MachineGhost's Research Depot

[MachineGhost]

[img width=800]http://i.imgur.com/TcKORNS.png[/img]

The Risk Parity has higher risk than the orthodox PP, in terms of volatility, maximum drawdowns and portfolio duration.  To derisk the portfolio down to match the orthodox PP and put any difference into T-Bills:

Code: Select all

Risk Parity Rescaled to PP Volatility (6.33%)
Stocks		24.04%
LT Bonds	34.56%
Gold		21.12%
T-Bills		20.27%

Risk Parity Rescaled to PP Duration (18yrs)
Stocks		22.56%
LT Bonds	32.43%
Gold		19.82%
T-Bills		25.18%

Risk Parity Rescaled to 1945 Duration (12.24yrs)
Stocks		15.54%
LT Bonds	22.35%
Gold		13.66%
T-Bills		48.45%

Risk Parity Rescaled to 1981 Duration (6.81yrs)
Stocks		8.65%
LT Bonds	12.43%
Gold		7.60%
T-Bills		71.32%

Risk Parity Rescaled to PP MaxDD (-20.24%)
Stocks		25.46%
LT Bonds	36.60%
Gold		22.37%
T-Bills		15.56%

Risk Parity Rescaled to -15% MaxDD
Stocks		18.87%
LT Bonds	27.13%
Gold		16.58%
T-Bills		37.42%

Risk Parity Rescaled to -10% MaxDD
Stocks		12.58%
LT Bonds	18.09%
Gold		11.05%
T-Bills		58.28%

The above is just a derisking example.  The latest multiplicative factors are in this post:

[fi50@fi2023]

What is your takeaway from all of the variations you present here?

[MachineGhost]

What is your takeaway from all of the variations you present here?

Pick your poison.  Strategy diversification may be prudent.  I think most of them have flaws and the PP is superior, excluding my own of course!  Besides the Browne Permanent Risk Parity, I feel only the Clive is unique enough to stand on its own.  While it isn't risk parity, it is like "economic climate parity".  Probability of occurence is not a risk factor accounted for in traditional portfolio theory.  Still, it's not like the correlations are going to be anything but high among them since the PP covers the macro assets already.

[MachineGhost]

So if the 20yr bond duration decrease between 1945 (16.42yrs) and 1981 (6.25yrs) resulted in a 60% real maximum drawdown, I believe everyone could shift their PP's bond duration so that it is much closer to the peak in 1981.  Since 7 or 15 year bonds don't exist, lets settle for 10 year bonds.  Here's the nitty gritty (click to enlarge):

[img width=800]http://i.imgur.com/volRoZR.png[/img]

Currently, the HBPP with 20yr has a bond duration of about 15.46yrs; a 30yr bond duration is 19.925.  So 1945 with a virtual 30yr bond might have been 21.16yrs depending on the yield curve.

Even after switching to 10yrs, the portfolio durations are still high because of the absurd duration extreme that stocks are currently at compared to 1945 (1.97x) or 1981 (2.80x).  If you're not going to retire in at least 15yrs, better figure a way to lower your stock exposure.

OTOH, since the MaxDD didn't change much vs holding 10yr, 20yr or 30yr bonds, duration may be pointless to worry about beyond retirement timing.  It is just another risk parameter.

[MachineGhost]

I have updated the first post with a graphic so it is easier to see the portfolios.  And updated the other posts to reflect.

Work is still in progress.

As a note of interest, the Browne Permanent Risk Parity beats buy and hold of 100% into the S&P 500 with less than half the maximum drawdown.  Fancy!

[MachineGhost]

As an alternative to switching Treasury maturies / bond funds or engaging in tactical trend following, one can simply manage the duration according to quant duration funds and put the difference into T-Bills as my latest update shows.  Example:

30year Treasury has a 19.625 duration.  Average of 4 and 4.5 quant durations is 4.25.  4.25 / 19.6925 * 43.35% = 9.36% weight.

