The Volatility Parity PP Sr. and The Volatility Parity PP Jr.

[MachineGhost]

I have split the "risk parity" concept into a new Volatility Parity PP Jr. and a Volatility Parity PP Sr..  Both are essentially the same as the former Risk Parity PP, but the Jr. includes T-Bills whereas the Sr. does not.  When I originally forced the Risk Parity PP to be at the Browne PP's volatility level so it would include T-Bills, that was not technically volatility parity since T-Bills do have volatility that was being ignored.  So without further ado, here are the weights:

Volatility Parity PP Jr.: 2.27% real CAGR (1968), -4.65% real MaxDD, 2.93% volatility
Stocks 7.67%
T-Bonds 10.80%
Gold 6.30%
T-Bills 75.23%

Volatility Parity PP Sr.: 5.52% real CAGR (1968), -17.60% real MaxDD, 9.46% volatility
Stocks 42.76%
T-Bonds 31.60%
Gold 25.64%

P.S.  These are end of year MaxDD values, not intrayear.

[Cortopassi]

MG,

I know you went through a lot of analysis with this, but I wanted to ask, when comparing these to the standard PP using Peak to Trough, this is the result with 25/25/25/25:

CAGR                 8.80%
Starting Capital 10,000
Ending Capital         556,230
Total Return         5462.30%
Max Drawdown 18.40% (1980-01-21 - 1980-03-27)
DD > 10% Count 7
Annualized Std. Dev 6.55%
Sharpe Ratio         0.52

Significantly better CAGR, about similar max drawdown, only ? is the volatility, and I need to ask why one would choose the Sr. Volatility Parity portfolio vs. standard PP?

Thanks.

[MachineGhost]

Significantly better CAGR, about similar max drawdown, only ? is the volatility, and I need to ask why one would choose the Sr. Volatility Parity portfolio vs. standard PP?

Peal2Trough is showing nominal returns; I'm showing real.  Too confusing?  I could go back to using nominal but I don't think that really shows the full negative impact of MaxDD's.  It is necessary so the full impact of withdrawals can be measured.

[MachineGhost]

Can you please clarify why LTT has a higher allocation than stocks in Jr, but a lower allocation in Sr? That doesn't seem to make sense.

You're thinking in terms of volatility alone.  The risk parity concept factors in both correlations and volatility.  So a relatively lower volatility asset is decreased in weight if its too correlated with other assets and vice versa.

[Mark Leavy]

These are great discussions, MG.  Thank you for continuing to pursue these ideas.

I did a back of the envelope calc and balanced the assets proportional to their historical drawdowns - and got nearly the same ratios as you did.  I thought that was interesting.

[MachineGhost]

These are great discussions, MG.  Thank you for continuing to pursue these ideas.

I did a back of the envelope calc and balanced the assets proportional to their historical drawdowns - and got nearly the same ratios as you did.  I thought that was interesting.

Volatility is a peak-to-trough or trough-to-peak measurement of price deviating from a moving average, so it will be similar to MaxDD so long as both the peak and the trough are present in the measurement period.

VaR (Value at Risk) is also similar to MaxDD, but it filters out all but the worst 95% or 99% of all losses in a given time frame, such as monthly.  MaxDD is by definition a 100% VaR until it isn't.

What's usually missing from MaxDD within a portfolio concept is the correlation with the other assets.  However, you can get an idea of a worst case scenario where all assets become 1:1 correlated and decrease all at the same time by summing up the individual portfolio weighted VaR's.

[MachineGhost]

When I mix his ideas (more stock, no cash) with mine, I come up with this mix:
50% equity, spilt equally among TSM, SCV and EM.

I think we should keep tilting out of this discussion.  This is only one dimensional.  The second and third dimensions are best left for other threads.

[Reub]

How many dimensions are there?

[MachineGhost]

How many dimensions are there?

Four:

One - Asset Allocation
Two - Asset Tilting
Three - Stock Picking
Four - Downside Risk Management

[koekebakker]

I've never heard of those 4 dimensions. Where can we read more about those dimensions? Tried to google it but no luck...

[Reub]

Google "string theory".

[EdwardjK]

Don't forget the 5th Dimension and the Age of Aquarius.

[MachineGhost]

How many dimensions are there?

Four:

One - Asset Allocation
Two - Asset Tilting
Three - Stock Picking
Four - Downside Risk Management

Five - Volatility Sizing
Six - Tax Loss Harvesting

Can someone think of a seventh dimension? 

Let's rearrange this so it makes more sense and guess at the expected additional returns over a buy and pray/fold portfolio:

1. Volatility Sizing (Huge, but varies)
2. Downside Risk Management (3.36%)
3. Stock Picking (.72%)
4. Asset Tilting (.68%)
5. Tax Loss Harvesting (.53%)
6. Optimal Asset Allocation (.34%)

[Jack Jones]

Let's rearrange this so it makes more sense and guess at the expected additional returns over a buy and pray/fold portfolio:

2. Downside Risk Management (3.36%)

I call baloney. Did you pull this from Hedgeable's whitepaper or something?

[MachineGhost]

I call baloney. Did you pull this from Hedgeable's whitepaper or something?

