Alternative PP Models

[EdwardjK]

Although the traditional Permanent Portfolio has a great track record, I have modeled a bunch of alternative approaches to see if the results can be improved.

I used VTI (Equities), TLT (Treasuries), GLD (Gold) and SHY (Cash) as asset classes, I downloaded Yahoo Finance adjusted closing price data from January, 1, 2006 to December 31, 2016 into my models. I started with January 1, 2006 because not all of the ETFs have a longer history.

To create a benchmark for the traditional Permanent Portfolio, I started my investment on January 1, 2006 equally allocated between the four asset classes. Rebalancing only occurred when one or more assets fell below 15% or above 35% of total portfolio value. Here are the results:

Year Return CAGR
2016 8.2% 7.27%
2015 -3.0%
2014 11.0%
2013 -2.0%
2012 6.1%
2011 12.7%
2010 16.0%
2009 8.3%
2008 3.7%
2007 13.9%
2006 10.7%

MaxDD -13.77%

Of the many models I created, one particular model stood out. I would characterize this as a "momentum model" which starts with calculating the 21 day moving average of three asset price along with the 63-day asset return. The three are VTI, TLT & GLD. SHY is considered the "safe haven" investment. On the measurement date (defined in a moment), if the current price exceeds its 21-day moving average and the 63-day return is greater than "zero", the asset is considered for investment. For each asset that passes this test, its 63-day return is ranked. The top two of the three are the selected investments. If only one of the three passes the test, then the balance of the "failing" asset is invested in SHY. If none of the three pass the tests, then the entire portfolio is invested in SHY.

The measurement dates are Feb 15, May 15, Aug 15 and Nov 15, or the date closest to the 15th. I prefer quarterly updates and believe off-quarter and mid-month updates avoid all the turbulence associated with money managers updating their portfolios.

If it happens that the entire portfolio is in SHY at the next measurement date, then the portfolio is equally split between the next two "winning" assets at the next measurement date.

Having said all that, here are the results:

Year Return CAGR
2016 6.0% 9.60%
2015 1.5%
2014 11.2%
2013 6.3%
2012 5.9%
2011 26.9%
2010 29.3%
2009 5.2%
2008 9.6%
2007 6.3%
2006 3.6%

MaxDD -12.53%

As you can see, the CAGR is higher than the traditional Permanent Portfolio. the MaxDD is lower, and there are no loss years. It's hard not to like this approach. FYI-no transaction costs were included.

In a third model, I modified my momentum approach to rebalance at every measurement date - equally between the two "winning" assets. The results further improved:

Year Return CAGR
2016 5.9% 10.29%
2015 1.5%
2014 11.2%
2013 6.4%
2012 6.3%
2011 26.4%
2010 21.6%
2009 4.2%
2008 14.8%
2007 6.4%
2006 13.7%

MaxDD -11.75%

For those with better modeling skills and data, I welcome feedback on how both approaches look over longer time periods.

If anyone is interested, PM me your e-mail address and I will share my Excel file.

[ochotona]

Go to Portfoliovisualizer.com, to the Timing Models section, and see if you can replicate your method there. It's a good user interface. If you register, you can save your models. It's all free.

[Pet Hog]

...I downloaded Yahoo Finance adjusted closing price data from January, 1, 2006 to December 31, 2016...

I've noticed that the Yahoo "adjusted close prices" have, incorrectly, been equal to the "close prices" for quite a while. Perhaps it's just my browser. You might want to check your spreadsheet to see if dividends have been included. Hopefully your yields will improve!

[EdwardjK]

Go to Portfoliovisualizer.com, to the Timing Models section, and see if you can replicate your method there. It's a good user interface. If you register, you can save your models. It's all free.

I tried but the available options do not match the approach in my spreadsheet. Thank you for the suggestion though.

[EdwardjK]

...I downloaded Yahoo Finance adjusted closing price data from January, 1, 2006 to December 31, 2016...

I've noticed that the Yahoo "adjusted close prices" have, incorrectly, been equal to the "close prices" for quite a while. Perhaps it's just my browser. You might want to check your spreadsheet to see if dividends have been included. Hopefully your yields will improve!