Beyond this point there is tilting and multistrategy fortress pyramid portfolios.  I'm not ready to work on the former and I'm actively implementing the latter.  Before I go back into hibernation, does anyone have any further questions?

[Cortopassi]

MG,

Those Pacer funds are interesting, a managed way to implement the 200 day MA in/out strategy for what seems to be a reasonable .60% fee.  Looks like the strategy have saved a few % points of loss vs. the corresponding Wilshire index.

Are there specific ETFs that correspond to the Quant Treasury and Gold Tactical entries?  I was not able to find anything.

[MachineGhost]

Are there specific ETFs that correspond to the Quant Treasury and Gold Tactical entries?  I was not able to find anything.

Nope.

[MachineGhost]

I've added the Browne MarketCap, Browne Minimum Risk (formerly the Volatility Parity Jr) and attributed the Gone Fishin'.

Hopefully, this will be the end of any updates for a few months.

Lastly, the Fortress size equalization (20% exposure to each market cap Mega to Micro) was done with these free Schwab ETFs: RSP, SCHX, SCHM, SCHA and WMCR.

[MachineGhost]

Below is a summary of all my research into the PP so far.

Asset Risk Equalization: 30.17% Stocks, 43.35% T-Bonds, 26.49% Gold

Equity Tilting: 60% domestic, 25% international, 10% emerging, 5% frontier

Domestic Size Equalization: 39.11% SCHX, 32.40% SCHA, 14.06% RSP, 9.33% WMCR, 5.10% SCHM.

Foreign Size Equalization: TBD

Sequence of Returns Risk: 538 trading days

Investment Horizon: 20 year duration

Worst Maxiumum Drawdown: -23.97%

Expected Forward Real Return: 2.168%+ CAGR

Derisking Factors: .7973 to PP volatility, .8444 to PP MaxDD, .7861 to PP duration

[img width=800]http://i.imgur.com/n20jQkq.png[/img]
[img width=800]http://i.imgur.com/YFXYYUQ.png[/img]

[mukramesh]

I might not be reading your charts correctly, but couldn't expected out-performance of the 3x Parity PP and the Tilt PP relative to the HBPP be due to not having a cash/short term treasury component?

[MachineGhost]

I might not be reading your charts correctly, but couldn't expected out-performance of the 3x Parity PP and the Tilt PP relative to the HBPP be due to not having a cash/short term treasury component?

Certainly!  But it's only .05% of missing return per 25% weight, i.e. not a high burden to overcome.

[Cortopassi]

I don't understand the 10 yr expected real return.  You have numbers of 1.7, 1.4, 2.0, 2.6  The numbers above are in addition to whatever the stock PP's CAGR is?

Or are you saying those are somehow the real CAGR's for all the options?  If so, doing PtoT just for the past 5 years on the HBPP shows 5.64%.  What am I missing?

[MachineGhost]

I don't understand the 10 yr expected real return.  You have numbers of 1.7, 1.4, 2.0, 2.6  The numbers above are in addition to whatever the stock PP's CAGR is?

Or are you saying those are somehow the real CAGR's for all the options?  If so, doing PtoT just for the past 5 years on the HBPP shows 5.64%.  What am I missing?

The latter. They are forward real CAGR based on all of the factors you can see in the first set of charts. 

The roll return and collateral reflects the use of futures contracts for commodities.  To reflect pure gold spot, subtract .40% from the HBPP; subtract .16% from Desert and subtract .432% from the risk parities.  For Desert's IT Treasuries, subtract .06%.

Desert, what is your exact tilt breakdown?

[Cortopassi]

MG,

Forgive me, I should have taken a finance class in college. 

I will just flat out say I do not know what that means:  Forward real CAGR. 

My only comparison point is the HBPP, last 10 years from PtoT is a 7.22% CAGR.

You are saying the same allocation, 25% each, going forward will only produce a 1.7% CAGR?

