Gyroscopic Investing

Someone here, I believe Slotine or  melveyr, has done some work showing how one can introduce pseudo-leverage into the PP by simply choosing more volatile funds, an example being EDV over TLT.

More volatile cash doesn't make sense so I guess you could probably remove that.

As for gold, I could see the argument that gold itself is already hyper-volatile and that no tweak is necessary, but what about stocks? Maybe a small-cap-cap value fund? That gives you a bit more volatility, but breaks the sector neutrality of a total stock index.


Man, what a crazy PP you would have with EDV, VBR, and GLD.

Pointedstick wrote: Someone here, I believe Slotine or  melveyr, has done some work showing how one can introduce pseudo-leverage into the PP by simply choosing more volatile funds, an example being EDV over TLT.

More volatile cash doesn't make sense so I guess you could probably remove that.

As for gold, I could see the argument that gold itself is already hyper-volatile and that no tweak is necessary, but what about stocks? Maybe a small-cap-cap value fund? That gives you a bit more volatility, but breaks the sector neutrality of a total stock index.


Man, what a crazy PP you would have with EDV, VBR, and GLD.

For what you just laid out, you might also want to add in a sprinkle of some silver. It has a little more kick than gold. I don't think GLD has a volatility on par with EDV or VBR (or just bump up the GLD weighting).

Also I have found it very useful to think of cash as negative leverage, or leverage as negative cash... It makes sense because most people pay for leverage by borrowing at the short term rates, and contrarily cash earns interest at the short term rates.

So if you are levering up a portfolio, the lowest hanging fruit is to remove cash (as you did).

No one really responded to this thread of mine but you might find it interesting. It explores how the PP doesn't have to be a one size fits all strategy (a weakness in the strategy from my perspective):
http://gyroscopicinvesting.com/forum/ht ... 148#p50148

As for gold, I could see the argument that gold itself is already hyper-volatile and that no tweak is necessary,

What about 50/50 mix of a 2X daily gold fund and a gold mining shares fund for "volatile gold"? How much more volatile is gold (as measured by SD) than the other two assets?

but what about stocks? Maybe a small-cap-cap value fund? That gives you a bit more volatility, but breaks the sector neutrality of a total stock index.


Man, what a crazy PP you would have with EDV, VBR, and GLD.

Could anyone really have stomached holding this PP from 1996 to 1999? Using DFSVX for small value (I don't think VBR existed back then), and BTTRX (with BTTTX for 1996 since BTTRX only started in February of that year) for zeros, I get the following returns for those years for a 33/33/33 mix of SV, gold, and zeroes

1996 = 2.97%

1997 = 13.05%

1998 = 4.43%

1999 = -2.42%

The zeroes smacked you in 1996 and 1999; the gold hurt you big time in 1997, and the SV killed you in 1998....all of this when the broad market as a whole was giving 20%+ returns each year (and let's not even get into what large growth or the NASDAQ were doing). You actually woud have LOST money in 1999. Honestly, I think you'd have done better these four years in CDs or t-bills. The PP's greatest weakness is people quitting it due to underperformance during bull markets and this "improved" version would have been even more vulnerable than usual to such tracking error.

I would suggest replacing half of the SV with either a NASDAQ-100 fund (or even a 2X NASDAQ-100 fund) or with a high beta ETF. Either one of those would have done far better during the mid to late 90s bull market (and in 2003-07, and 2009-10) than a PP with just SV for its stock allocation.

melveyr wrote:
For what you just laid out, you might also want to add in a sprinkle of some silver. It has a little more kick than gold. I don't think GLD has a volatility on par with EDV or VBR (or just bump up the GLD weighting).

I double this idea. Silver is like gold on steroids, despite being a somewhat more industrial metal. But I think silver is the only physical asset that is both close to gold and more volatile.

I have been thinking that instead of an enemy that volatility in each sector of the PP (except for cash) might be a friend.  Thinking that someone here had already toyed with this idea I came upon this thread.

For example, I ran the following:

1.  VOLATILE 4x25 PP at etfreplay.com from Jan 29, 2008 (farthest allowed back because of one of the ETFs used) EDV, VBR, IAU, and SHY - CAGR 7.8%, Max. Drawdown 13.55%  (SPY - 5.6% Max DD - 51.49%).

2.  Regular 4x25 PP at etfreplay.com from Jan 29, 2008, using VTI, IAU, TLT, and SHY - CAGR 3.1% Max Drawdown 3.85 %

Note how the volatile 4x25PP beat SPY quite handily with much less max DD. 

Has anyone done some serious back testing or have more thoughts on increasing the volatility of the PP components?

BP wrote: I have been thinking that instead of an enemy that volatility in each sector of the PP (except for cash) might be a friend. 

Isn't the whole basis of the PP that the 4 assets are volatile and uncorrelated?

Yes.  This is increasing the volatility still more.

BP wrote: Has anyone done some serious back testing or have more thoughts on increasing the volatility of the PP components?

Bad idea.  You can expect the 4x25 vanilla PP to have had a 22%-25% or so MaxDD.  It's very unbalanced in terms of gains to risk.  You should read the "Engineering Targeted Returns" thread.

Love the idea.  You'll recall that Harry Browne originally advocated trying to pick a basket of volatile stocks, but later fell back on index funds.

You'll note that not only do you get the performance boost from the extra volatility, but there will be a rebalancing benefit as well.  That's something that etfreplay doesn't let you model, at least without paying some dough.  I bet that would reduce volatility a bit....anyone?  (Sorry, just enough time to weigh in but not enough to put $$ where mouth is and do the backtesting myself.)

I will do the backtesting and we'll settle this issue once and for all.  It needs to be done before I implement factor tilting anyway.

Thanks to you both.  I will read the thread as suggested.  Looking forward to your research results MG.