Yes, I am at the mercy of Yahoo's accuracy. Although there are subscription services that provide much more accurate data, I do not have access to them. Call me cheap.

[Kbg]

I'll give this a shot. Could you be a bit more explicit on when one of the assets "fails" and what happens when the full portfolio is in SHY. Really I need to be clear on what percent of the portfolio is invested when. Assuming above 21ma and 63d return is positive it looks to be 50% each in the top 2. If only 1 asset passes then 50% to it and 50% to SHY. Correct?

I don't get what happens when the port is all in SHY on a rebalance date. Would not the same above rules be checked again and then assets be allocated as per the rules?

[EdwardjK]

kbg,

The rebalance dates are Feb 15, May 15, Aug. 15 and Nov. 15, or the dates closet to the 15th of these months. On these dates, the 21-day moving average price is calculated for each asset. For all assets where the price on the rebalance dates exceed its 21-day average, the 63-day return is calculated. For all assets whose 63-day return is positive, the returns are ranked and the top two selected for investment.

There are 4 possible outcomes:

1. All 3 assets (VTI, TLT, GLD) pass both tests. The assets with the highest positive 63-day return are selected for investment.
2. Only 2 assets pass both tests. These two assets are selected for investment.
3. Only 1 asset passes both tests. That asset is selected for investment. The rest goes to SHY.
4. No assets pass both tests. All investment goes to SHY.

I'll answer your other questions by way of examples.

Assume today is the rebalance date. Yesterday the assets were invested in VTI & GLD. Today the tests say the new investments are still VTI and GLD. In this case, no action is taken.

Yesterday the assets were invested in VTI & GLD. Today the tests say the new investments are TLT and GLD. In this case, the balance in VTI is sold and invested in TLT. The existing balance in GLD is untouched.

Yesterday the assets were invested in VTI & GLD. Today the tests say the new investments are TLT and SHY. In this case, the balance in both VTI and GLD are sold and equally invested in TLT and SHY.

Yesterday the assets were invested in VTI & GLD. Today the tests say the new investment SHY. In this case, the balance in VTI and GLD are sold and invested in SHY.

Yesterday the assets was invested in SHY. Today the tests say the new investments are TLT and GLD. In this case, the balance in SHY is sold and equally invested in TLT and GLD.

This method resulted in the portfolio with the 9.6% CAGR and -12.53% MaxDD.

Afterward, I went back and added a rule to rebalance each asset to 50% on all rebalance dates (for possible outcomes 1, 2, and 3).

This method resulted in the portfolio with the 10.29% CAGR and -11.75%% MaxDD.

I believe this covers all your questions. Let me know if you need anything else.

[EdwardjK]

I thought I would take this opportunity to explain why I did what I did with these models.

I like the HBPP concept and use it for a sizable portion of my current portfolio. But I feel SHY (and cash in general) is a wasted investment with near-zero interest rates, unless used as a safe haven. That is why I decided to focus on VTI, TLT and GLD as the targeted assets. When I did include it as a targeted asset, I found that 50% of the assets were invested in SHY for a substantial portion of the time, thus reducing overall return. Is that risky? Yeah, maybe, but that is what I did.

I chose to use a 63-day return test because I see this used in other models/strategies and I guess there is some science to it. As for the 21-day moving return, I chose that because it represents about one month of trading activity and reasonably reflects what some would call "momentum".

I chose off-quarter, mid-month rebalance dates to avoid any distortion of asset prices caused by end-of-month and end-of-quarter portfolio rebalancing that goes on at those times.

I chose a January 1, 2006 start date because that's as far back as I can go to obtain data for all 4 assets. Some ETFs were not in existence prior to that point.

Finally, I chose quarterly rebalancing because I like certainty. Meaning I know in advance the exact dates that I need to update my data and take action. I really do not like the idea of continually updating my data to see if my portfolio has fallen outside some rebalance bands.

If anyone has better modeling skills and access to more historical data for comparable assets, then please take the time to test the model and share the results.

