Moving Averages/Market Timing

[gap]

This topic may be anathema to many HB purists but I would like your thoughts on the subject anyway. I was surprised to get a null result when I did  a search with "market timing" and also with "moving average" while looking for discussions on the topic. With the small historical data  (1998 onwards) to which I have access, it appears taking a 200 day average and staying in ST/Cash positions when the value of PP is lower than the 200 day average offered better returns.less Draw Down and a higher Sharpe ratio than holding PP all the time.In fact it seems to work for PRPFX too

It is so  simple a strategy that I have this recurrent feeling that it may be a quirk of the paucity of data I have and is possible too good to be true.

Curious about your views

[rickb]

There was a discussion a while ago, see this thread.

[Bonafede]

Also may want to check out this one.

[clacy]

As with just about any strategy, the value of using MA's will vary greatly from time period to time period.  A MA filter such as the 10 month average like Mebane Faber uses, will keep you out of severe declines, so for periods like 2007-8 it would have out-performed.  However, you would miss much of the upside that took place in March-June of 09.  As mentioned in the threads that were linked, you have to be able to handle some whipsaws during volatile time periods as well.

Perhaps the best approach might be to use a timing method such as MA's or Relative Strength for half of your portfolio and use the HBPP by-the-book approach for the other half.

[gap]

Thanks for all the comments and especially pointers to previous threads which were very enlightening

[AdamA]

Speaking only from my own personal experience, I tend to be happiest with the PP (or any investment strategy) the less buying and selling I have to do.

It would not surprise me if some these strategies outperformed the PP slightly over the very long term (in tax protected accounts), but I think you'd have to be very disciplined to reap the benefits.  Once you add something like 200 day MA, it becomes very tempting to try to add other things or even just shoot from the gut.  

It's also psychologically tough to stick with MA strategies when they are underperforming and taking losses.  It's not fun to get whipsawed in and out of unprofitable trades for a year or two.

[craigr]

Market timing doesn't work.

If you used 200 day MA on bonds you maybe sold out in December 2010. But then when to get back in? May 2011? It crossed again so you buy. But wait it now went below 200MA again so you sell. But then a month later it crosses again. Ha! You won't be fooled again and you wait.

Big mistake.

They shot through the roof right after that.

There are two decisions involved in this question and therefore multiple points where you can question yourself (Should I sell now? Should I wait? Should I buy now? Should I wait?) If you waited too long you lost a lot of money. They are now up +29%.

Then there is the idea that you wouldn't have rebalanced into them in January because they are, after all, now under the 200MA line. So in other words the best time you could have bought them when they were the cheapest, was the exact same time your 200MA told you not do it!

(The above is assuming 2010-2011. The general condition still applies regardless of year.)



Then you get the transaction fees and taxes on top of it. No thanks.

Market timing always can look great in hindsight. But in practice I don't know of any investor who actually gets good results doing it.

A Permanent Portfolio with zero trades this year is up over 12% doing nothing at all.

[craigr]

I just created this graphic of how 200 day moving average market timing really works. This is the iShares Treasury Long Term ETF:

[moda0306]

Prettiest thing I've seen in a long time.

[craigr]

I just created this graphic of how 200 day moving average market timing really works

? ? ?

How weekly reviewed timing really worked for TLT over that period :

I would be suspicious if an indivudual trades that way. Too many emotions involved.

Even then, if an investor just bought at the beginning of the chart and held on they not only had the excellent price appreciation, but also were getting the bond interest payments the entire time which are often not included in timing charts.

 

[craigr]

So why is 20 weeks the right number? Why not 10 weeks? Or 30? Or 43?

Why is 200 days moving average the right one? Why not 150 days? Why not 638? Why not every leap year? 

Point being that we don't know what the right number is for the moving average until after the fact.

[TripleB]

So why is 20 weeks the right number? Why not 10 weeks? Or 30? Or 43?

Why is 200 days moving average the right one? Why not 150 days? Why not 638? Why not every leap year? 

Point being that we don't know what the right number is for the moving average until after the fact.

To further this point, suppose 200 days is the magical number. If true, then some smart hedge fund investors would use a 199 day moving average, and beat everyone else by 1 day and make a killing.

Then even smarter hedge fund people would start using a 198 day moving average and beat those people.

It's the same thing for why there's no free money to be had in dividend stocks. People say that the stock doesn't drop by an equal amount after the ex-dividend day. Bullshit. The market does things, as the market does, but the stock will drop by the dividend amount, in addition to market forces of that day. If it didnt, then people would buy the stock one day before the ex-dividend day and make free money. And smarter people would recognize that and buy 2 days before the ex-dividend date, and so forth.

Also, don't you find it coincidental that 200 days or 20 weeks is considered special, and we have 10 fingers and use a base 10 number system? If we had 9 fingers would 198 days be the magical number, because in base 9 it would be a round number? Same principle applies to 1000 point intervals on the DJIA as being special. It's only because we have 10 fingers. We might as well call 12,139.45 the special threshold, but we call 12,000 special because it looks round.

