Gyroscopic Investing

Wonk wrote:

Gosso wrote: Wonk, something seems wonky with your 2013 numbers.  I track a 3x33% of the 1x, 2x, and 3x PPs using Google Finance.  Here are the 2013 results:

1x = -3.8%
2x = -8.3%
3x = -11.3%

All of the portfolios were rebalanced in mid-May, and include interest and dividends.  Maybe your portfolio rebalanced at a better time than mine?  I doubt SHY would make that much of a difference...

I think it has to do with rebalancing. 

The 2x PP triggered a rebalance after 2010 & 2011, but not after 2012.  At the beginning of 2013, the stock portion was just under 30% of the portfolio so it was overweight.  With the strong year in stocks, it pulled the other assets up.  By the end of 2013, stocks represented 44% of the porfolio and triggered another rebalance.  Rebalancing was end of year only, so that might have an influence on the numbers as well.

Wonk,

Just to be clear on the rebalancing issue - There are basically three schools of thought as to how often and when to rebalance a portfolio:

One, rebalance at the end of every year back to the original starting percentages regardless of how well/poorly each component has done and how much of the overall portfolio each component makes up before the rebalance.

Two, rebalance when any one component gets outside of a certain "band" (in this case more than 35% or less than 15%) and immediately do it whenever this violation of the rebalancing bands occur, even if it means several rebalancings per year if necessary.

Three, rebalance only at the end of the year if and only if any one of the components is above/below the rebalancing band limits (again, 15/35 in this case) as of the 31st of December of that year; otherwise, do not rebalance at all and simply wait one more year (until the 31st of December of next year) to see--using the test above--whether rebalancing is warranted, and so on.

I am presuming (perhaps incorrectly) that the UGL/UBT/SSO/SHY portfolio is rebalanced as per #3 above?

D1984 wrote: I am presuming (perhaps incorrectly) that the UGL/UBT/SSO/SHY portfolio is rebalanced as per #3 above?

Yes, that's correct.  Although, the portfolio has a start (& rebalance) date in January instead of December.  If the 15/35 bands are not breached, no rebalance takes place.

modeljc wrote: I would have thought decay would have shown up in four years of returns?  The annual 15/35 rebalance seems to help.  Appears to have a 13.8% annual return.

From what I can see, decay happens in isolation.  In a leveraged portfolio with equally non-correlated assets, they seem to cancel each other out.  Pure speculation though...

Wonk wrote:

modeljc wrote: I would have thought decay would have shown up in four years of returns?  The annual 15/35 rebalance seems to help.  Appears to have a 13.8% annual return.

From what I can see, decay happens in isolation.  In a leveraged portfolio with equally non-correlated assets, they seem to cancel each other out.  Pure speculation though...

Decay is also a function of volatility.  For example, if GLD goes down 10%, then up 10%, repeated every day, you will have a lot of decay.  If it just goes down .00001% every day, you will have no decay.

edit:  so my point is maybe daily volatility has been relatively low.

dragoncar wrote: Decay is also a function of volatility.  For example, if GLD goes down 10%, then up 10%, repeated every day, you will have a lot of decay.  If it just goes down .00001% every day, you will have no decay.

edit:  so my point is maybe daily volatility has been relatively low.

I agree that it is all about volatility.

I was comparing the returns of SPY and UPRO over the past 12 months and noticed that UPRO has performed greater than 3x the SPY:

SPY = 20.1% (+2% Dividend) x3 = 66.3%
UPRO = 75.5%

The reason of course is because UPRO maintains a constant 3x leverage and is reset daily.

I'm considering converting my Traditional IRA portfolio to a 2x PP (technically a 2.66x PP) using 2x ETFs, at least right now when interest rates are low:

33% SSO
33% UGL
33% UBT

I won't do this in my taxable accounts.  First, I can't touch this money until I'm 59.5 without penalties so I might as well take higher risks with it.  Second, it's gonna require more frequent rebalancing which will not be tax efficient anyways.

