Performance of Individual Components During Market Crashes

[MachineGhost]

I did this fun exercise mainly to see how bonds respond during overvalued market crashes.

1973: Stocks -18.19%, Bonds -5.66%, Gold 72.35%, Cash 2.46%
1974: Stocks -29.81%, Bonds -0.05%, Gold 60.00%, Cash 4.61%
(Nifty Fifty Implosion: http://www.ozy.com/flashback/nifty-fift ... 37.article)
(Gold was not legal to own again until January 1, 1975 and only thanks to Ron Paul's diligent legislative efforts, so you'd have crushing PP losses unless you had invested in other real assets like, say, Chinese ceramics.  Not kidding.)

1981: Stocks -9.05%, Bonds -3.79%, Gold -32.60%, Cash 9.82%
(Tight money to break inflation expectations; gold imploded previous year.)

1987: Stocks 4.15%, Bonds -10.67%, Gold 19.98%, Cash 2.88%
(US Dollar Implosion: https://en.wikipedia.org/wiki/Black_Monday_%281987%29)
(The year Harry Browne came out with the current PP allocation.)

1994: Stocks 1.31%, Bonds -12.87%, Gold -2.25%, Cash -0.16%
(Bond Implosion: http://money.cnn.com/magazines/fortune/ ... /17/79850/)

2000: Stocks -8.26%, Bonds 21.66%, Gold -5.53%, Cash 4.58%
2001: Stocks -9.57%, Bonds 1.35%, Gold 1.91%, Cash 5.05%
2002: Stocks -22.70%, Bonds 13.28%, Gold 24.65%, Cash 2.12%
(Tech Implosion.  Without 09/11 fear, it seems likely gold would not have responded so favorably.  That is also the year I first bought gold, so it had to have been a fear trade.)

2008: Stocks -36.19%, Bonds 38.97%, Gold 2.60%, Cash 3.89%
(Subprime Implosion.)

2014-:
(TBD Implosion.)

So it seems to me that the market could vaciilate between either bonds or gold in providing crash protection.  I don't think we can universally say that bonds always hedge equity, as I previously thought.  In some cases, it is the cash yield alone that barely just saves the overall PP at the end of the day.  Can the same be said of now with interest rates at nearly zero if both stocks and bonds crash at once?  My worst fear is a worser repeat of 1981 or 1994 where nothing responds; I see nothing to suggest we're not going to have such a "Tight Money" repeat and it'll probably be far worse since I don't believe we've had both stock and bonds overvalued at the same time before along with zero paying cash.  That puts a tremendous amount of pressure on gold to outperform against a massive deflationary headwind.  And pigs fly.

[annieB]

Excellent research.Thanks..

[barrett]

MG, thanks for posting that.

A couple of thoughts...

Both 1982 and 1995 were really strong years for the PP. Also, when money comes out of assets through selling, it has to go somewhere. Apart from Chinese ceramics (Damn! Why didn't I buy?), some of the obvious choices are real estate (already getting high as well), gold & cash (terrible yields for the moment). So maybe there won't be a crash in both stocks and bonds because the alternatives are not great.

[dragoncar]

Sounds like we are overusing the word "implosion"

[MachineGhost]

Both 1982 and 1995 were really strong years for the PP. Also, when money comes out of assets through selling, it has to go somewhere. Apart from Chinese ceramics (Damn! Why didn't I buy?), some of the obvious choices are real estate (already getting high as well), gold & cash (terrible yields for the moment). So maybe there won't be a crash in both stocks and bonds because the alternatives are not great.

One can always hope.  The alternative seems like a slow death by a thousand cuts, Japanese-style.

[MachineGhost]

I've noticed an interesting effect of using monthly closing data, even if you buy on the open of the next day.  It somehow seems to literally halve historical MaxDD's.  I cannot quite figure out why this is so, but it makes suspect all systems based on monthly data like "A Quantitative Approach to Tactical Allocation".

In the meantime, the best rotational/momentum system when optimized has approximately the same MaxDD as the vanilla PP but about twice the CAGR.  But if you don't use monthly data, then the MaxDD goes past the native PP and sometimes into truly stupendous levels.  This is really only obvious if you only test on long periods of history.  10 years that every so-called "blog authority" uses is a joke not even worth considering.  I once again suspect the issue is in the lag between asset switching which is not a problem with the vanilla PP as it is always invested.

Last year, I feel the PP dodged a bullet because stocks should have tanked along with bonds and gold on such a "tight money" condition I've been worrying/warning about as it did in the same circumstances in the past, such as in 1969, 1981 and 1994.  It is only due to QEternity yield chasing bubblemania that investors decided to ignore reality.  So the quesetion is, will the PP perform as it should when the party ends, i.e. negatively?  There seems to be an average of 12-13 years between deflationary "tight money" style events.

Anyway here's another one for the list:

1969: Stocks: -11.42%, Bonds: -11.29%, Gold: -15.79%, Cash: 0.715%
I've no idea what happened in 1969.

So after all this, I've ironically come around to thinking the vanilla PP just cannot be beat.  It may not be very efficient.  It is certainly far from optimal.  But it is sure as hell robust.  The offsetting of all the assets work very, very well...  except when it doesn't.  But those don't seem to happen often enough to worry about?