I do not understand how you arrive at such a radically lower number than what history has shown?  I'm the last guy to place too much credence in backtesting, but I just really don't understand the 1.7% number.  You don't have to explain, just telling you that I am one guy who looks at this and kind of goes, ok, I don't understand that, so I am pretty much going to ignore it, even though I know you've put a lot of effort in.

Mike

[MachineGhost]

I will just flat out say I do not know what that means:  Forward real CAGR. 

It means the projected, expected, forward nominal returns less the projected, expected, forward rate of inflation, the net of provided on a CAGR basis.

My only comparison point is the HBPP, last 10 years from PtoT is a 7.22% CAGR.

You are saying the same allocation, 25% each, going forward will only produce a 1.7% CAGR?

No it is 1.3% expected real CAGR for HBPP as I explained in a previous post.  It is so low vs recent experience because unique in history, both stocks and bonds are simultaneously overvalued.  So both won't be contributing much towards future returns, although the situation is much more severe with stocks than bonds.

I do not understand how you arrive at such a radically lower number than what history has shown?  I'm the last guy to place too much credence in backtesting, but I just really don't understand the 1.7% number.  You don't have to explain, just telling you that I am one guy who looks at this and kind of goes, ok, I don't understand that, so I am pretty much going to ignore it, even though I know you've put a lot of effort in.

Let's try breaking it down individually:



That clearer?  Again, ignore the roll return and collateral on gold since that is only applicable to futures, not spot.  I'd expect 1-year T-Bills to have exactly a 0% expected real CAGR.

[Cortopassi]

MG,

Almost there....

The expected real return graphical blocks for each of the assets -- your own development or from some financial site that you use which puts out the projection?

The 10 year forward prediction uses how much of the recent past to make the prediction?

And the big question:  Do you believe it?  If so everyone is in deep doo doo.

[MachineGhost]

Some math problems here ... you have PP at 1.7, Desert at 1.4, yet the only difference is Desert has less of the negative projected return asset (gold), replaced with small positive return assets (Treasuries and S&P).  Maybe you're doing something with rebalancing "bonus."

Subtract the numbers I mentioned previously to completely net out the influence of the commodity futures.  The problem with your portfolio as far as 30% being into Large Cap is it has negative expected returns so the HBPP with 5% less weighting to that will win out.  But we know that's not your actual portfolio since you've tilted, so I expect the tilts to beat out to beat HBPP.  Again, what is your tilting?

It looks like the spot valuation for gold is assuming a mean reversion to its real average price of just the last 10 years to account for changes in the production cycle.  That seems reasonable to me.  After all, the cure for low prices is low prices.

[MachineGhost]

MG,

Almost there....

The expected real return graphical blocks for each of the assets -- your own development or from some financial site that you use which puts out the projection?

The 10 year forward prediction uses how much of the recent past to make the prediction?

And the big question:  Do you believe it?  If so everyone is in deep doo doo.

I guess you missed the other thread.  The returns are from here: http://www.researchaffiliates.com/asset ... rview.aspx

Click on the Library icon then you can read about the methodology used for each asset class.

Of course I believe it, but keep in mind the expected real CAGR is a mean not a predestination.

Either way, as Desert indicated, the expected real return is way too low to even sustain a safe or sustainable withdrawal rate.  If you're not in retirement and you can live with opportunity cost, I guess the orthodox PP is still fine but it sure is flirting with disaster not to make any sensible changes.

[MachineGhost]

Anyone that wants to integrate Wellesley (VWINX) into their PP, it is 88.19% Equity, 10.92% T-Bonds, .89% T-Bills.

[Dieter]

Thanks! Looking forward to stock tilts away from S&P 500 helping again....

[lazyboy]

Anyone that wants to integrate Wellesley (VWINX) into their PP, it is 88.19% Equity, 10.92% T-Bonds, .89% T-Bills.

MG, I haven't a clue about what this means. How does this bash with Wellesley make up of 35% Large Value stocks and 65% Investment grade bonds?