My tentative rules:

1) When any of the 2x ETFs deviates more than 10% from its target weight ( <23% or > 43%), a rebalance is required.

2) A rebalance is also required every quarter (3 months) even if the rebalancing bands aren't triggered.  In my backtesting, I found that frequent rebalancing helps mitigate some of the tracking error but rebalancing too frequently may not be practical due to the commissions and slippage.

3) In #1 and #2, whenever a rebalancing event is triggered, you have the option of reverting to the 25/25/25/25 standard HP PP instead.

blackomen wrote: I'm considering converting my Traditional IRA portfolio to a 2x PP (technically a 2.66x PP) using 2x ETFs, at least right now when interest rates are low:

33% SSO
33% UGL
33% UBT

I won't do this in my taxable accounts.  First, I can't touch this money until I'm 59.5 without penalties so I might as well take higher risks with it.  Second, it's gonna require more frequent rebalancing which will not be tax efficient anyways.

My tentative rules:

1) When any of the 2x ETFs deviates more than 10% from its target weight ( <23% or > 43%), a rebalance is required.

2) A rebalance is also required every quarter (3 months) even if the rebalancing bands aren't triggered.  In my backtesting, I found that frequent rebalancing helps mitigate some of the tracking error but rebalancing too frequently may not be practical due to the commissions and slippage.

3) In #1 and #2, whenever a rebalancing event is triggered, you have the option of reverting to the 25/25/25/25 standard HP PP instead.

blackomen, do these etf's have enough data or is there a way to backtest further?  If two of them go bad at once, can be ugly.

I'm just curious

Yes but just for the time period since 3x ETFs have been around. I don't have my computer with me but if I recall you get about 4% more per year with a good jump in volatility and drawdown as compared to the 2x version. Most academic studies on leverage indicate much over 2x leverage is not beneficial over the long haul. The short backtest seems to line up with the academics. I personally use the 2x version for my PP and have been quite happy with it. For me I know there's no way I could stick with a regular PP during a long bull run and the leveraged version hangs plus some.

Did some backtesting (2011 to now only)...at least for the past 3 years and change periodic rebalancing does a much better job of improving CAGR over band rebalancing. I was a bit surprised by this.

Kbg, what kind of results did you get with our backtest?

Thanks,

Using 2x ETFs

Rebalance bands from +5% to 15%: range 7.2% to 10.43%.

Monthly to annual: range 8.6% to 10.9%

Drawdowns are comparable but slightly improved by periodic rebalancing. Interestingly, adding XIV and allocating to 5 equal parts of 20% doubles CAGR with only a slight increase in DD for periodic rebalancing. I highly recommend at least a dash of XIV even if you did perhaps a 4 or 8% allocation with 24/23% for the rest.

For a taxable acct I think I'd just stick with annual rebalancing.

The much better comment was a bit in error. It is much better if you have an XIV component. Obviously the impact is not as much for the bands vs. Periodic if XIV is not in the mix.

Here is the backtest using XIV w quarterly rebalance:

[img width=870 height=855]https://dl.dropboxusercontent.com/u/28352540/PPwXIV.PNG[/img]

Quarterly rebalance no XIV:

[img width=870 height=855]https://dl.dropboxusercontent.com/u/283 ... P%202x.PNG[/img]

SPLX 2x SP500 is a new ETN that does monthly resetting instead of daily. I'm not aware of any other monthly reset leveraged ETFs appropriate for PP.

So maybe at least you could replace 1/3 of your buckets with an ETN that's not subject to the daily decay problem.

Then again, maybe there's more counter party risk with an ETN vs. ETF.

What's the long-term rationale for holding XIV?

We've seen a period of record low volatility, which has helped XIV the past few years, but if we have another 2008, right when all this leveraged ETF paper starts to enter uknown territory, XIV would spike down, right when you need it least.

What am I missing? How is XIV not just curve fitting?