I hope this helps explain my thinking.

[blue_ruin17]

Some questions:

• If your modified PP under-preformed the HBPP for a period of weeks, months or even years, would you be able to stick with your system, or would you bail?
• What are the real-world costs of running this modified PP? How much will transactions cost you? How many extra taxable events will occur?
• Would committing to this system force you to only invest in GLD to the exclusion of physical gold? What are the implications of owning no physical gold for a portfolio which is designed to survive serious political upheavals?
• Cash may feel like a "wasted asset" now, but what if interest rates spiked for a prolonged period of time and cash was once again an asset that spun-off meaningful income for your portfolio? Your system appears to be biased against this possibility.

Just some thoughts. I'm saying your system is "wrong" or "bad", but I would say that it introduces to complexities which probably void the HBPP warranty.

"Complexity kills returns" is a principle that I always have to remind myself about. I've backtested dozens of mutations of the HBPP that have outperformed it, but I never committed to any of them because they all "beat" the HBPP with some sort of after-market risk, complexity or 'hack', and I was never able to put absolute faith into any of my modified PP's because of this. I always knew, deep down, that I wouldn't be able to endure significant or prolonged periods of tracking error, and that I would bail from my system if it under-preformed the HBPP for even just a few months.

I believe that if you can't commit to an investment strategy with zen-like discipline and peace-of-mind for at least a decade, knowing that you won't bail on your strategy no matter how much it over/under preforms other strategies, it probably isn't a system that you should use. You're just going to end up island hopping between 'hot' strategies every two or three years, usually entering too late and exiting too early, and at the end of the decade you'll have nothing to show for it but below average returns and above average expenses.

[EdwardjK]

blue_ruin17,

Thank you for your message and questions.

While I admire Harry Browne's strategy and invest a portion of my portfolio using an adapted-Permanent Portfolio, I do not worship at the altar of Harry Browne. In part, Browne and his apostles prepared for a post-apocalyptic world requiring physical gold to survive. I just do not believe the Walking Dead will one day be a reality show.

In his writings, Browne said the PP is suitable for the money you cannot afford to lose. He did not say that you should invest your entire portfolio using his 4x25 strategy. He did not say his strategy's return would beat the market every year. Nor did he say that his strategy would never change over time. What he did say is that his strategy did appear to earn a reasonable return with low volatility over time.

Your first question asks whether I would stick with my "system" if it under-performed the HBPP for a period of weeks, months or even years. If the HBPP was my benchmark, then "no", I would not stick with any strategy that failed to meet or exceed that benchmark over the long-term.

As to your second question, we all face the the real-world costs of investing. All investors face transaction fees, income taxes, and fund/ETF fees. Fortunately, we all have access to either 401(k), Roth 401(k), IRA and/or Roth IRA accounts to at least minimize income tax costs.

I prefer to use ETFs rather than the traditional equity, Treasuries, physical gold and cash components. I imagine it's a challenge and expensive to store and transport (when necessary) physical gold. ETFs are easy.

Your last is a good question. If and when interest rates increase, I will consider cash as something other than safe haven. I have fond memories of opening a CD in 1981 when short-term interest rates were 16%. Of course, my very first mortgage interest rate was set at 14.25% in 1981 and that sucked.

I hope this is responsive to your questions. Let me know if you have any other questions or thoughts. Thanks.

[dualstow]

In part, Browne and his apostles prepared for a post-apocalyptic world requiring physical gold to survive.

Just one nitpick: Browne created the p.p. to diversify out of gold, not out of stocks and into gold.
Also, in his radio archives, you can hear Harry cautioning against at least one apocalyptic title (I forget the name at the moment).

[blue_ruin17]

While I admire Harry Browne's strategy and invest a portion of my portfolio using an adapted-Permanent Portfolio, I do not worship at the altar of Harry Browne. In part, Browne and his apostles prepared for a post-apocalyptic world requiring physical gold to survive. I just do not believe the Walking Dead will one day be a reality show.

The notion that physical gold is held solely to hedge against a 'Mad Max' scenario is a misrepresentation of very practical, utilitarian role that gold plays in the HBPP (or any other portfolio that holds gold).

