Must Read

[hrux]

Butler & Philbrick are putting out some of the best research on a consistent basis that I have seen anywhere.  They have reservations about it's future returns.

Part 1
http://gestaltu.blogspot.ca/2012/08/per ... art-1.html

Part 2
http://gestaltu.blogspot.ca/2012/08/per ... rt-ii.html

Japan PP
http://gestaltu.blogspot.com/2012/09/th ... anese.html

[Pointedstick]

Whenever I encounter a skeptic, my urge is to ask, "well, then what do you propose?" Usually the answer is a stock/bond mix that underperforms the PP (or only overperforms during cherry-picked time periods) , or a complicated market timing strategy that has you trading all the time.

Their "Tactical Asset Allocation" is the latter.

[hrux]

The authors make some valid suggestions for their tactical overlay

[melveyr]

If I had clients to impress and fees to justify then I would probably introduce formal risk parity and some attempt at constant volatility. It couldn't hurt much and it would keep me busy.

Luckily I don't have to impress anyone and so vanilla PP will be fine  Plus, as craigr notes in his signature, complexity kills returns. Commissions are no joke and are rarely included in simulations.

[craigr]

I have been investing for years. I have never once seen market timing work. It's all bunk.

If you want to market time, do it with money you can afford to lose!

[Pointedstick]

The authors make some valid suggestions for their tactical overlay

This is what I think of when I hear "tactical overlay":



Sounds like just another fancy term for "market timing" to me. As melveyr points out, the fees from all that trading are conveniently left out. That could be hundreds of dollars a year you're paying your broker just to use this extreme spec-ops tactical ninja asset matrix whatever.

[MediumTex]

Investment professionals whose livelihoods depend on sounding smart in front of their clients are rarely going to care much for the PP.  It's everything they aren't.

It would be like an elevator operator telling you all the reasons that self-service elevator controls are doomed to fail.  

BTW, melveyr is currently on a "very insightful PP post" streak.  I'm really enjoying reading them.  

[MachineGhost]

BTW, melveyr is currently on a "very insightful PP post" streak.  I'm really enjoying reading them.  

Ah, to be young and aggressive again...

[MediumTex]

The authors make some valid suggestions for their tactical overlay

As I recall, you did some serious PP tire kicking a while back, but decided to go with active management with a money manager who was a relative (forgive me if I'm not remembering all of that correctly).

How have the returns from the strategy you decided on measured up against the PP's returns over the same period?  As I recall, it was at least a year ago that you made your allocation decision.

[MediumTex]

In the end, it's not a PP.  It's just a timing model that spends most of its time in cash with enough assets that at least one of them is moving up for harvesting.   I'd be tempted to replace the non-working cash component of a vPP with this - IF transaction costs are manageable - but not more than that.

Even if it does "work", it sounds like it is a lot of work to maintain (maybe that's what you pay the manager to do for you).

It also sounds like it's not terribly tax efficient.

[stone]

With any strategy that entails stop loss abandonment of a falling asset you run the danger of being stopped out multiple times in quick succession -loosing a chunk with each stop out. Clive posted a lot of stop loss strategies on here. As far as I could see they were just exchanging an intermingled sequence of small gains and losses (such as the HBPP experiences) for a long sequence of near perfect incremental gains punctuated by occasional  calamatous sequences of multiple stop out "whip saw trap" losses. It wasn't reducing risk, it was just redistributing where in time the risk was suffered. It is actually worse to have all of you bad luck come at once but that is exactly what the stop loss strategy is designed to give you.

[MachineGhost]

Even if it does "work", it sounds like it is a lot of work to maintain (maybe that's what you pay the manager to do for you).

Checking whether two squiggly lines have crossed over each other once a month is hardly "a lot of work".  The problem is moving averages lag too much.  By the time you get a signal in either direction the market has a tendency to reverse, causing you to sell low and buy high.  It doesn't mean moving average crossovers don't improve risk adjusted returns, just that its not an improvement of orders of magnitude.

Transanction costs are also minimal these days.  Its not like pre-May Day when it cost $100 to trade even just 1 share of a stock.

Seriously, in this day and age of technological novelty, you guys sound like a bunch of fuddy duddies when it comes to figuring out how to logistically apply the most basic and elementary precept of technical analysis!  I suppose its fitting then that the market is like the Borg.