Yep!  Why do you think its baloney?

Here's something to consider.  Missing the 10 worst days in the stock market from 1993 to late 2010 would have generated a 4.88% CAGR outperformance.  That's enough to sextuple your starting balance instead of merely tripling it.

[Pointedstick]

I call baloney. Did you pull this from Hedgeable's whitepaper or something?

Yep!  Why do you think its baloney?

Here's something to consider.  Missing the 10 worst days in the stock market from 1993 to late 2010 would have generated a 4.88% CAGR outperformance.  That's enough to sextuple your starting balance instead of merely tripling it.

By the same token, I imagine that missing the best 10 days would demolish your CAGR.

[MachineGhost]

By the same token, I imagine that missing the best 10 days would demolish your CAGR.

That's a -50% gain difference on the starting balance (or +150%), so there's more "juice" in avoiding the 10 worst days than worrying about missing the 10 best.  The thing is, the 10 best days almost always follow the 10 worst days.  I haven't seen any analysis about how much the non-worst-day-best-days outliers contribute.  I suspect it is that and a combination of preserving a larger capital base for the sextupling.

[Pointedstick]

So… how do we avoid the worst days while being invested to benefit from the best days?

[Libertarian666]

So… how do we avoid the worst days while being invested to benefit from the best days?

You can't. If you could, they wouldn't be the worst days. 

[MachineGhost]

So… how do we avoid the worst days while being invested to benefit from the best days?

You can't. If you could, they wouldn't be the worst days. 

Indeed, I don't believe it is possible to do it consistently.  At least not without some really uber advanced Artificial Intelligence Sentience that's far off in the flung future...  should I give it a go?  Heh!

Look, people have enough problems with just the concept of wrapping their head around protecting the downside because of all the endless Boglehead propaganda, and you're asking about the Holy Grail?!! 

[MachineGhost]

[img width=800]http://i.imgur.com/K02Ap1A.png[/img]

[MachineGhost]

These are great discussions, MG.  Thank you for continuing to pursue these ideas.

I did a back of the envelope calc and balanced the assets proportional to their historical drawdowns - and got nearly the same ratios as you did.  I thought that was interesting.

[img width=800]http://i.imgur.com/VQdzjJ5.png[/img]

I'm truly done squeezing the PP.  There is no blood left!

[Mark Leavy]

I'm truly done squeezing the PP.  There is no blood left!

In addition to your PP blood squeezing, you've demonstrated some simple and sensible approaches to looking at any passive portfolio.  Much appreciated.

[MachineGhost]

I'm truly done squeezing the PP.  There is no blood left!

In addition to your PP blood squeezing, you've demonstrated some simple and sensible approaches to looking at any passive portfolio.  Much appreciated.

I aim to please!  But seriously, my final thought on this project is I want to clarify that my version of risk parity or volatility parity or even MaxDD parity is using the covariances between the assets.  From what I understand, the big players that use risk parity have traditionally used 1 / volatility normalizing all asset weights to the lowest volatility asset which doesn't account for the covariances.  Hence, why you wind up with a relatively large overallocation to T-Bonds that will kill the portfolio if yields rise, as they did to Dalio's Bridgewater All Weather fund in 2013.

So in the end, it looks like a 50%/50% mathjak portfolio except with longer duration bonds and some gold in place of equity as optimal.  But with no cash or other assets, the optimizer has to put the funds somewhere.  If you feel that more diversification would enhance risk-reward and provide less concentration, then by all means, test it out!  The PP is a rather risky and sloppy binary approach in having only one asset class per type: large cap for stocks, gold for real and the longest duration bonds for risk-free Treasuries.  I plan on testing that concept out next, i.e. Most Diversified Portfolio (MDP) which is the #1 portfolio that beat out the 1/v risk parity (#2) and the PP (#3) in the 300-year Monte Carlo backtest.

P.S.  I still have some doubts MaxDD works within this framework, but since I don't use returns at all, I can't find any flaw with it in theory.

P.P.S.  The essential difference between volatility and MaxDD parity looks to be 5% between T-Bills and T-Bonds.  I just don't have a high enough confidence that the covariances since 1968 are representative of the future risk in T-Bonds.  If there were a quantitative duration managed Treasury bond fund available, it would solve the issue nicely.  Only Hussman has come close so far but he pollutes the fund with gold stocks and other fixed income types, i.e. a self-standing strategy not an equity hedge (although don't ask me what you do when you got a 50-year duration in stocks with a 5-year duration in T-Bonds when current conditions dictate as they do at the moment).  Another half-baked solution may be the PLW ETF.

P.P.P.S.  What I'm saying about T-Bonds is the Black Swan Fat Tail Outlier is not in the historical data.  I could be wrong, but we're at a 5,000 year low in interest rates.  Do you want to bet on it not happening???  Use your intuition and gut feel and lower the T-Bonds either in weight or duration.

P.P.P.P.S.  The PP is certainly fine for the Set It And Forget It For Dummies types.  Just don't be surprised at the suboptimal growth and risk as budd was.  The all weather core is what makes it robust.  Over and out!