The Correlated Risk Parity PP

[MachineGhost]

Not sure I ever posted this in the Achilles Heel thread, but a picture is worth a thousand words they say:

[align=center][img width=800]http://s22.postimg.org/ivna7iztt/riskparity.png[/img][/align]

[MachineGhost]

Is the left-most column the equal risk portfolio?  The equal volatility portfolio (both)? 

If so, what is that column at the far right with the cash?

Left is the equal risk contribution (ERC) portfolio using daily volatility for the risk metric.

And the right is the ERC portfolio scaled [down] to the 25x4 PP's level of daily volatility with the difference going into cash.

[MachineGhost]

This isn't that far off from the weightings I talked about in my risk parity / volatility thread, which multiple people condemned as too bond heavy (because rates have nowhere to go but up TM).

I think this makes a lot of sense vs 4x25.

Just keep in mind the bond volatility does not include a peak-to-trough bear market.

[MachineGhost]

Ah, I understand now.  That's very interesting.  And what period did you use to evaluate volatility? 

Please tell me it started in ~1975, and not back in the fictional gold years....

Its not fictional!  The London Gold Pool collapsed in 1968 and gold became a free market.  Legality technicalities of any one country are irrelevant when gold is a worldwide commodity.  Besides, didn't you see all the loopholes I listed in the other thread?  And only one person ever got prosecuted under FDR's order (and I think he won anyway).  It was just a stupid law with no real enforcement other than via self-fear.  All it did was prevent gold from being legally sold here in the open by dealers, not to citizens per se.  What would be the point of writing a book talking about the coming collapse of the dollar in 1970 if you really couldn't buy gold to protect yourself?

[MachineGhost]

Well, that's true if a typical investor could access the commodity.  I suppose a U.S. back-tester could convince himself that the 1971 end of the dollar peg could be a starting point.  Before that, a gold investment was equal to a cash investment.  And even after 1971, we saw a once-in-a-lifetime free market adjustment to the dollar/gold ratio.  To assume otherwise requires that you assume we could be re-pegged to the dollar (a return of the gold standard, however briefly).  So yes, I view the pre 1971 values as cash, and the most of the '70's as a once-in-a-lifetime adjustment in value.  We certainly don't want to measure returns or volatility in that special era and try to extrapolate it into the modern era; that would be very misleading, no?

I think you're under some kind of misapprehension that gold didn't flunctuate in U.S. dollar terms between 1968 and 1971.  So cash was hardly the same thing as gold.  Your cash bought less and less gold all the way up until 1980.

I could argue that when the 30yr bond collapses in a sovereign debt crisis under our floating exchange rate system in the near future, it will be a once in a lifetime, special era event.  Does that mean all prior history then becomes useless?

What we're really interested in with history is the extremes to get an idea of the realistic amount of risk that can be expected in the future.  Nothing more, nothing less.

And BTW anyone that owns 25% gold is not a typical investor, so its sort of a red herring to be worrying about it.  99% of investors back then would not have had a PP and if they were smart enough to have had one, they were smart enough to have figured out how to own the gold.

[MachineGhost]

Look, I like what you're doing, matching volatilities.  But to ignore a major singularity like losing the dollar/gold peg, is just not rational. 

I'm not ignoring it.  I want to capture it in the data.  It's a real event that happened.  I don't understand why you would want to sweep it under the rug?  Are you afraid gold will never be that volatile again and you'll underweight it in your portfolio?  Then why only a 10% allocation?  Look, this is what gold did before and after the pool collapse / depegging:

Code: Select all

1958	0.00%
1959	0.00%
1960	3.55%
1961	-2.74%
1962	-0.42%
1963	-0.28%
1964	0.28%
1965	0.42%
1966	-0.28%
1967	0.28%
1968	22.54%
1969	-5.75%
1970	-5.12%
1971	14.65%
1972	43.14%
1973	75.83%
1974	66.15%

It's not even that significant since the gains and losses between 1968.25-1971.75 were easily trumped by what came later including very recently with the 40% peak-to-trough drawdown.