[hrux]

The authors make some valid suggestions for their tactical overlay

As I recall, you did some serious PP tire kicking a while back, but decided to go with active management with a money manager who was a relative (forgive me if I'm not remembering all of that correctly).

How have the returns from the strategy you decided on measured up against the PP's returns over the same period?  As I recall, it was at least a year ago that you made your allocation decision.

FYI, for the past 18 months my advisor managed portfolio has underperformed the PP by 6% however the volatility was also reduced.  However, I am running a portfolio for my mother with 1/3 dedicated to a PP, 1/3 to a 60-40 boglehead portfolio and the remaining third in a few actively managed funds by Doubleline, Merk, Pimco, Gateway and AQR.  So in essence I wanted her to have strategy diversification.  I have also dedicated 5% to peer to peer lending via Lending Club. 

[notsheigetz]

I'll bet there was at least one person in Holland who sold his tulips at the right time so you can't just say it doesn't work.

But I know you mean it doesn't work as a long term strategy unless you have a time machine. 

I have been investing for years. I have never once seen market timing work. It's all bunk.

If you want to market time, do it with money you can afford to lose!

[Gestalt]

As the author of the studies mentioned above, I'm glad our PP investigations have sparked such a debate here on the PP forum. That was the explicit intention.

I'm a little surprised by the general cynicism in this forum about the possibility of incremental improvements to the superb general framework described by Harry Browne. While the framework he proposed is conceptually sound, and has good ex post performance (from a U.S. perspective) since it was introduced, I think it naive in the extreme to assume that there is no way to improve upon it. Bridgewater has compounded its All Weather portfolio, which is a much better conceived rendition of the PP, since 1996 at 9.4% annualized with 12.4% volatility, with only one negative year (<-5% in 2008), so it may make sense to keep an open mind.

I also want to clear up a misunderstanding about the risk managed approaches we described in our articles. We did not use ex ante volatility to manage the allocation decision in retrospect. I know many 'risk parity' articles do use reverse engineered ex ante volatility to create allocations, but we are major skeptics of this approach.

Rather, we use simple observed 60 day realized historical volatility to dynamically size allocations, which accounts for the stochastic nature of volatility, and de facto serves to stabilize 'bet size' for each asset class within the portfolio. If the edge to the strategy is consistent through time, and Thorpe mathematically tied a statistical edge to an optimum constant bet size, then why not make the bet size constant through time, rather than allowing changes in volatility alter the risk profile of the portfolio over time in an uncontrolled way?

Also, the transaction cost objections in these posts are unfounded in our experience. We manage $120 million in much more active strategies and have yet to experience any major slippage. With a little creativity, commissions and slippage costs are negligible for liquid asset class exposure in contemporary markets.

Also, there is a very well founded research platform for 'trend' based investing, such as that described in our very simple PP tactical examples. Most recently, AQR published a paper involving an extremely simple trend following approach using 100 years of data for dozens of asset classes. The results below summarize their results, which include extremely onerous trading frictions:





The PP has a long history and a solid foundation, but it is difficult to argue with over 100 years of evidence for trend based approaches. We are broad skeptics ourselves, but there is a difference between a skeptic and a cynic. We would hope that this forum might be more dominated by the former rather than the latter.

Best of luck to all,
Gestalt

[Ad Orientem]

Medium Tex: It would be like an elevator operator telling you all the reasons that self-service elevator controls are doomed to fail.  

LOL I am so stealing that quote.

notsheigetz: But I know you mean it doesn't work as a long term strategy unless you have a time machine.

Advice for time travelers... set your wayback machine for 1920. Buy RCA (approximately $1.25 / share). Sell August - September 1929 (approximately $550 / share adjusted for splits). Buy long term non-callable tax exempt bonds issued by any state other than Arkansas. Buy 1930 Duesenberg J and beach front villa. Sip lemonade and enjoy the good life while waiting for prohibition to end.

[melveyr]

Gestalt,

Thank you for your reply! We are accustomed to giving the financial community a good ribbing, especially in ways that make us DIYers feel smart 

I think your risk parity tweaks are reasonable, and don't strike me as pure data-mining backtesting. I see the PP as one of the first risk-parity portfolios. Although it is not dynamically adjusted, Harry Browne got closer to risk parity through the use of long bonds and gold than most others. We implicitly assume that the 25% split is "neutral" because of the reasonably close volatilities of the components. I understand that this "neutrality" is disrupted if one of the components becomes more or less volatile than the others.