[buddtholomew]

Congratulations and welcome.

[MachineGhost]

Congratulations and welcome.

Thank you.  I figure I've got better things to do than deal with the constant stress of a tactical PP, such as focusing on the Variable Portfolio (which is my "career" now).  Gsundheit!

[julian]

Your 1981 numbers seem off

[iwealth]

I've noticed an interesting effect of using monthly closing data, even if you buy on the open of the next day.  It somehow seems to literally halve historical MaxDD's.  I cannot quite figure out why this is so, but it makes suspect all systems based on monthly data like "A Quantitative Approach to Tactical Allocation".

Indicative of a consistent market behavior at month's end? But either way, I much prefer to backtest using random start days and variable holding periods. Surely makes for a more robust system.

[MachineGhost]

Your 1981 numbers seem off

In what way?

[MachineGhost]

Indicative of a consistent market behavior at month's end? But either way, I much prefer to backtest using random start days and variable holding periods. Surely makes for a more robust system.

No, I believe by using a varying end date, you can eliminate the day before or day after, etc. which results in smaller losses either by random chance or by optimization.  The outperformance is a mirage, in other words.  The results could possibly be replicated if you knew ahead of time what the new signal would be and buy before the close on the last day of the month.  This all speculation, of course.

[iwealth]

MG, I'm trying to understand what you are saying..

If one backtests a timing or rotation system based on end of month closing numbers and performs the transactions the following trading day, you are seeing a vast reduction in historical drawdown?

So if one backtests that same system using varying start dates such as the 2nd of the month, 3rd of the month, etc., assuming results are consistent with trading on the first of the month, wouldn't that eliminate doubt about the robustness of the system?

[MachineGhost]

If one backtests a timing or rotation system based on end of month closing numbers and performs the transactions the following trading day, you are seeing a vast reduction in historical drawdown?

No, you're seeing a vast reduction in maximum drawdown by buying on the close of the last trading day of the month.  The overnight gap between that close and the open of the next day can kill the return and increase the maximum drawdown.  And I don't believe you can actually get the open print price either, so there's even more slippage to consider.

So if one backtests that same system using varying start dates such as the 2nd of the month, 3rd of the month, etc., assuming results are consistent with trading on the first of the month, wouldn't that eliminate doubt about the robustness of the system?

Maybe.  A different start date would just demonstrate the path dependency.  I'd increase the exit date proportionately.

[vnatale]

I did this fun exercise mainly to see how bonds respond during overvalued market crashes.

1973: Stocks -18.19%, Bonds -5.66%, Gold 72.35%, Cash 2.46%
1974: Stocks -29.81%, Bonds -0.05%, Gold 60.00%, Cash 4.61%
(Nifty Fifty Implosion: http://www.ozy.com/flashback/nifty-fift ... 37.article)
(Gold was not legal to own again until January 1, 1975 and only thanks to Ron Paul's diligent legislative efforts, so you'd have crushing PP losses unless you had invested in other real assets like, say, Chinese ceramics.  Not kidding.)

1981: Stocks -9.05%, Bonds -3.79%, Gold -32.60%, Cash 9.82%
(Tight money to break inflation expectations; gold imploded previous year.)

1987: Stocks 4.15%, Bonds -10.67%, Gold 19.98%, Cash 2.88%
(US Dollar Implosion: https://en.wikipedia.org/wiki/Black_Monday_%281987%29)
(The year Harry Browne came out with the current PP allocation.)

1994: Stocks 1.31%, Bonds -12.87%, Gold -2.25%, Cash -0.16%
(Bond Implosion: http://money.cnn.com/magazines/fortune/ ... /17/79850/)

2000: Stocks -8.26%, Bonds 21.66%, Gold -5.53%, Cash 4.58%
2001: Stocks -9.57%, Bonds 1.35%, Gold 1.91%, Cash 5.05%
2002: Stocks -22.70%, Bonds 13.28%, Gold 24.65%, Cash 2.12%
(Tech Implosion.  Without 09/11 fear, it seems likely gold would not have responded so favorably.  That is also the year I first bought gold, so it had to have been a fear trade.)

2008: Stocks -36.19%, Bonds 38.97%, Gold 2.60%, Cash 3.89%
(Subprime Implosion.)

2014-:
(TBD Implosion.)

So it seems to me that the market could vaciilate between either bonds or gold in providing crash protection.  I don't think we can universally say that bonds always hedge equity, as I previously thought.  In some cases, it is the cash yield alone that barely just saves the overall PP at the end of the day.  Can the same be said of now with interest rates at nearly zero if both stocks and bonds crash at once?  My worst fear is a worser repeat of 1981 or 1994 where nothing responds; I see nothing to suggest we're not going to have such a "Tight Money" repeat and it'll probably be far worse since I don't believe we've had both stock and bonds overvalued at the same time before along with zero paying cash.  That puts a tremendous amount of pressure on gold to outperform against a massive deflationary headwind.  And pigs fly.

I'd rate this a "classic" study that somehow needs to be more easily found.

It does not seem that since this was written we've had any market crashes of any type?

And, couldn't the case be made that both stocks and bonds are overvalued even more albeit with cash no longer paying near zero?

Vinny