The Golden Butterfly Portfolio

[MachineGhost]

If something has a 1% probability, but wipes out 90% of the value, that's a bigger deal than something that has a 20% probability and stands to wipe out 10% of the value.  In the end it would probably take some serious actuarial-style math to come up with it.  And we here could probably do well enough with our experience and intuition, using just the probabilities as givens, and adjusting from there.

Do Clive's numbers have some relation to the occurrences of inflation, deflation, prosperity, and tight money?

I think it was SOLELY the number of occurences on a yearly basis, i.e. what was the winning asset class.

If you want to nitpick magnitude and not just probability, why the hell are you even in the PP?  It's so out of whack on many levels of risk.    I don't hear anyone complaining about the 25% "weighting for dummies" except myself.  Are you a new convert, sir?

There is only non-significant marginal improvements to be had once you risk normalize the PP to bother with more complex asset allocation schemes.  You're better off putting your energy into tilting and downside risk management.  I've done the latter but I'm still too lazy to get enthused about the former since we're talking marginal improvements.  Maybe .5% to 1% a year at best.  Wake me up when the show is over.

[I Shrugged]

And, I like the Golden Butterfly a lot.

Without having gone through my own plan above, my first instinct as a taxable investor is to make it a Global Butterfly:

20% US large stocks
20% Intl large stocks
20% LTT
20% STT or general ST for more income
20% Gold

[MachineGhost]

20% US large stocks
20% Intl large stocks
20% LTT
20% STT or general ST for more income
20% Gold

This is just changing the risk exposure to 40% equity to explain the higher returns.  It is more risky than the PP as a result.  When I say tilting, I mean keeping it within the fixed strategic allocation to Prosperity which if you're doing it on a risk normalized basis to PP's risk, is still 25%.

[dutchtraffic]

If something has a 1% probability, but wipes out 90% of the value, that's a bigger deal than something that has a 20% probability and stands to wipe out 10% of the value.  In the end it would probably take some serious actuarial-style math to come up with it.  And we here could probably do well enough with our experience and intuition, using just the probabilities as givens, and adjusting from there.

Do Clive's numbers have some relation to the occurrences of inflation, deflation, prosperity, and tight money?

I think it was SOLELY the number of occurences on a yearly basis, i.e. what was the winning asset class.

If you want to nitpick magnitude and not just probability, why the hell are you even in the PP?  It's so out of whack on many levels of risk.    I don't hear anyone complaining about the 25% "weighting for dummies" except myself.  Are you a new convert, sir?

I, for one, have adjusted the 4x25 to what I consider to be volatility/parity/risk improved. 35/50/15 stocks/LTT/gold,  cash separate but plentiful, although I am thinking about reducing the treasury duration a bit and reducing the gold allocation slightly. Maybe after the next stock crash. 

Why reduce the only asset that's not in the extreme-bubble zone?

[MachineGhost]

I, for one, have adjusted the 4x25 to what I consider to be volatility/parity/risk improved. 35/50/15 stocks/LTT/gold,  cash separate but plentiful, although I am thinking about reducing the treasury duration a bit and reducing the gold allocation slightly. Maybe after the next stock crash. 

I don't know where you came up with that because even the Volatility Parity Sr. PP isn't that high in T-Bonds!  Even adjusted down to PP's risk level, its still only 35%.  I'm curious what would happen to the other assets if I put on a lower constraint based on the "intuitive art" of not having enough historical data to simulate a full bond bear market.  A lot could happen in a year's time before anything new in the data was baked into the cake for your annual rebalancing.  Or maybe I could use 20-year T-Bonds and go back further in time...  anyone know what the duration of 20-years is???

I wish there was a quantitative duration-managed bond fund without all the unnecessary dreck.  I think the most practical solution at this point is PLW which is a 1-30 year Treasury ladder ETF.  However, I don't think the maturity exposure is equalized, but the duration is still only about 10 years which is less risk than unmanaged T-Bonds at 17 year duration.  I'd prefer real downside risk management but this is the best that can be done at the moment.

EDIT: The duration of 20-year T-Bonds is about 15.5 years.  But that brings up once again the drastically lower durations of 30-yr T-Bonds in the 70's and 80's...  perhaps a compromise is to target the duration in 1987 when the PP was invented.  I hate "art".

[MachineGhost]

Read the OP of this thread I started back in June. There is a link to the original source there.

http://gyroscopicinvesting.com/forum/pe ... olatility/

Oh, I remember that.  That's how I found Hedgewise but I must have got distracted by that and forgot to followup your original question!

Anyway, in their article, they are using naive risk parity (1/volatility) which doesn't take interasset correlations into account as the Volatility Parity PP does.  This is hugely problematic when in risk off mode and all assets increase in correlations.  And rebalancing the forward test on a monthly basis which is incredibly unpractical when transaction and taxes are factors.

Obviously, they're likely keeping true correlated risk parity for their advisory service.  At least, I hope so, because that has been my working impression to date and why I'm waiting for them to get out of beta and lower their minimum.  But now I have some doubts.  If they're crazy enough to offer only naive risk parity, then I ought to start my own advisory business... 

[MachineGhost]

Without posting charts or tables, can you simply give us your own preferred allocation to stocks/LTT/gold/cash based on optimization of risk parity/volatility/interasset/whatever? If you think intermediate bonds is better than a barbell, what duration is ideal?

It is still the same as before: 25% stock, 35% T-Bonds (30yr), 20% gold, 20% T-Bills.

Changing the duration would change the interasset correlations.  Using 20-year T-Bonds from 1968 to date as opposed to 1968 to 1977 results in a 5% increase to T-Bonds and a -5% decrease in cash.

I get a trailing volatility of 10.78% for point-in-time T-Bonds back to 1950.  Almost 3% difference from Hedgeable's analysis is a huge gap.  30yrs are simply more volatile.  Let's make an educated guess that to account for the duration difference between 20yr and 30yr if the latter had existed between 1968 and 1977, we lop off 5% from the T-Bond allocation.  That is the difference between my results and Hedgeable's for almost 3% volatility and 2 years in duration.

Or if you would prefer, going back to 1950 using official gold prices (it did flunctuate very minorly since it was a fixed exchange rate no different than China defending the yuan's "allowed" range against the USD): 24% stocks, 35% T-Bonds, 21% gold, 20% T-Bills.  But Treasuries were not fungible money before 1971, so I expect their behavior to be more volatile under floating exchange rates.

[MachineGhost]

Thanks for the clarification. That's not far off from my posted allocation. If you leave cash out of it as I did, your allocation is 31% [25/80] stocks, 44% [35/80] LTT, and 25% [20/80] gold. My bond duration is shorter than the equivalent 30 year, so I have added more bonds at the expense of gold, although I keep a small part of the stocks in international as a dollar hedge since I hold less gold.

The point being is that small increases in T-Bond allocation have a disproportionate impact on the portfolio risk.  So long as you decrease the duration at the same time I think it could be a net wash.  We really should decide on an official orthodox duration target and make sure all heterodox PP portfolios reflect it.