One concern I have with risk parity is this: if an asset class drops significantly then its volatility often becomes elevated. With risk parity you would be selling this asset after it had fallen. Does risk parity often cause a sell low and buy high behavior?

Thanks,
Ryan

[MediumTex]

Gestalt,

Thanks for joining us here.  I think that we always strive for courtesy here toward different points of view.  I suspect I would have been a bit more diplomatic had you been presenting this research personally.

As the author of the studies mentioned above, I'm glad our PP investigations have sparked such a debate here on the PP forum. That was the explicit intention.

I'm a little surprised by the general cynicism in this forum about the possibility of incremental improvements to the superb general framework described by Harry Browne.

For me, I think that skepticism is a better word than cynicism.  So many people in the investing world wander into trouble by failing to just stick with a simple and reliable strategy like the PP over time.  I hate the thought of a person starting up with the PP and before he ever even gets comfortable with its long term deployment, he's on to something else.  I think that this pattern is what you see in many people's overall investment career (just bouncing from one strategy to another without a lot of rhyme or reason).

While the framework he proposed is conceptually sound, and has good ex post performance (from a U.S. perspective) since it was introduced, I think it naive in the extreme to assume that there is no way to improve upon it. Bridgewater has compounded its All Weather portfolio, which is a much better conceived rendition of the PP, since 1996 at 9.4% annualized with 12.4% volatility, with only one negative year (<-5% in 2008), so it may make sense to keep an open mind.

I don't think that anyone would say that the PP can't be improved upon.  What I would be more inclined to say is that I'm not smart enough to consistently improve upon it, and this is the experience that I think many smart people have had over the years.  What seems to happen is that a more highly engineered strategy will beat the PP for a while, but between the management fees and previously brilliant market insights bumping into unexpected patches of reality, keeping up with the PP year in and year out is a bit like John Henry trying to keep up with the steam-powered drill.

I also want to clear up a misunderstanding about the risk managed approaches we described in our articles. We did not use ex ante volatility to manage the allocation decision in retrospect. I know many 'risk parity' articles do use reverse engineered ex ante volatility to create allocations, but we are major skeptics of this approach.

Rather, we use simple observed 60 day realized historical volatility to dynamically size allocations, which accounts for the stochastic nature of volatility, and de facto serves to stabilize 'bet size' for each asset class within the portfolio. If the edge to the strategy is consistent through time, and Thorpe mathematically tied a statistical edge to an optimum constant bet size, then why not make the bet size constant through time, rather than allowing changes in volatility alter the risk profile of the portfolio over time in an uncontrolled way?

That's a pretty complicated sounding approach.  Is this something that investors could consistently do (and understand) on their own, or would it be necessary to pay a professional to help them deploy, monitor and modify it as needed?

Also, the transaction cost objections in these posts are unfounded in our experience. We manage $120 million in much more active strategies and have yet to experience any major slippage. With a little creativity, commissions and slippage costs are negligible for liquid asset class exposure in contemporary markets.

What about management fees?  How much does that cost?

Also, there is a very well founded research platform for 'trend' based investing, such as that described in our very simple PP tactical examples. Most recently, AQR published a paper involving an extremely simple trend following approach using 100 years of data for dozens of asset classes.

When people talk about how easy it is to make money trend following, I'm reminded of Where Are the Customers' Yachts?.  Each person must make up his own mind regarding these things.

The PP has a long history and a solid foundation, but it is difficult to argue with over 100 years of evidence for trend based approaches. We are broad skeptics ourselves, but there is a difference between a skeptic and a cynic. We would hope that this forum might be more dominated by the former rather than the latter.

There are many things that can come between an investor and his money, and it is a rare investor who finds himself filled with regret at having been too skeptical.  I think that you will find that there are a lot more skeptics here than cynics.

The PP is NOT the only game in town when it comes to conservative strategies, and I don't have any reason to think that you may not have actually built a better mousetrap with your modified PP approach.  The fact that I believe the "un-tweaked" PP to be an outstanding strategy when ALL things are considered doesn't mean that it's the right decision for every investor seeking a simple way to obtain reliable inflation-adjusted returns.