The world does not have to "end" for gold to save your portfolio's ass. Just ask an investor who ran the HBPP in Iceland in 2008, or an American investor in the 1970s, or an investor in Argentina, Cyprus or Greece, and so on. The world didn't end in any of these cases, but lots of portfolios sure did get wiped out. From a macroscopic, multi-decade perspective, paper-portfolio busting events happen all the time. They don't usually happen globally, all at once (though there is nothing stopping that from occurring, either, as history reminds us), but every year that you run a portfolio, you subject your portfolio to the risk of happening to be invested in a region where "the end of [your portfolio's] world" does occur. And often, gold is the asset that would have saved your portfolio.

Hedging against uncertainty with gold isn't 'crazy', or ideologically motivated, or unreasonable. It is entirely logical and prudent.

There also is something to be said about being invested in an asset which conceivably could survive a very real regional apocalypse. For example, a German investor who allocated 25% of his capital in physical gold that he held in a Swiss vault in 1928 would have had a portion of his wealth survive the equity annihilating Great Depression and the total annihilation of bonds (to mention nothing of real estate and infrastructure) concluded in 1945 with the assistance of Bomber Command. Every other German investor would have been ruined, totally and absolutely. But the prudent German investor who held capital in gold which survived the war was able to re-deploy his assets and invest in what became known as the "German economic miracle" while most of his investor peers were genuinely lucky to still own a good pair of shoes.

As a side note, I would also point out that the equivalent to gold ETFs for the aforementioned German investor would have been obliterated along with the rest of his paper capital. Only physically held gold would have sufficed.

Even if such an event never occurs in the region where you invest, it is certain that, over the course of several decades, significant periods of uncertainty and fear will occur, and holding gold will allow you to profit from these periods and re-balance into great deals on stocks and/or bonds. In the unlikely (but by no means unrealistic) event that a generational event occurs which permanently destroys paper-portfolios, your wealth will be preserved through holding physical gold and you'll be positioned to invest in the economic 'spring' which inevitably presents itself as a ripe opportunity for anyone who was lucky (or prudent, rather) enough to have capital survive the event.

In his writings, Browne said the PP is suitable for the money you cannot afford to lose. He did not say that you should invest your entire portfolio using his 4x25 strategy. He did not say his strategy's return would beat the market every year. Nor did he say that his strategy would never change over time. What he did say is that his strategy did appear to earn a reasonable return with low volatility over time.

I think it would be fair to say that, after a lifetime of considering the mechanics and underlying premise behind the PP, that Harry Browne had effectively settled on a program that he was satisfied with. Once Harry Browne consolidated his thinking about asset allocation into the 25/25/25/25 system, he never again made a major revision to it.

Obviously, you don't have to allocate all your capital the way Harry Browne recommended. That's why he integrated the Variable Portfolio concept with his PP system (with just a couple rules for guidance).

But going even further, you don't have to commit any capital to the PP, if you don't want to. The HBPP isn't a religion. It's just a portfolio. For some, the logic behind the PP is extremely attractive when evaluated from the perspective of their unique goals and individual risk tolerances, and it is natural for them to comfortably allocate substantial portions of their wealth towards the PP.

For others, the PP just doesn't yield enough alpha. The HBPP just doesn't work for them, considering their unique investor goals and risk tolerances. That isn't a criticism of either the HBPP or this particular investor. I don't criticize those who buy mini-vans just because I don't require the utility provided by a mini-van.

However, I would say there is great danger in making substantial modifications to the PP and then presuming that you have preserved the capital preserving function of the pure HBPP. Modified, souped-up, alpha generating versions of the PP may beat the original for a while, but you might not be as safe as you think you are the next time we hit a major economic or geo-political speed bump.

There is some room for tinkering with the HBPP (i.e. 30/20 re-balancing instead of 35/15, or whatever), but once you start venturing into asset rotation strategies, you aren't running an HBPP anymore. And maybe you're perfectly aware and okay with that! Just make sure that you don't put undo faith in the reliability of your portfolio compared to a portfolio that has nearly half a century real-world, real-capital invested track record of safety, stability and real-returns.