[amp]

Completely understand your questioning of the 200 MA, I'm right there with ya.  However, I'm equally (if not more so) intrigued by:

1.  Why 25/25/25/25?  Haven't others (Clive, I think) shown that better returns with slightly more volatility can be had by reducing cash and increasing equities?  And don't give me any "backtesting" justifications.  Why are we concerned about backtesting when one of the main tenants of the PP is that the future isn't knowable?

2.  Why rebalance everything to 25% as opposed to underweighting (maybe 20% instead of 25%?) the leading asset and overweighting the lagging asset(s)?  Somehow 25% percent is always the magical starting regardless of current asset prices (e.g. current long bond prices) and it's also the magical rebalancing point.

3.  How about the rebalancing bands?  Isn't choosing 20/30 vs. 15/35 pretty arbitrary?

So there are things taken as gospel in every strategy that may not make total sense to the uninitiated.  They just seem to "work" and usually go unquestioned by the converts.  I really like the PP strategy, but it's just as questionable and arbitrary as pretty much every other strategy I've researched.  Although I will admit that PP followers (at least the participants on this board) are far better informed about the risk/reward profile and potential pitfalls than followers of other strategies.

[Ad Orientem]

Market timing is pure bunk.  This is one case where I am going to say skip Harry Browne and read Jack Bogle.  The man has spent most of life building just about the most damning case I have ever read against trying to outsmart the financial markets. There are a lot of things I will concede are open to reasonable debate.  I don't think that market timing is one of them.  Bogle has blown it out of the water.

[craigr]

Completely understand your questioning of the 200 MA, I'm right there with ya.  However, I'm equally (if not more so) intrigued by:

1.  Why 25/25/25/25?  Haven't others (Clive, I think) shown that better returns with slightly more volatility can be had by reducing cash and increasing equities?  And don't give me any "backtesting" justifications.  Why are we concerned about backtesting when one of the main tenants of the PP is that the future isn't knowable?

2.  Why rebalance everything to 25% as opposed to underweighting (maybe 20% instead of 25%?) the leading asset and overweighting the lagging asset(s)?  Somehow 25% percent is always the magical starting regardless of current asset prices (e.g. current long bond prices) and it's also the magical rebalancing point.

3.  How about the rebalancing bands?  Isn't choosing 20/30 vs. 15/35 pretty arbitrary?

So there are things taken as gospel in every strategy that may not make total sense to the uninitiated.  They just seem to "work" and usually go unquestioned by the converts.  I really like the PP strategy, but it's just as questionable and arbitrary as pretty much every other strategy I've researched.  Although I will admit that PP followers (at least the participants on this board) are far better informed about the risk/reward profile and potential pitfalls than followers of other strategies.

All valid points. The difference is that these allocation points are not imposing large transaction fees or emotional inflection points in the decision process.

I'd love it as much as the next guy if there were an accurate system or market timing that allowed me to avoid large losses and lock in big gains. But as much as I've looked, I've not found it yet. 200 day MA may seem like a solution, but the costs involved are quite large in tems of whipsaws, transaction fees, taxes, etc. I just don't think any of it works as well as buy and hold.

There could be an argument for 20% of X and 30% of Y as a better approach. But ultimately I have control over very few things in the market and two of them are taxes and transaction fees. As an investor then it makes sense to optimize around what we have control over and not sweat what we don't control (future asset returns). So as much as I'd like to believe that some kind of market timing is going to give me greater profit, the rational side of me that has looked extensively at the data must conclude that it simply isn't reliable enough to use to make an excess profit. 25% splits may not be optimal, but it is far more reliable, and cheaper, than relying in market timing to accomplish the same goals.

I just don't think market timing works. I have, not in all the years of my experience, ever seen a market timing strategy produce consistent real-world results. Ever. Not once. Backtesting is one thing, but applying that knowledge to the future is basically not profitable in my experience. I'd love for someone to prove me wrong, but I've never once seen it.

[stone]

Clive, is it fair to say that the world isn't populated with people who have made fortunes through being retail momentum traders? To my mind that observation says a lot. It suggests to me that the supposed potential of the approach is just a mirage. Holding a PP won't make anyone a fortune but hopefully will preserve what you have. Perhaps trying for more than that only ends up worse

[stone]

The PP incorporates elements of timing - suggesting that you reduce once the weighting of a single asset rises to 35% or adding when the weighting declines to below 15%. Without such timing the PP is a lot more volatile - which is often considered as being riskier.
Timing can, and often does, reduce risk.

