Market Timing Schmarket Timing

[ERD50]

...

As I suggested, M-T might find decent discussion on Morningstar or similar site, where there are forums set aside for those who want to discuss a revolutionary concept.

Well there's a self fulfilling prophecy for ya! 'Go take the discussion elsewhere, so we can talk about how no one can talk about that subject here!'?

I'm interested to hear his ideas, please don't chase him away.

-ERD50

 

[target2019]

Well there's a self fulfilling prophecy for ya! 'Go take the discussion elsewhere, so we can talk about how no one can talk about that subject here!'?

I'm interested to hear his ideas, please don't chase him away.

It won't be me doing the chasing.

Just noticed that his introduction was spliced into a thread called Market Timing Schmarket Timing. Maybe that will throw up a caution flag, and he will no longer discuss. We will wait and see.

 

[LOL!]

What I should do is manage a small portfolio and create my own market timing newsletter.

I would enjoy more market timing newsletters over in the Stock Picking forum.

As you know, I have one there which is now in the top-5 most-viewed of non-sticky threads for that sub-forum: http://www.early-retirement.org/forums/f44/lol-s-market-timing-newsletter-57042-11.html

My rule there is that I try to post moves before I submit my orders for them.

 

[Animorph]

As a forum for generally do-it-yourself non-professional retirement investors, where many new members are just starting to handle their own investments, it makes little sense for us to advocate market timing without a specific algorithm, several academic studies supporting it, and some reason to believe the assumed conditions will continue 30 years into the future.

From Morningstar investor returns we already know that, on average, without a plan, most people seem to be really bad at it.

So I'm happy to advocate for buy and rebalance with index funds as a baseline strategy, and I would discourage "market timing" fairly strongly, especially if it is just aimed at avoiding big bear market drops by getting out of the market 100%.

That's a little different than executing a market timing strategy that follows algorithmic signals and doesn't pull you 100% in and out of the market on a moment's notice. But even something like that might be unsuitable for someone who might be reluctant to sell when stocks are high and buy when they are low. Which seems to be true for many.

The more sophisticated investors know who they are and can ignore our more simplistic advice and play with their own money in a way that makes sense to them. And if someone has a nice market timing algorithm with some type of pedigree, by all means give us a link to it and lets us know why you like it. Give us a demo, like LOL!'s posts.

 

[Lsbcal]

As a forum for generally do-it-yourself non-professional retirement investors, where many new members are just starting to handle their own investments, it makes little sense for us to advocate market timing without a specific algorithm, several academic studies supporting it, and some reason to believe the assumed conditions will continue 30 years into the future. ...

I would be satisfied with an algorithm and back test results for at least 50 years.

Academics have a reputation to protect that may have nothing to do with market performance. The algorithm has to be logical i.e. correlation to something like butter production in Bangladesh is not going to hold water.

BTW, I do have a nice spreadsheet that tests a 200 day moving average methodology back to 1950. It is the sort of thing that convinces me not to do that sort of thing. One really does have to look at the moving parts to understand the nuances. Some moving average methods using monthly data might work best. One person on Bogleheads actually did a real time experiment. I could not believe anyone would bother without looking at the backtest results. Frequent whipsaws are really part of the game. I think he stopped after getting 1 whipsaw.

 

[Animorph]

I would be satisfied with an algorithm and back test results for at least 50 years.

Academics have a reputation to protect that may have nothing to do with market performance. The algorithm has to be logical i.e. correlation to something like butter production in Bangladesh is not going to hold water.

I thought about that, but if you throw enough variables into the mix it is very easy to get fantastic backtesting results. So if it was an algorithm that looked at just one raw data input I'd be pretty good with it, but five inputs probably not. Just a simple curve fitting problem.

As an engineer for a private company I had little use for most academic research. Most were just trying to find a problem for their solution. But I put that in to at least suggest a peer-reviewed algorithm that had shown positive results versus what several people threw at it. A single proponent could just be blowing smoke, but I might spend some time with it if several independent sources obtained the same results.

 

[target2019]

I would be satisfied with an algorithm and back test results for at least 50 years.

Academics have a reputation to protect that may have nothing to do with market performance. The algorithm has to be logical i.e. correlation to something like butter production in Bangladesh is not going to hold water.