Your first question asks whether I would stick with my "system" if it under-performed the HBPP for a period of weeks, months or even years. If the HBPP was my benchmark, then "no", I would not stick with any strategy that failed to meet or exceed that benchmark over the long-term.

Maybe I've misinterpreted what you've written, but it sounds like that is a pretty big red flag. If know you know today that you probably won't endure substantial tracking error compared to the HBPP, then when it happens for real you're very likely to bail out. Constantly changing asset allocation strategies is an absolute performance killer.

Avoid this by finding a portfolio you believe in, and therefore are likely to commit to for at least a decade, and you're likely to out-preform most individual investors for the simple reason that you stuck to your guns and ignore the fact that "everyone" is plowing into Uber Inc. (or whatever) because "it's the future". And then you agnosticly watch as Uber Inc (or whatever) dramatically out-preforms your chosen asset allocation...and then you agnosticly observe as your asset allocation dramatically out-preforms Uber Inc (or whatever) as it comes crashing down and investors flee like rats on a sinking ship.

Even the 60/40 portfolio was abandoned in droves in the 2008 crisis, and those investors were supposed to be in it "for the long run". They locked in huge losses because of their inability to stay committed to what is otherwise a very good portfolio. They didn't have enough faith in their strategy, and they paid for it, literally.

I prefer to use ETFs rather than the traditional equity, Treasuries, physical gold and cash components. I imagine it's a challenge and expensive to store and transport (when necessary) physical gold. ETFs are easy.

I use ETFs for 20% of my gold allocation, for ease of re-balancing and for taxable event avoidance purposes. But, as I described above, in the event of serious economic or political catastrophe, there simply is not substitute for physical gold. Physical price premium and costs of storage are simply the price to be paid for protection against systemic turbulence. I do avoid a substantial amount of expenses by never re-balancing from my physical allocation, though. That's the valuable role that gold ETFs held in tax-free accounts serve, for me.

[EdwardjK]

Blue-ruin17,

I believe in the Prime Directive.

Be well.

[Jack Jones]

Thanks blue_ruin. Very well put.

I'm reminded of my wealthy Latvian ancestors who lost everything (including their real estate holdings!) when the Russians invaded. If they only had some gold outside the country they would've had something to rebuild from. No zombies required in this story.

[EdwardjK]

Thanks blue_ruin. Very well put.

I'm reminded of my wealthy Latvian ancestors who lost everything (including their real estate holdings!) when the Russians invaded. If they only had some gold outside the country they would've had something to rebuild from. No zombies required in this story.

So why didn't your relatives have gold outside the country?

[Jack Jones]

So why didn't your relatives have gold outside the country?

They hadn't heard of geographic diversification I suppose.

[dualstow]

Ha. I wonder how many of us do. I haven't bothered. Maybe I should have something in Canada.

[Kbg]

kbg,

The rebalance dates are Feb 15, May 15, Aug. 15 and Nov. 15, or the dates closet to the 15th of these months. On these dates, the 21-day moving average price is calculated for each asset. For all assets where the price on the rebalance dates exceed its 21-day average, the 63-day return is calculated. For all assets whose 63-day return is positive, the returns are ranked and the top two selected for investment.

There are 4 possible outcomes:

1. All 3 assets (VTI, TLT, GLD) pass both tests. The assets with the highest positive 63-day return are selected for investment.
2. Only 2 assets pass both tests. These two assets are selected for investment.
3. Only 1 asset passes both tests. That asset is selected for investment. The rest goes to SHY.
4. No assets pass both tests. All investment goes to SHY.

I believe this covers all your questions. Let me know if you need anything else.

How's this sound:

Step 1 is a filter: non SHY ETFs must be above their 21 day MA

Step 2 is a sort: After step 1, pick the top 2 based on 63 day return if the return is also positive. 50% to each.

- If you have only 1 pick 50% to it 50% to SHY

- If no picks, 100% to SHY

Repeat quarterly

[EdwardjK]

kbg,

Yeah, that works too.