I posted on this in another thread...but the short answer is while things are good they are very good so good that it can make up for REAL bad plus quite a bit more.

Heck we don't even need a 2008 scenario to get crushed. 2011 tossed us a 78% loss in just a couple of months. 30%+ loses over a quarter are routine. The 913% percent return since 2011 has been quite nice though.

Really, I would not want anyone to invest in XIV without an in-depth understanding of what they are buying. For those in the advanced class do some math, explore various rebalancing and risk sizing approaches and let the numbers and your own risk tolerance be your guide.

I'm going to move my PP with an XIV Twist post to here since it isn't technically a PP but rather a VP. I will pick up the 2x reporting as well as a comparison.

VP 20% to each

DGP = Gold
SHY = Cash
UBT = Long Bonds
UWM = Stocks (Russell 2000)
XIV = Short Volatility

2014 2xPP w/XIV Twist Performance YTD: 17.49%
  - 7.74% Max DD

2014 2xPP Performance YTD: 11.93%
  - 5.99% Max DD

It was a very good quarter with everything up.

SHY + 0.14%
UWM + 2.83%
DGP + 6.71%
UBT + 9.31%
XIV + 41.27%

S&P 500 is up 6.76% by comparison. I would think next quarter's report won't be so good.

A quick mid-quarter update.

Using 7/1 & 8/15 closing prices - 2.42% with a max DD of 7.22%.

UBT and UWM have essentially cancelled each other out at about +/- 10.5% respectively. XIV is down a little over 8% while DGP is down almost 4%. SHY is a skosh positive.

I got lucky on timing and bought into...

TMF (+43%)
UPRO (+31%)
UGLD (-9%)

...back in Feb. I am up about 20% as a group. Looking forward to an equally exciting reverse at some point.

My PP is a mix of 1x, 2x  and these 3x funds (sorry for the leverage Harry) so far I am up 15% for the year :-0. And why did I come to the PP six years ago? To avoid volatility...oops

phil

Third quarter results:

VP 20% to each

DGP = Gold
SHY = Cash
UBT = Long Bonds
UWM = Stocks (Russell 2000)
XIV = Short Volatility

2014 2xPP w/XIV Twist Performance YTD: 9.59%
  - 8.90% Max DD

2014 2xPP Performance YTD: 5.75%
  - 6.92% Max DD

Not so good this quarter!

SHY + 0.01%
UWM - 16.67%
DGP - 18.00%
UBT + 7.27%
XIV - 17.10%

S&P 500 is up 6.70% by comparison.

Just for fun let's do an update to capture the recent "unpleasantness."

Assets: IWM, TLT, SHY, GLD (Classic PP w/small cap) and XIV (PP w/XIV Twist)
2x as above only 2x ETFs
3x as above only 3x ETFs

1x PP 4.71/-3.69 DD
2x PP 9.87/-6.92 DD
3x PP 14.60/-9.91 DD

1x Twist 4.43/-10.85 DD
2x Twist 8.39/-12.85 DD
3x Twist 11.99/-14.73 DD

SPY 6.76/-9.93 DD. Note the 3x PP...over double the performance with basically the same drawdown as SPY. Who knows what the market will do next, but if we go flat or up the Twist version is likely to really start piling on some outperformance compared to the standard PP.

Kbg wrote: Just for fun let's do an update to capture the recent "unpleasantness."

Assets: IWM, TLT, SHY, GLD (Classic PP w/small cap) and XIV (PP w/XIV Twist)
2x as above only 2x ETFs
3x as above only 3x ETFs

1x PP 4.71/-3.69 DD
2x PP 9.87/-6.92 DD
3x PP 14.60/-9.91 DD

1x Twist 4.43/-10.85 DD
2x Twist 8.39/-12.85 DD
3x Twist 11.99/-14.73 DD

SPY 6.76/-9.93 DD. Note the 3x PP...over double the performance with basically the same drawdown as SPY. Who knows what the market will do next, but if we go flat or up the Twist version is likely to really start piling on some outperformance compared to the standard PP.

