How to Normalize the PP's Risk

[MachineGhost]

There are several simple ways to fix the outsized influence of gold's risk so the portfolio's risk is normalized (or do I mean equalized?).

Trailing Volatility

This is the traditional approach.  First, set a target trailing volatility level, either a number you're comfortable with such as from bonds or the PP's average trailing volatility.  Let's say you decide on 10%.  So given that the trailing volatility of the risky assets are:

Stocks 20%
Bonds 12%
Gold 25%

Then:

Stocks 20% / 10% = 2.0 = 25% / 2.0 = 12.5%
Bonds  12% / 10% = 1.2 = 25% / 1.7 = 20.83%
Gold  25% / 10% = 2.5 = 25% / 2.5 = 10%
Cash  100% - Total of Above

To keep it robust, use the entire history for each of the risk assets so you're not curve fitting to a short-term economic subcycle that will prove to be sub-optimal weight going forward.

Maximum Drawdown

Similar to above, but rather than use trailing volatility, you use the maximum drawdown of each risky asset and decide on a portfolio maximum drawdown target aka portfolio heat.

Real Maximum Drawdown

Same as above, but you use the inflation-adjusted maximum drawdown of each risky asset.

Forward-Looking Volatility

Similar to above, but rather than use trailing volatility, you use the forward-looking implied volatility instead.  This is embeddded in long dated, at-the-money LEAP options for each of the risky assets.

Duration

Similar to the above, but rather than use volatility or maximum drawdown, you use the duration of each risky asset and decide on a portfolio weighted duration target.  This is most useful for those near retirement or with short-term goals to control potential capital losses before commencing withdrawals.  See: http://gyroscopicinvesting.com/forum/pe ... p/msg96254

Positive Years

Use an optimizer like Excel's Solver to come up with weights for each of the risky assets that targets for at least a 0% yearly PP return during each of the past 48 years.

Real Positive Years

Same as above, but use the inflation-adjusted yearly real return as a target instead.

Leverage

See: http://gyroscopicinvesting.com/forum/va ... #msg110582

[azmat9]

Awesome, thank you so much!

[azmat9]

How do you calculate the volatility of a given asset and do you normalize monthly or annually?

[MachineGhost]

How do you calculate the volatility of a given asset and do you normalize monthly or annually?

http://www.wikihow.com/Calculate-Histor ... Volatility

I would do it annually but using daily prices to make sure all moments are captured.

[frugal]

MG,

good morning!

But you still use 25% for all assets, right?


Best regards

[MachineGhost]

But you still use 25% for all assets, right?

Yup, if only because 1 / 4 = 25%.

[MachineGhost]

MG, in another thread [not sure which one] you came up with a division of

Stocks    33%
LTT        54%
Gold      13%

Which one of the risk normalizing methods was used to calculate that?

I do not remember.  Does it matter?

[MachineGhost]

Probably not, although I want to say that it was volatility, in which case the numbers don't jive with your first post above.

Also, in your opinion, is one of those methods preferable to the others?

Oh, I bet it was inverse volatility!  Can't see how the bond weight would be so high otherwise.

IMO, the forward-looking volatility is the most preferable.  It embeds the "wisdom of the crowd".  But personally I use maximum drawdown as a risk metric not volatility.

[MachineGhost]

What would be the breakdown using MaxDD?
And yet you still use 4x25 regardless?

I dont know as I don't set an explicit MaxDD target.  I finds ways to reduce the MaxDD without modifying the portfolio weightings.  That may change now as I think it makes sense to risk level the three assets.

[Lowe]

The OP is about deciding how far to let an asset get out of balance, before re-balancing the portfolio, right?

So gold has the ordinary 10% bands, and stocks have 12.5% bands, and bonds have 20.83% bands.  This smooths gold's price swings, relative to stock's smaller swings, or bond's even smaller swings.  Correct me if I am misunderstanding the strategy.

I have done stuff like this already.  Recently I let stocks ride higher than I was supposed to.  I did not have a system, though.  I eventually got scared and rebalanced.

[MachineGhost]

The OP is about deciding how far to let an asset get out of balance, before re-balancing the portfolio, right?

So gold has the ordinary 10% bands, and stocks have 12.5% bands, and bonds have 20.83% bands.  This smooths gold's price swings, relative to stock's smaller swings, or bond's even smaller swings.  Correct me if I am misunderstanding the strategy.

No, it's about modifying the strategic weights from 25% to whatever is necessary to level the risk contribution of the three assets to match each other.  In the standard HBPP, gold contributes more risk exposure than stocks and bonds; stocks more than bonds, etc..

I like the idea of using custom rebalancing bands for each asset, but the concept needs to prove itself worthy in backtesting.

[Lowe]

Haha, never mind then.  I have no idea what is going on.

If you decrease gold to 10%, based on a volatility estimate, then are the re-balance bands are then proportional to that ( 10 is to 25 what 4 is to 10 )?  You have probably talked about this elsewhere.

[MachineGhost]

Haha, never mind then.  I have no idea what is going on.

If you decrease gold to 10%, based on a volatility estimate, then are the re-balance bands are then proportional to that ( 10 is to 25 what 4 is to 10 )?  You have probably talked about this elsewhere.

No the rebalance bands would stay the same, and in theory all assets would then have an equal chance of hitting them unlike in the standard HBPP.