[D1984]

Look, I like what you're doing, matching volatilities.  But to ignore a major singularity like losing the dollar/gold peg, is just not rational. 

I'm not ignoring it.  I want to capture it in the data.  It's a real event that happened.  I don't understand why you would want to sweep it under the rug?  Are you afraid gold will never be that volatile again and you'll underweight it in your portfolio?  Then why only a 10% allocation?  Look, this is what gold did before and after the pool collapse / depegging:

Code: Select all

1958	0.00%
1959	0.00%
1960	3.55%
1961	-2.74%
1962	-0.42%
1963	-0.28%
1964	0.28%
1965	0.42%
1966	-0.28%
1967	0.28%
1968	22.54%
1969	-5.75%
1970	-5.12%
1971	14.65%
1972	43.14%
1973	75.83%
1974	66.15%

It's not even that significant since the gains and losses between 1968.25-1971.75 were easily trumped by what came later including very recently with the 40% peak-to-trough drawdown.

MG,

Are your gold prices from LBMA? For 1970, they show gold closing at $37.38 on the PM closing fix of 12-31-1970; given that it closed 1969 at around $35.20 (after opening the year at just under $42....around $41.90 IIRC) shouldn't the table you posted show a gain of 6.19% for 1970 and a loss of 15.98% or 15.99% for 1969? The LBMA London price was the "free market" price during this time period, was it not?

[MachineGhost]

Are your gold prices from LBMA? For 1970, they show gold closing at $37.38 on the PM closing fix of 12-31-1970; given that it closed 1969 at around $35.20 (after opening the year at just under $42....around $41.90 IIRC) shouldn't the table you posted show a gain of 6.19% for 1970 and a loss of 15.98% or 15.99% for 1969? The LBMA London price was the "free market" price during this time period, was it not?

Thanks for pointing that out!  It seems that LBMA has finally updated their gold prices to go back to 1968 in line with other providers.  After bitching to them, I had been forced to use an average December (or yearly?) gold price for 1968-1972.  This only affected working with annual returns not daily data.

Code: Select all

Year	HBPP	HBPPCD	Desert	RPPP	RP25PP
1968	8.95%	9.31%	6.99%	8.04%	8.29%
1969	-5.96%	-5.31%	-5.25%	-5.94%	-4.88%
1970	8.07%	7.37%	11.84%	8.73%	8.18%
1971	11.37%	11.42%	10.93%	11.23%	10.79%
1972	19.44%	20.00%	12.39%	17.46%	17.19%
1973	13.83%	14.41%	3.94%	9.18%	10.41%
1974	12.23%	12.16%	1.23%	8.66%	9.30%
1975	6.15%	6.11%	12.46%	7.49%	7.76%
1976	10.82%	10.82%	15.24%	12.53%	11.31%
1977	5.31%	6.15%	0.57%	4.15%	4.37%
1978	11.83%	12.83%	5.79%	9.56%	10.25%
1979	37.92%	38.50%	20.56%	30.99%	32.01%
1980	13.30%	13.90%	11.66%	11.48%	13.10%
1981	-5.18%	-5.91%	-0.90%	-4.38%	-2.72%
1982	22.46%	22.05%	26.52%	25.43%	22.29%
1983	3.11%	3.97%	7.17%	3.41%	4.31%
1984	2.65%	2.99%	8.08%	4.62%	4.08%
1985	19.96%	19.99%	24.31%	22.68%	20.07%
1986	18.81%	18.55%	18.48%	20.52%	18.24%
1987	7.06%	7.72%	3.67%	4.80%	6.11%
1988	3.74%	4.71%	6.94%	5.08%	4.76%
1989	14.33%	14.23%	18.47%	16.11%	14.88%
1990	1.57%	1.47%	3.45%	1.84%	2.13%
1991	11.94%	11.43%	17.68%	13.82%	12.72%
1992	3.39%	3.62%	5.20%	4.22%	3.89%
1993	13.06%	13.38%	11.93%	14.14%	12.34%
1994	-2.39%	-1.08%	-3.70%	-3.52%	-2.16%
1995	19.61%	19.53%	24.20%	22.74%	19.85%
1996	5.09%	5.27%	6.91%	4.77%	5.58%
1997	8.23%	8.45%	13.76%	10.73%	9.55%
1998	12.98%	12.76%	15.86%	14.61%	13.27%
1999	3.26%	3.75%	3.38%	1.69%	3.42%
2000	2.94%	2.99%	5.19%	5.11%	3.48%
2001	0.21%	-0.58%	0.82%	0.35%	0.51%
2002	3.13%	3.23%	2.46%	2.47%	1.94%
2003	13.32%	13.61%	11.26%	12.64%	12.39%
2004	5.99%	6.65%	5.39%	6.53%	5.79%
2005	7.79%	8.45%	4.26%	7.49%	6.99%
2006	11.41%	11.65%	8.73%	10.28%	10.40%
2007	13.16%	12.62%	11.01%	12.41%	11.87%
2008	2.97%	2.30%	0.32%	6.60%	2.94%
2009	5.64%	5.98%	5.09%	1.54%	4.39%
2010	13.38%	13.62%	11.33%	12.86%	11.92%
2011	11.09%	11.15%	9.23%	13.95%	10.65%
2012	6.46%	6.55%	6.45%	6.22%	6.05%
2013	-2.43%	-2.27%	3.16%	-2.53%	-1.06%
2014	10.36%	10.61%	8.46%	13.13%	10.36%
Total	422.39%	431.12%	412.94%	425.94%	409.34%