Welcome to the forum!

[craigr]

As the author of the studies mentioned above, I'm glad our PP investigations have sparked such a debate here on the PP forum. That was the explicit intention.

Welcome to the forum.

I'm a little surprised by the general cynicism in this forum about the possibility of incremental improvements to the superb general framework described by Harry Browne.

Tex summed this up, but I'll say the same thing. I personally harbor very strong skepticism towards any idea that offers to beat the market, especially when it involves things like market timing. The idea of market timing certainly wasn't lost on Harry Browne either as he spent a large amount of time discussing it in his books and other writings.

That said, can the strategy be improved as markets change? Of course. I've talked about this myself in the past, and we even touch on it in the book with respect to no longer using Swiss Francs that were in the original Permanent Portfolio allocation. Reason? The Swiss Franc has experienced significant changes to how it is managed culminating in 2011's decision by their national bank to basically peg it to the Euro. It is no longer the same asset it was 30 years ago when first recommended. So things change and when they change the strategy should reflect that.

However I just don't see how market timing is an improvement, especially for an individual investor. There are a number of reasons for this:

1) The average person has a hard enough time deciding even on a basic allocation, let alone one that is going to shift around due to timing signals.
2) Every buy/sell opportunity in the timing strategy introduces a new decision point for the investor to second-guess themselves and be influenced by other sources of information that say do the opposite/do something else.
3) These strategies require frequent monitoring of the portfolio and the more an investor looks at their portfolio the more likely they are to be emotionally affected by things like losses and volatility. Investors should look at their portfolios less, not more.
4) These strategies all, by their very nature, have higher transaction costs, taxes and other fees that are never deducted from the theoretical returns.
5) The strategies are often heavily data mined to draw their conclusions and the assumptions used may not last going forward.

Even if market timing could be shown to work (and I've not seen that in any strategy going forward), the only people I think that have even a reasonable chance of pulling it off are institutional investors. An individual trying to do it is going to get socked with the points above, plus many more. Plus, the big problem with looking at past volatility behavior to make future guesses on what the volatility will be is a clear case of the Problem of Induction.

Also, the transaction cost objections in these posts are unfounded in our experience. We manage $120 million in much more active strategies and have yet to experience any major slippage. With a little creativity, commissions and slippage costs are negligible for liquid asset class exposure in contemporary markets.

What are your annual management fees? What is your after-tax return?

BTW. There is no problem discussing market timing here. We have a full sub-forum dedicated to it:

http://gyroscopicinvesting.com/forum/ht ... .php?f=0.0

But pointing out the overall failure of market timing to work consistently over time is not being cynical, it's just being realistic. If I thought market timing was a good idea, I'd be doing it myself. Trust me. But there are huge problems in terms of investor emotional management and reliability in the approaches.

I don't mind if people want to market time, but they should only be doing it with money they can afford to lose.

[MediumTex]

Craigr,

To be fair, we rarely even discuss taxes on the HBPP, except as a hand-wave that we trade less frequently and hence no one should worry about it.  But that's not entirely correct as anyone who's had a good sizable cap-gain on rebalance can attest to.  Especially as they get bumped up through the marginal rates while at the same time looking at those tantalizing unrealized cap-losses.

The best tax layout is going to be different for everyone, and forcing other strategies to include it in their posted returns while we blissfully ignore it is disingenuous.  Given the tax rate structure, there are many people who are better off with more frequent small gains than one large one.  I'd even argue that the target audience (retirees possibly?) would benefit the most.

It seems to me that only realizing gains in a diversified portfolio every 2-3 years (with the possibility of tax loss harvesting to offset gains) seems like about as good as it gets when it comes to tax efficiency (short of buying Berkshire Hathaway in 1972).

Point five sits uneasy with me as even the PP is data mined - even if you don't want to admit it.  It may be based on theoretical under-pinnnings, but at the end of the day there's some unaccounted for lag between the different components kicking in.  Settling on 25% equal weight components being robust enough is data mined.  Settling on 25+ year maturities is data mined.  To the extent that Gestalt's strategy data mines - I'd say it's less reliant on it than even the PP.  The ex-post target volatility is the only data-mined value, and even that you could go blindly and set it at whatever target you feel comfortable with and still be relatively happy - as long as you don't know whether the PP could have done better or not.