[Kbg]

My logic is a wee tad different just to make the coding easier.

12/30-05 - Now

PP Annual Rebalance - 6.87% CAGR/-14.02% MaxDD (intraday)
PP Qtr Rebalance - 6.80/-15.08

EdwardJK System
Qtr 1st of Month - 8.35/-14.92
Qtr 15 of Month - 7.36/-21.58

Personally I would be surprised if there was a full 3-5 CAGR point outperformance as reported (10-11% CAGR). However, it does look like this system does add some to the bottom line.

Here is the exact code I am using for anyone who can decipher it. The main difference in this code is that it is possible to have 50% SHY only position if nothing else qualifies and if a positive return does not beat SHY's 63 day return then SHY will be selected.

PositionScore =
IIf( C < MA( C, 21 ), 0, //C < Mov Avg = no position
IIf( ROC( C, 63 ) > 0, ROC( C, 63 ), 0 ) ); //ROC 63 < 0 = no position

SetPositionSize(50, spsPercentOfEquity);
SetOption("HoldminBars", 57);
SetOption("MaxOpenPositions", 2);


Buy = C > 0;
Sell = Day() >=15;
//Sell = Month() != Ref(Month(),-1);

[Kbg]

To eliminate SHY bumping a positive 63 day ROC symbol I removed SHY from the symbol list and get the following:

8.98/-15.12 1st of month
8.01/-25.61 15th of month

And I'm 99.99% sure that in the above where SHY is not an option and we would have added back in that CAGRs of 10 and 11 are not possible. Interest rates have been way too low to make up that difference.

So my take - a good tweak to the standard PP but with the chance of more volatility. Highly doubtful that EJKs numbers are accurate.

[Kbg]

Just for fun I eliminated the C > 21d MA requirement and trading at the first of the month I get 12.08/-15.85. If one trades on the 15th we still get that big dip and a 7.42 CAGR

[EdwardjK]

Highly doubtful that EJKs numbers are accurate.

Kbg,

I'd be happy to share my Excel file with you. Perhaps you can compare the data to your analysis to see where I may gone astray. PM me your e-mail address if interested.

Thanks.

[Kbg]

I don't really want to invest the personal time but likewise if you would like to go over the trades I'm showing, happy to provide you the results.

[EdwardjK]

For those interested, here is the asset selection for the portfolio achieving the 10.29% CAGR.

The most interesting observation is that this portfolio is in at least 50% SHY for 26 out of 44 rebalancing points.

If you are inclined, please validate the results and report back on any discrepancies.

Date Asset 1 Asset 2
11/15/16 VTI SHY
08/15/16 VTI GLD
05/16/16 GLD TLT
02/16/16 GLD TLT
11/16/15 SHY SHY
08/14/15 TLT SHY
05/15/15 VTI SHY
02/13/15 VTI SHY
11/14/14 VTI TLT
08/15/14 VTI TLT
05/15/14 TLT SHY
02/14/14 VTI GLD
11/15/13 VTI SHY
08/15/13 SHY SHY
05/15/13 VTI SHY
02/15/13 VTI SHY
11/15/12 TLT SHY
08/15/12 VTI GLD
05/15/12 TLT SHY
02/15/12 VTI TLT
11/15/11 TLT VTI
08/15/11 GLD TLT
05/16/11 TLT SHY
02/15/11 VTI GLD
11/15/10 VTI SHY
08/16/10 TLT GLD
05/14/10 GLD TLT
02/16/10 VTI GLD
11/16/09 GLD VTI
08/14/09 VTI SHY
05/15/09 VTI SHY
02/13/09 GLD SHY
11/14/08 TLT SHY
08/15/08 TLT SHY
05/15/08 VTI TLT
02/15/08 SHY SHY
11/15/07 TLT SHY
08/15/07 SHY SHY
05/15/07 VTI SHY
02/15/07 GLD VTI
11/15/06 VTI TLT
08/15/06 TLT SHY
05/15/06 GLD SHY
02/15/06 VTI SHY
12/30/05 GLD VTI