Just a quick thank you, kbg, for your introduction to XIV and for your abject honesty in the highs and lows of your schemes.  I have learned a lot from all of your posts.

As a quick aside...

Using XIV, you can simulate a 2x PP without Leverage.

10 year backtesting (huge alarm here!!!) shows a 20% max draw down, and 20% CAGR for:

50% TLT
35% GLD
15% XIV

(N.B. This backtesting requires the purchase of simulated historical values for XIV).

If you compare the above portfolio with the PP - this is the same maximum Draw Down with double the CAGR.
Even the day to day movements are nearly always 2x plus and minus.

Please don't try this at home.  Obviously a 10 year backtesting using simulated data is highly suspect in multiple ways.
I wouldn't bet my future on it, but it is fascinating to consider the highly volatile XIV as a proxy for the first derivative of equities.

And I mostly wanted to use this post to thank kbg for forcing me to examine areas of the market which I had not considered.

Again, don't do this at home.  I am not recommending it.  I am just thanking kbg for his insights.

[img width=600]http://i61.tinypic.com/2qd03mw.png[/img]

In the above chart, rebalance bands were set to be whenever XIV halved or doubled from 15%.
Black is total portfolio value.
Blue is TLT
Gold is GLD
Red is XIV
Green is accumulated cash from TLT dividends
Purple verticals are rebalance events
Thin black exponential line is a best fit CAGR ~= 20%
Thin gray horizontal line is maximum draw duration.  Largest ~= 18 months.


I haven't updated the chart in a couple of months.  It could use an update.

- updated - fixed multiple typos.

- And another update...

In the interest of full disclosure:

Upon reviewing my spreadsheet, rebalance bands were set for:
XIV = 7.5%: 15% : 30.0%
GLD = 17.5% : 35% : 50%
TLT = 30% : 50% : 70%

There were a few times when the rebalance was triggered by a different asset than XIV.
Again... don't trust curve fitted backtesting.  Smoking causes cancer.  Don't drink while driving.  Kiss your babies.

Excellent Mark! There are some interesting backtests out there that advocate a Cash or Bonds and XIV "Barbell Portfolio" meaning super safe and super risky/volatile with periodic rebalancing.

And now for the fun...

2x w/Twist +13.70%
3x w/Twist +19.79%   

Mid 4th quarter update...not sure why the SPY DD is different than posted earlier. A data issue likely. This has not been a good year for volatility or small caps. Just swapping UWM out for SSO bumps up the 2xPP to 13.7/-5.5 DD and the twist to 13.99/-10.75 DD. In any event it has been a good year for large caps.

Given the stability of the PP I think the start of this thread as well as the updates since makes a relatively good case for doing a leveraged version, particularly if you are in your asset accumulation years. There is a good case for putting in a dash of XIV/SVXY as well even for a conservative person...maybe like 5% out of your stock allocation. If one were to chose this route the volatility component can not be looked at (ever) in isolation of the over portfolio construction. I'm a late comer to PP but I like it a lot but also agree with William Bernstein as noted in the thread in the regular section...when the stock market is doing well for an extended period of time can you really stick with a PP? I think this version enables me to as I know it is going to be competitive with good market (SPX) years and absolutely trash it in bad market years.

1x PP 4.15/-3.70 DD
2x PP 8.78/-6.92 DD
3x PP 13.12/-9.91 DD

1x Twist 6.41/-10.85 DD
2x Twist 10.33/-12.87 DD
3x Twist 13.42/-14.76 DD

SPY 11.64/-7.7 DD

P.S. If you are wondering why the returns dropped since the last post with an up market since then ...I'm pretty sure I was looking at the annualized CAGR and typed that instead of the percent return thus far.

Really interesting stuff -- keep it coming!  I've been doing a low-leveraged PP where I invest my expected savings for the next year at the beginning of each year.  Ends up only being like 1.25x leverage, but interested in alternative implementations.