Duration Hedged PP

[MachineGhost]

This is in beta testing.  The idea is to have a duration hedged PP so that the durations of the assets are not so out of whack with each other.  This is especially acute between equities and bonds.

[img width=800]http://i62.tinypic.com/352hmk4.png[/img]

The performance as of this posting is 9.51% CAR and -13.79% MaxDD.

[buddtholomew]

I thought it was just me  ...glad to see that I am in such good company.

Enjoy your perspective MG, but what are you attempting to convey in this visualization? I suspect it has something to do with bond duration.

[dragoncar]

MG, you post some awesome data, so thanks for that. Maybe it's just me, but I have no idea what that table represents, which is often the case when I look at your charts and tables.

A post like this would be infinitely more useful of there was some clarification as to what exactly the numbers represent, and how to implement the conclusion in a practical way.

You have to subscribe to his newsletter :-P

[MachineGhost]

MG, you post some awesome data, so thanks for that. Maybe it's just me, but I have no idea what that table represents, which is often the case when I look at your charts and tables.

A post like this would be infinitely more useful of there was some clarification as to what exactly the numbers represent, and how to implement the conclusion in a practical way.

The results show the compounded annual returns, as well as the monthly gains and monthly averages.  This is beta, so there is no practical application yet.

The idea is to normalize the modified duration of all the components of the PP to each other, so that you can adjust the duration up or down of the portfolio as a whole to exactly match your cash flow needs, such as retirement.  As it stands now, the duration is very out of whack.  Stocks are 50 years, bonds are 16, gold is infinite and cash is 0.  I cannot find an imputed duration from an imputed yield on gold in any white papers because such an idea seems to not have occured to anyone in academia or Wall Street yet.  I don't have mad math skillz to be able to come up with it myself.

[MachineGhost]

I'm pleased to announce that I have finally determined an effective portfolio duration of the 25x4 PP!

It is 42.07 years as of this posting date.

Each 5% increase in cash and a proportional decrease in the other three assets reduces the duration by 2.77 years.

For a duration parity PP using bonds as the target duration, the weights are:

13.39% Stocks
21.69% Bonds
5.24% Gold
59.68% Cash

More to come.

[cnh]

MG, I'm very intrigued by your posts in this thread...particularly your latest.  I don't have "mad math skillz" either, but I'd be very interested to learn how you arrived at your effective portfolio duration for the 25X4 PP. I've been messing around with duration-based portfolio allocation, including duration parity portfolios (though I hadn't really tried to deal with gold duration). I'd be interested if you have an approach to deriving a PP allocation based on a varying target duration (e.g., based on someone's situation, the target or desired portfolio "duration" is 10 years...if so, what's the appropriate...or optimal...PP allocation?).

[Mark Leavy]

MG, I'm very intrigued by your posts in this thread...particularly your latest.  I don't have "mad math skillz" either, but I'd be very interested to learn how you arrived at your effective portfolio duration for the 25X4 PP. I've been messing around with duration-based portfolio allocation, including duration parity portfolios (though I hadn't really tried to deal with gold duration). I'd be interested if you have an approach to deriving a PP allocation based on a varying target duration (e.g., based on someone's situation, the target or desired portfolio "duration" is 10 years...if so, what's the appropriate...or optimal...PP allocation?).

Along with cnh, I am also curious about your results.

Depending on your "mad math skillz", it can be convenient to move between volatility and duration and CAGR and maximum draw down.

Lately, I've been using 10 year maximum draw down as my proxy for all of the above - but I would love to hear what you have come up with, MG.

Thanks,
Mark

[buddtholomew]

Can someone explain to me in layman's terms the point of this thread? I have a good grasp of fixed income duration, but how is the idea extended to stocks and gold?

[EdwardjK]

MG,

I chuckled when I read this post, shortly after submitting my own new post, "What Would Browne Have Changed?"  Trying to "improve" the Permanent Portfolio strategy is fraught with risk, especially with the Crawling Road crowd.  Why change what works so well?

I share others comments about what you are trying to achieve with equal duration of the strategy's components.  But if you are interested in potential tweaks, how about adding a momentum variable to the strategy to assess its impact on trading frequency and returns? 

The variable would increase the percentage in that asset which is growing the fastest at a point in time.  For example, if equities were growing faster than the other assets, the percentage allocation to equities might be 30% instead of 25%, with the offset being a reduction in the allocation to the slowest growing or the losing asset.

I think this retains the core Permanent Portfolio strategy and provides a potential for enhanced returns for a  bit more risk.

What do you think?

[EdwardjK]

Pugchief,

I would not say it is the "buy high, sell low" approach.  Rather, it may be the buy higher and sell even higher approach. 

Look back into 2013 when equities increased 30%+.  Would you have rebalanced out of equities when they were up 10% early in the year?  Or would you have waited until the upward trend slowed down later in the year?  Or would you have waited until sometime in 2014 in order to capture all the gains?  I suspect we all have different rebalancing time frames.

Yeah, you could call this market timing, whatever.  My comments were in response to the equal duration exercise being discussed, and offered that momentum can be considered as well.

Thanks for your comments.

[MachineGhost]

I think this retains the core Permanent Portfolio strategy and provides a potential for enhanced returns for a  bit more risk.

What do you think?

It sounds like you're advocating market-cap weighting of the PP, rather than equal-weighting.  I don't know how i feel about that considering equal weight beats out market-cap over time.  But a simple way to implement it would be when an asset first reaches the 30% band (which is equivalent to annual rebalancing).

[MachineGhost]

Can someone explain to me in layman's terms the point of this thread? I have a good grasp of fixed income duration, but how is the idea extended to stocks and gold?

Think "Target Date" or "Retirement".

[cnh]

Can someone explain to me in layman's terms the point of this thread? I have a good grasp of fixed income duration, but how is the idea extended to stocks and gold?

Buddtholomew, here's a very general and brief overview of the idea of stock duration: http://www.traders.com/Documentation/Fe ... inton.html.  Hussman also discusses it here: http://www.hussmanfunds.com/wmc/wmc040223.htm.  If you want to go beyond this, there's more here http://www.bus.umich.edu/FacultyResearc ... ration.pdf, here http://papers.ssrn.com/sol3/papers.cfm? ... id=2085954, and here http://www.russell.com/au/assets/pdfs/i ... ration.pdf. It's evolving "science."

MG, would you share what you ultimately used for the duration of gold in your calculations?

[MachineGhost]

MG, would you share what you ultimately used for the duration of gold in your calculations?

It's nothing revolutionary, but with NEM pegging its dividend to the price of gold, I realized I could just take the average yield of the gold mining industry.

[cnh]

MG, would you share what you ultimately used for the duration of gold in your calculations?

It's nothing revolutionary, but with NEM pegging its dividend to the price of gold, I realized I could just take the average yield of the gold mining industry.

So, something like this? Twelve-month yield for Market Vectors Gold Miners ETF (GDX) is 1.03%, which results in a rough, proxy duration of 97 years for gold?

[MachineGhost]

So, something like this? Twelve-month yield for Market Vectors Gold Miners ETF (GDX) is 1.03%, which results in a rough, proxy duration of 97 years for gold?

Something like that but I used a pure gold mining ETF not ones that mine copper, etc..

[MachineGhost]

I'd be remiss if I didn't point out that Roche has determined that stocks have a duration of 25 years based on a heuristic method of drawdowns to breakeven.  This is quite a substantial drop from the simple 1 / yield method that Hussman advocates.  So this makes the HBPP have a duration of about 10.5 years.