[Cortopassi]

I guess you missed the other thread.  The returns are from here: http://www.researchaffiliates.com/asset ... rview.aspx

Click on the Library icon then you can read about the methodology used for each asset class.

Of course I believe it, but keep in mind the expected real CAGR is a mean not a predestination.

Either way, as Desert indicated, the expected real return is way too low to even sustain a safe or sustainable withdrawal rate.  If you're not in retirement and you can live with opportunity cost, I guess the orthodox PP is still fine but it sure is flirting with disaster not to make any sensible changes.

You've somehow come upon this Research Affiliates group as the basis of your determination that going forward pretty mush all asset classes will have poor returns, except EM Equity and a few select others.

I skimmed their methodology, and my immediate thought was that I could quickly find at least a handful of other reputable firms whose forecasts for the next 10 years were completely different.  And I did.

What does this all mean?  Goes back to what a lot of us here say all the time, the past can't really predict the future (though many believe that), and you just flat out can't predict the future anyway.  You are here putting out well thought out and researched efforts into what you think the future will hold, and I respect that, but it has just a good of a chance of turning out right as of turning out completely wrong.

So I have to step back and wonder -- Is MG right?  Is Desert right?  Is HB right?  Is the Golden Butterfly the better choice?  And on and on.  Or, should I leave it all alone because no one knows and I'd rather be equally spread around?  I continue to lean toward the latter, with maybe a little bit of more equity and international exposure than the standard HBPP.

[MachineGhost]

MG, I haven't a clue about what this means. How does this bash with Wellesley make up of 35% Large Value stocks and 65% Investment grade bonds?

It means if you invest $25000 of your PP into Wellesley, then $22047.50 of it will be allocated to Prosperity, $2730 allocated to Treasuries and the remaining -- if want to bother -- allocated to T-Bills.

[MachineGhost]

You've somehow come upon this Research Affiliates group as the basis of your determination that going forward pretty mush all asset classes will have poor returns, except EM Equity and a few select others.

I skimmed their methodology, and my immediate thought was that I could quickly find at least a handful of other reputable firms whose forecasts for the next 10 years were completely different.  And I did.

What does this all mean?  Goes back to what a lot of us here say all the time, the past can't really predict the future (though many believe that), and you just flat out can't predict the future anyway.  You are here putting out well thought out and researched efforts into what you think the future will hold, and I respect that, but it has just a good of a chance of turning out right as of turning out completely wrong.

So I have to step back and wonder -- Is MG right?  Is Desert right?  Is HB right?  Is the Golden Butterfly the better choice?  And on and on.  Or, should I leave it all alone because no one knows and I'd rather be equally spread around?  I continue to lean toward the latter, with maybe a little bit of more equity and international exposure than the standard HBPP.

I would be careful about what firms you pay attention to.  If their model of expected returns has no correlation to actual subsequent returns historically, then they're just engaging in marketing fiction for whatevers self-serving reason.  Between Research Affiliates, Hussman and GMO, a composite of all three sources is pretty clear that future expected real returns are going to be very low.  It's not rocket science, just common sense.  After all, stocks are nothing more than a legal claim on a very long-term stream (certainly a lot longer than bond durations at present) of discounted cash flows.  The price you pay today for that stream of cash flows determines the return you will get in the future.

[img width=800]http://z822j1x8tde3wuovlgo7ue15.wpengin ... 2/6040.png[/img]

As for predictability, that's why RA gives a 95% confidence range.  If 95% of all possible scenarios result in a low real return mean, how can anyone act as if they're going to be outlier exceptional and achieve the high end (or the low end)?  That's just the confirmation bias at work and gambling.  You want to see that the mean is at least high enough to reach your investment goals.  If the returns continued lower in the short-term future, you would have to re-adjust the portfolio as necessary. 

In a way, people retiring in the present need their future returns now so if they continue to bid up overvalued securities to get it, they are doing so at the expense of everyone else who is retiring later.