The difference with Gestalt's strategy (and every other modified PP) seems to be that ultimately it leans in one direction or another, while the PP maintains its radical neutrality, plus there is really no theoretical underpinnings to these modified PP approaches other than the belief that there is a more optimal approach, but any effort I have seen at optimization still seems to lean in one direction more than another.

Ultimately, too, I think that the PP only needs to be good enough, since we are always in the process of preparing for an unknown future, and thus simply staying in the ballpark is all most people should be trying to do.

For someone who has a knack for investing and successful speculation, and who is willing to spend an inordinate amount of time studying and monitoring his investments and the markets, the PP may actually be a terrible choice for him.

To me, the PP is striving to be a Honda, not a Ferrari.  As strong as a Ferrari can run when it's tuned properly, it's going to require a lot of maintenance if you expect to drive it every day.

I'm fine with being left in the dust by the PP tweak du jour if I get to sleep more soundly when the Big Bad Wolf shows up.  I admit that I fear losses more than I crave gains, which is something I had to learn the hard way several times before I figured it out.  I like having a strategy that I know I can understand and stick with no matter how crazy the markets get (and how rattled I may become).  That's an important feature for me.

I don't have anything bad to say about what Gestalt is proposing, other than it sounds complicated, expensive and riskier than the PP, and thus it would not be appealing to me personally.

[craigr]

People are free to try market timing if they want. The research in this area, however, is pretty conclusive that market timing doesn't work. Believe me, if I thought it would work, I'd be doing it. But I have looked at this issue forwards and backwards and have yet to see any compelling case when all factors are taking into account:

1) Fees.
2) Trading costs.
3) Taxes.
4) Human Emotions.
5) Reliability of the system to keep working as it did in the past.

I encourage anyone that thinks they can beat a passively run portfolio with market timing to please only do it with money you can afford to lose.

[hrux]

The authors make some valid suggestions for their tactical overlay

As I recall, you did some serious PP tire kicking a while back, but decided to go with active management with a money manager who was a relative (forgive me if I'm not remembering all of that correctly).

How have the returns from the strategy you decided on measured up against the PP's returns over the same period?  As I recall, it was at least a year ago that you made your allocation decision.

IMO, If forward returns are expected to be low on the 4 asset classes (assuming one agrees with that), that increases the odd of beating them. And that’s why I personally have not fully embraced the PP and continue to believe it is not the best way to approach this market.  Investors must make long term decisions, never knowing if they are going to be right until well into the future. They will remain disciplined for a few years then panic and do the wrong thing at the wrong time. PP investors will go through the same someday and won’t sleep as well as they think they will. No strategy is perfect, but the most important thing is to pick sensible strategies you believe in and stay with it. And why not take diversification to a new level by diversifying your strategies. That way you get additional non-correlation and are protected in the event one of them fails to work as anticipated.

[Pointedstick]

IMO, If forward returns are expected to be low on the 4 asset classes (assuming one agrees with that), that increases the odd of beating them. And that’s why I personally have not fully embraced the PP and continue to believe it is not the best way to approach this market.

If stocks, bonds, gold, and cash are under-performing, what's going up? Real estate? Corporate bonds? International stocks? Platinum?

[Tortoise]

And why not take diversification to a new level by diversifying your strategies. That way you get additional non-correlation and are protected in the event one of them fails to work as anticipated.

Can you give me an example of a strategy that is uncorrelated with the PP? I'm assuming it would have to be a strategy that holds no stocks, no LTTs, no gold, and no cash?

"Strategy diversification" is a bit of a tricky concept. If you "diversify" equally among three stock-heavy portfolios and the PP, for example, what you effectively end up with is... one big stock-heavy portfolio. That's just one example.

[craigr]

And why not take diversification to a new level by diversifying your strategies.

Depends what you mean. If you mean diversifying strategies by picking active managers, you have compounded the complexity and risk, not reduced it.

For instance, instead of picking stocks yourself, you have inserted a manager to pick stocks for you. But there are thousands of stocks and perhaps tens of thousands of money managers just in the U.S. alone. Now you have added a layer of manager and it's impossible to say what they'd do in selecting those stocks. If picking the right assets is hard, how much harder must it be to pick the right manager to then go and pick the right stocks for you!

I strongly feel that investors are best served by eliminating layers between them and their assets, not increasing them.