Clive are you saying that say 30% to 20% bands make the PP a lot more volatile? I thought there was little in it? To my mind there are two broad approaches to trading- hedging or momentum. The PP with 30%to20% bands would be firmly in the hedging camp. The 35% to 15% bands are more of a compromise. Descision Moose is an out and out momentum trading approach. Clearly sardines manage to shoal and starlings manage to fly as a flock. A sardine turns in the water in response to what it observes its shoal mates have done up to that moment in time. The sardine can predict the immediate future on the basis of the immediate past. You can't do that with a flipping coin or radioactive decay etc etc. Whether momentum trading makes any sense at all entirely depends on whether there is any momentum to asset prices rather than them just being Brownian motion. I got the impression that there was a bit of momentum but also plenty of random noise and the relative levels of momentum and noise vary over time. The PP will survive even when "momentum is in the air" and will prevail when things go Brownian. Momentum trading will prevail when there is momentum and will get crushed when things go Brownian. My impression was that when ever momentum leaves the markets, the momentum traders give back their prior gains. 
Let's face it, a lot of finance is based on the assumption that there is no momentum:
http://en.wikipedia.org/wiki/Black%E2%80%93Scholes
"The Black–Scholes model of the market for a particular stock makes the following explicit assumptions:
The stock price follows a geometric Brownian motion with constant drift and volatility."

[stone]

There not much difference between trading rules based on relative value (or weighting) to that of trading rules based on relative strength. Additional rules such as amounts traded etc. depict whether the trading is being employed in a defensive or aggressive fashion (or maybe a combination of both).

I do think that there is a massive difference between simultaneously holding two assets and alternately holding one or other of them based on which has been doing best up to that point. The former is a hedging strategy with the rationale that the past is not predictive of the future. That is an explicit assumption of the Black Scholes model for what that's worth:
http://en.wikipedia.org/wiki/Black%E2%80%93Scholes
"The Black–Scholes model of the market for a particular stock makes the following explicit assumptions:
The stock price follows a geometric Brownian motion with constant drift and volatility."

Alternately switching (and suffering the consequent lack of rebalancing gains) only works if the price movements are not Brownian. You are relying on the other market participants following the schoal like sardines and not all acting any which way like cats. I suspect those hedgefunds and investment banks that do use momentum trading always keep a very close eye out for when momentum has evaporated.

[stone]

Clive, check out:-
http://en.wikipedia.org/wiki/File:Wiene ... imated.gif

Even pure randomness can trick us into perceiving a trend if we are not vigilent.

[stock chartist]

Has anyone here tried using a market timing approach using multiple moving averages? I use 4 MAs (50, 100, 200 and 300 dmas) and have come up with a set of conditions which, if the discipline is followed rigorously will produce outstanding results far exceeding a buy-and-hold approach.  I tested it over the whole period 1963-current and for 10-year intervals with excellent results.  The discipline is clearly an "alarm" system for periods that includes Bear Markets and Crashes and doesn't help much when the market is trending up (e.g., 1990's).

[MachineGhost]

The "Top 1%" are populated with people like that.  They're called hedge fund traders.

It's much more accurate and less deceptive to say market timing "doesn't work" for those without the emotional and intellectual fortitude to deal.  That seems to be the vast majority of PPers from what I've seen so far.  There's nothing wrong with this.  It's better to recognize your weakness than to be in denial about it and have it lose you money.

Generally, it does not make sense to combine market timing and rebalancing.  Rebalancing is an act of [hopefully] buying low and selling high with noncorrelated investments.  Market timing only buys when there is a trend and by definition, the trend has already been in motion to get to that recognition point.  So it will always buy moderate to high.  Market timing is a "growth" strategy; rebalancing is a "value" strategy.  They both have their places in investing, but they require different skill sets.

MG

Clive, is it fair to say that the world isn't populated with people who have made fortunes through being retail momentum traders? To my mind that observation says a lot. It suggests to me that the supposed potential of the approach is just a mirage. Holding a PP won't make anyone a fortune but hopefully will preserve what you have. Perhaps trying for more than that only ends up worse

[MachineGhost]

Stock movements are not Brownian motion.  They are serially correlated, because people collectively herd in response to new data.  The literature has been documenting this momentum effect for decades.  Theres lots of other "irrational" effects too, like under-reaction to earnings releases, etc..  In other words, the market is only in a state of homeostasis when not being emotional.  Most of academia is still stuck in an ivory tower that nils that effect out from assuming that everyone is Homo Rationalis, but it has been slowly growing up with more and more revelations from behavioral finance.

Keep in mind though that the majority of acadamia does not personally care about the profit motive and does not eat its own cooking.  The incentives are completely different, more political, that of power, prestige and envy.  So "proof" is defined differently than in the real world.

MG

http://en.wikipedia.org/wiki/Black%E2%80%93Scholes
"The Black–Scholes model of the market for a particular stock makes the following explicit assumptions:
The stock price follows a geometric Brownian motion with constant drift and volatility."

[MachineGhost]

Doesn't this introduce a lot more lag?  How are you using it in conjunction with rebalancing?

MG

Has anyone here tried using a market timing approach using multiple moving averages? I use 4 MAs (50, 100, 200 and 300 dmas) and have come up with a set of conditions which, if the discipline is followed rigorously will produce outstanding results far exceeding a buy-and-hold approach.  I tested it over the whole period 1963-current and for 10-year intervals with excellent results.  The discipline is clearly an "alarm" system for periods that includes Bear Markets and Crashes and doesn't help much when the market is trending up (e.g., 1990's).