Looking at 1969 (Fed hike), 1981 (Fed hike), 1994 (Fed hike) and 2013 (QEternity Will End), I think its still clear that all variations of the PP get whacked on rising real rates aka "Tight Money".  We like it loosy and goosy.  Is there a magical asset that could turn those years profitable?

[Kbg]

Total 422.39% 431.12% 412.94% 425.94% 409.34%

Wow I think Faber is on to something...all that and the standard PP lands smack dab in the middle of performance.

[MachineGhost]

Total 422.39% 431.12% 412.94% 425.94% 409.34%

Wow I think Faber is on to something...all that and the standard PP lands smack dab in the middle of performance.

I suspect the Marc "Dr. Doom Gloom and Boom" Faber portfolio would get whacked in "Tight Money" far worse than the PP.  50% real assets is a doom porn bet.  I could actually backtest it if you're interested.

[MachineGhost]

Yesterday, I did some volatility checking of bonds from 1925 to date and various subsets, especially the bond bear from 1946 to 1981 which had low rates at the outset comparable to today. 

Unfortunately, it seems that monthly data really masks the volatility one would experience on a daily basis as I was getting low volatility values not in any way comparable to 1977 to date (13.09%).  Isn't it normal to annualize by multiplying the square root of a value by the standard deviation of the returns representing the sampled temporal interval?  i.e. 12 for monthly data?

I don't know what the answer to the bond risk conundrum is, but I do know one thing.  The 1970's is no longer representative because it was high rates increasing higher (minimal duration).  We're now at 1940's low rates that are going to increase higher (maximum duration).  If anyone is not scared, they should be.

I don't see that the minor increase in volatility compensates for the over doubling increase in the duration.  I suspect HB may have been a victim of data mining.

On the other hand, someone could just properly calculate the volatility from 1946 to 1981 on T-Bonds and that should be the worst case scenario.  Then again, maybe not:

[quote=http://awealthofcommonsense.com/a-histo ... rrections/]An intriguing question, for those who maintain a static bond allocation, is how far out to go in duration.

In publicly-disclosed unleveraged versions of Ray Dalio’s All Weather Portfolio (such as the one described by Tony Robbins and reviewed here on Nov 18, 2014), a 40% weighting in long Treasuries (i.e. 20-year maturity) is used. Since even long Treasuries are still less volatile than equities, they require both long duration and a heavy weighting to serve as an effective counterbalance to stocks.

On the other hand, 7 to 10 year Treasuries historically have offered about the same return as 20 to 30 year Treasuries, with lower volatility. On a standalone basis, Treasury investors hugging the right-hand edge of the yield curve don’t get paid for the extra volatility they endure.

