBT / UGL / CASH
13.84% up to this day

Congratulations!

[Gosso]

I would be interested in implementing some sort of leverage scale based on draw down.  So if the standard 4x25 HBPP draws down 10%, you add 25-50% leverage (so using 1.25-1.5x leverage at that point).  Then if the standard portfolio experiences another 10% decline, you increase leverage to 1.5-1.75x).

It would be interesting to hear how that worked over the past 40 years.  I'm not equipped to backtest something that elaborate, but I'm sure many here are.

I was thinking of doing this as well, but then I worried about being able to pick the bottom and top.  I think it'd be easier to follow a simple system of using the 200 day moving average.  This takes all the guess work out of it.

Have a look at this:



The blue line = 50% 1x Gold / 50% 1x S&P500 = 10.8% CAGR and 16% Stdev
The green line = 50% 3x Gold / 50% 3x S&P500 = 13.2% CAGR and 51% Stdev
The orange line = 50% 3x S&P 500 / 25% 3x Gold / 25% Cash (with trend following of the 200 day SMA, fees/taxes are excluded) = 18.1% CAGR and 27.2% Stdev

I forgot to mention in the previous post that I have also subtracted the borrowing costs (1 year treasury rate (times two for 3x ETF's)) from the levered ETF's.

Based on this chart it seems we are due for some sort of crash in the near future to bring the levered players back to the B&H...but I may be overly pessimistic.  I suppose it is possible we see another massive spike in gold, which would be fantastic for the levered ETF's...until it crashes.

[Gosso]

Clive, I seem to recall you being a fan of Mebane Faber, do you still feel that a trend following system can work (sort of putting you on the spot )?  With my backtesting I have been able to reduce the trades down to once a year for each asset by applying a +/-3% buffer on the 200 day moving average, and only checking at the end of each month (or end on the week works well too).  This maintains the overall return and volatility while decreasing the number of trades.  Then leverage can be applied to match the desired return and volatility.

Keep in mind that this is for the VP.  Although I'm starting to feel more comfortable in a trend following system over a buy & hold strategy...but this is because I know I have the discipline to follow it (I hope ).

[systemskeptic]

The trick with leveraged ETF's is to use them not to attempt to scale up rewards, but to scale down risk.

Clive,

Does something intrinsic break when you use the 2x as a full part of your portfolio?  I have read some of your posts where you show 50% in a 2x and 50% in cash matches the 1x pretty closely.  Here are some of my comments.

1.  From a risk perspective, I am okay with an asset allocation that has an expected return:worst year return ratio of a little over 1:1 while also having no single year worse than -15%.  The PP at 9.65% CAGR and -4.13% worst year is already 2:1 (Compare to stocks at 9.92% CAGR 37.04% worst year, or a 1:4 ratio).  Using leveraged ETFs the ratio changes only slightly due to decay/borrowing costs (if everything is working properly) but it does double the worst year.

2.  If you use a 50% 2x PP + 50% cash, what do you do with the remaining cash?  If we assume this allocation is the best for risk/reward, we would take the remaining amount and put 25% in the 2x PP and 25% in cash -- and so on until all the cash is gone.  Ultimately what you would be left with is a 100% 2x PP.

3.  One situation where I can imagine this failing would be if the leveraged paper market blows up in a black swan event.  For this, I would probably want to hedge with an additional allocation of physical gold.

[clacy]

One thing that I've thought about and tested on etfreplay is something like this:

20% DEF (Guggenheim Defensive Equity ETF).... fairly low volatility stock ETF
5% SPXL (3x S&P)

20% LTPZ (Long term TIPS ETF)... much lower volatility than gold
5% UGL (2x gold)

20% IEF (7-10 year treasury ETF)
5% TMF (3x LTT)

25% SHY

This portfolio had a little better total return and almost 1% better CAGR, with lower draw down and volatitlity.  Also better sharpe than compared with a standard HBPP.

This would give you 85% of your assets in solid, low volatility assets.  But you have the 3x versions to capture upside.

This portfolio only dates back to Sept of 2009, so there isn't a lot of data.  Not to mention how would the 3x ETF's do in a fall of 08/spring of 09 type scenario.

It does give you a little leverage, but pretty limited downside risk.  It's not a lot different than running a 90% standard PP and then going into 3x ETF's for the VP.

I would like to see how this performs in a 10-15% DD of the standard version however....

[Khisanth]

I'm going to put my money where my mouth is here, and try the 3xETF PP out. Purchases done today.

10% TMF @ 93.3522 73.2906
10% UPRO @ 73.2906 93.3522
10% UGLD @ 50.7389

70% SHV

Rebalance bands for this will be 7% or 13% 6% and 14%.

Thanks for the great thread here! We'll see first/second hand how the leverage and decay track.

EDIT: Mixed up prices of TMF and UPRO

EDIT Oct 31 2012 End of Day:

Code: Select all

       Gain/Loss   Port%    NoCash%
Cash     0.00%   72.21%
UPRO    -8.96%    8.88%     31.94%
UGLD    -8.85%    8.89%     31.99%
 TMF     2.38%   10.02%     36.07%

TOTAL   -1.48%

[dragoncar]

I've been watching the 2x verion and the 2x treasuries don't seem to be doing very well at hitting the full 2x... seems you'd have to overweight it.  Does the 3x have a similar problem?

[dragoncar]

I mean that, even on a daily basis, the 2x treasury only appears to be hitting 1.5x TLT.

[Gosso]

I mean that, even on a daily basis, the 2x treasury only appears to be hitting 1.5x TLT.

It's because UBT only trades 5 or 6 times a day, so the price swings are delayed and won't match TLT on a minute by minute basis.  However, if you look at the bid/ask price for UBT then you'll see that its 2x TLT.  Also, if you zoom out to a few months then you'll see that UBT is pretty close to 2x TLT.

[clacy]

I mean that, even on a daily basis, the 2x treasury only appears to be hitting 1.5x TLT.

Maybe look into TMF (3x LTT etf). It has 5-10x the volume of UBT

[Wonk]

It's been 1 year since I tracked all three portfolios together.  Here are the results as of today:

1xPP: 6.42%
2xPP: 11.26%
3xPP: 13.53%

The sideways action of the first 6 months was interesting to observe.  There wasn't a decay problem that seemed to be a predominant fear going in.  The 3x did exhibit what Clive had guessed--more of a 2x to 2.5x return rather than 3x.  Before the PP beatdown of the last few weeks, the 2x was almost exactly 2x the 1x and the 3x was about 2.50-2.75x the 1x.

On the longer timeline, I started tracking the 1x vs 2x PP on Jan 21, 2010.  Here are the results to date:

1xPP: 31.75%
2xPP: 77.41%

I'd be curious to see what the returns will be once we get back to positive real interest rates where the borrowing environment is not as desirable.  From the looks of things, we might still be 5 years out.