Ultimately it’s a question of context and correlations. If Treasuries continue to be mildly anti-correlated to stocks, then longer-term Treasuries can improve the Sharpe ratio of a stock-bond mix. But if we should return to the bad old days of 1979-1980, which produced the worst drawdown ever in Treasuries at the same time stocks went south, shorter maturities will be the best place to hide.

What to do, what to do?
[/quote]

[mathjak107]

Despite larry swedroe coming up with it there is no way i would ever use a larry portfolio in retirement unless i was only taking 2% withdrawals.

Success rates in firecalc have been pretty poor

[mathjak107]

Well no one can predict but i would say a anything with a 146 year history spanning 111 rolling 30 year periods and a 90% plus success rate would likely still play out well.

While i couldn't take the volatility anymore 100% equity would still likely do okay.

It has always done well because without the weight of cash and bondss the up markets go up so much more they cushion any downturns fairly well

[mathjak107]

We can think of risk anyway we like but if  we kept that vision of japan in our minds the last 25 years we would never be investors. That thinking would historically  left you a whole lot poorer.

There are volumes written about just why we are not japan.

[Pointedstick]

We can think of risk anyway we like but if  we kept that vision of japan in our minds the last 25 years we would never be investors. That thinking would historically  left you a whole lot poorer.

I think investing in that stock market is what would have made you a whole lot poorer.

[mathjak107]

Yep , which is why a portfolio needs to be flexible . Japan was hot when i first starting investing but after the 87 crash lots of things changed in that big picture and it was time back then to move elsewhere.

They tightened their money after the crash and drove themselves in to a death spiral.

Don't confuse risk with volatility.there is a big difference. Japan iis a risk ,the
s&p 500 is volatility.

Dealing with index's and diversified funds is dealing with volatility and market cycles. If the investment is matched to the right time frame it is only a volatile investment.

[Pointedstick]

Yep , which is why a portfolio needs to be flexible . Japan was hot when i first starting investing but after the 87 crash lots of things changed in that big picture and it was time back then to move elsewhere.

They tightened their money after the crash and drove themselves in to a death spiral.

That's easy to say with the benefit of hindsight. But if you were a Japanese investor in 1986, how would you know that? You've said many times before that stock market crashes should just be waited out because you're almost always made whole in 5 years or so. This is a case where that's not true. What are you supposed to do?

[mathjak107]

I also said anything i say applys to here and not other countries.

You have corruption ,political risk ,overthrowing gov't risk ,currancy risk  and a whole lot of other risks elsewhere besides just market volatility

[Pointedstick]

I also said anything i say applys to here and not other countries.

You have corruption ,political risk ,overthrowing gov't risk ,currancy risk  and a whole lot of other risks elsewhere besides just market volatility

Do any of those apply to Japan?

[mathjak107]

all you need to do to maintain that rosie retirement scenario and still have that 4% safe withdrawal rate hold true is a mere 2% average real return over the first 15 years.  that is all it takes because that rosie scenario you speak of is actually based on the worst conditions we have ever had.

which historically has left you with more money 90% of the time than you even started with using a 50/50 mix and 67% of the time left you with more than 2x what you started with after a 30 year period of spending.

when it comes to risk people tend to believe their own bull sh*t.  they  try to take every scenario out there that can play out and use that as ammunition as to why they are afraid to committ .

most of the time it isn't even risk , it is just the natural swings in the markets which are just volatility.

it is only when folks do the wrong thing and behave badly as investors or are speculators  that they get burned and lose money.

[MachineGhost]

More fuel for the fire:

[quote=http://www.valuewalk.com/2015/01/balanc ... idgewater/]Investors know they should diversify their portfolios—the question is how. Conventional wisdom holds that a properly diversified portfolio is 60% stocks and 40% bonds. This allocation assumes that steady bond returns will help offset the volatility of stocks. Alex Shahidi, taking his cue from the Bridgewater All Weather Fund, argues against this conventional wisdom. In Balanced Asset Allocation: How to Profit in Any Economic Climate (Wiley, 2015) he suggests that a truly balanced portfolio will include fewer stocks and more bonds, both treasuries and TIPS, along with some commodities. His sample balanced portfolio has 20% equities, 20% commodities, 30% long-term treasuries, and 30% long-term TIPS.

Overweighting bonds is not an intuitive asset allocation strategy, so Shahidi takes pains to explain its rationale. He sets the stage with a description of Ray Dalio's economic machine, more vividly shown in Bridgewater's 30-minute animated video. He then explains why a 60/40 portfolio is not well balanced. First, “the impact of an asset class on the total portfolio is only dependent on two factors: (1) how volatile the asset class is, and (2) how much of the total portfolio is weighted toward it. … The more volatile asset class should get a lesser weight to make up for the fact that it is more volatile. The less volatile segment should receive a higher allocation so that its impact on the portfolio matches that of the higher-volatility asset class.” Second, “the traditional 60/40 allocation is 99 percent correlated to the stock market!” (p. 24)

Instead of viewing an asset class as something that offers returns, the Bridgewater model looks at it as “something that offers different exposures to various economic climates.” (p. 26) Some sources of volatility, such as the future cash rate and risk appetite, are not diversifiable. But the most hazardous risk to investors--shifts in the economic environment--can in large measure be neutralized through proper asset allocation.

A simple two-factor model (growth and inflation) shows the economic bias of each major asset class. Equities want growth but not inflation, treasuries want neither, commodities want both, and TIPS want inflation but not growth.[/quote]

[quote=http://www.advisorperspectives.com/news ... sified.pdf]
[align=center][img width=800]http://s2.postimg.org/4y8tdaw2x/returns.png[/img][/align]
[/quote]

[MachineGhost]

I'm starting to come around to the idea that the best way to go may not be to put all your apples into one PP portfolio, but diversify among different "all weather" portfolios.  The composite should be stronger than any individual portfolio.  Here's the 15 and 30 year real returns for the Shahidi portfolio:

Code: Select all

1972	1986	3.02%
1973	1987	2.44%
1974	1988	3.25%
1975	1989	5.24%
1976	1990	4.83%
1977	1991	5.49%
1978	1992	6.09%
1979	1993	6.86%
1980	1994	6.64%
1981	1995	8.12%
1982	1996	9.31%
1983	1997	9.34%
1984	1998	9.18%
1985	1999	9.05%
1986	2000	8.51%
1987	2001	7.16%
1988	2002	7.68%
1989	2003	7.85%
1990	2004	7.31%
1991	2005	7.51%
1992	2006	6.63%
1993	2007	6.91%
1994	2008	5.81%
1995	2009	6.40%
1996	2010	5.78%
1997	2011	6.09%
1998	2012	5.63%
1999	2013	4.93%
2000	2014	5.20%

Code: Select all

1972	2001	5.09%
1973	2002	5.06%
1974	2003	5.55%
1975	2004	6.28%
1976	2005	6.17%
1977	2006	6.06%
1978	2007	6.50%
1979	2008	6.34%
1980	2009	6.52%
1981	2010	6.95%
1982	2011	7.70%
1983	2012	7.48%
1984	2013	7.05%
1985	2014	7.13%

$10K in 1987 turns into $109K.

[mathjak107]

i would agree . if i was not retiring now i would have split the money between the fidelity insight portfolio and the pp. the pp is great for preserving assets but is weak growing them over long accumulation time frames .

but i don't trust only a 25% equity position in the spending down phase  in lieu of the fact the pp spent pretty much all of it's 30 year time frames mostly through falling rate trends  .

i don't want to start spending down in 3 weeks with gold as a dead weight too.

there can be plenty of time down the road to add inflation fighting assets  but for now not something i can see adding to the already difficult problem of staying solvent early on in retirement.

[lordmetroid]

You can drop bonds as far as you please, even negative rates. Imagine if stocks and corporations were getting hammered much like the great depression. A lot of investors would be gung ho buying bonds with negative rates. Because it would be a lot safer to loose 1% per year instead of loosing it all.

[mathjak107]

while anything is possible you would be speculating on the flyer if you were counting on that to drive your portfolio.  realistically do you see that happening as opposed to rates continuing to rise on bonds as they have been for 6 months now ?