BTW, I do have a nice spreadsheet that tests a 200 day moving average methodology back to 1950. It is the sort of thing that convinces me not to do that sort of thing. One really does have to look at the moving parts to understand the nuances. Some moving average methods using monthly data might work best. One person on Bogleheads actually did a real time experiment. I could not believe anyone would bother without looking at the backtest results. Frequent whipsaws are really part of the game. I think he stopped after getting 1 whipsaw.

I think I read somewhere that whipsaws could be nullified by using longer timing signals, or setting up channels around the MA.

 

[dallas27]

I would discourage "market timing" fairly strongly, especially if it is just aimed at avoiding big bear market drops by getting out of the market 100%.

I think the only exception(i hope you agree) is the "blood in the street" opportunities when the market is crashing big-time, a la housing crash. We all know it's coming back eventually, the trick is do have cash or equivalent at those times. Buffet calls it the soft pitch. But timing/opportunity could also be used.


Sent from my iPhone using Early Retirement Forum

 

[Lsbcal]

I thought about that, but if you throw enough variables into the mix it is very easy to get fantastic backtesting results. So if it was an algorithm that looked at just one raw data input I'd be pretty good with it, but five inputs probably not. Just a simple curve fitting problem.

As an engineer for a private company I had little use for most academic research. Most were just trying to find a problem for their solution. But I put that in to at least suggest a peer-reviewed algorithm that had shown positive results versus what several people threw at it. A single proponent could just be blowing smoke, but I might spend some time with it if several independent sources obtained the same results.

I hope you do not think I'm being picky. I read several posts on Bogleheads saying it is easy to get fantastic backtesting results with several variables. But I've not seen a demonstration of this. I would really like to see a great backtest that covers several decades with fantastic results and maybe 5 variables.

 

[Lsbcal]

I think I read somewhere that whipsaws could be nullified by using longer timing signals, or setting up channels around the MA.

That would be interesting to see.

The SMA results I got were similar to a study by Siegel in his book Stocks for the Long Run (3rd edition). My modest study used a small band of 1% around the sell/buy points. Good results for the 2010's, poorer then buy-hold for the 1990's, etc. From 1950 - 2014 there were 106 sell/buy trades and 88 whipsaws (worst one was -5%). Trying to do this on a daily basis would be painful in my opinion. Also impractical as life events can get in the way.

 

[rayvt]

I Some moving average methods using monthly data might work best. One person on Bogleheads actually did a real time experiment. I could not believe anyone would bother without looking at the backtest results. Frequent whipsaws are really part of the game. I think he stopped after getting 1 whipsaw.

I have a backtest spreadsheet for the S&P500 (including re-invested dividends), monthly, Jan 1950 to Jan 2013. 10 month SMA, like Faber. But sell at 3% below the SMA, not at simply under the SMA. That simple modification eliminates almost all the whipsaws. 50 trades total (25 round-trips) in 63 years.

S&P500 buy-and-hold
CAGR 10.81%
StDev 10.80%
MaxDD -46.35%
Sortino 0.87 (3% MAR)

S&P500 with SMA timing. (1-yr Tbill when out)
CAGR 11.5%
StDev 12.1%
MaxDD -23%
Sortino 1.11 (3% MAR)

 

[Texas Proud]

I don't care to convince anyone to become a market timer. I merely stated what has worked for me in the past. No one knows what the future holds. It is quite possible that those who sit on a 50% decline in the future may have to wait decades to be made whole. Of course, this can happen to market timers as well.

What has not changed throughout history is that investors get greedy and fearful. This can be exploited. Even a simple 10 or 12 month moving average system in a diversified portfolio shows reduced drawdowns without hurting performance (Mebane Faber research). Portfolio insurance is a viable option for those who insist on being fully invested. The point that is beyond technical dispute is that some with the right temperament have been successful market timers. Some on this board disagree with this assertion.

Not to discount YOUR return.... but one person's returns does not a good investment style make... you have to compare many people and see if there is a trend to the timing...

As an example, my old boss sold everything he had just before the market crash.... he sat in cash... good move...

Then, he sat in cash... bad move...

Then he finally decided to invest again... bought silver and gold.... good move when he did it, but did not get out and is now a really bad move...

He is down from where he was in early 2008... by a lot...

I just kept investing... yep, lost a lot there for awhile... but kept putting in my monthly amount to my 401.... now I am UP big time....

Who was smarter in this example Why buy and hold of course...


So, personal examples are great and all.... but I would rather see a study done to see if there is a benefit in market timing.... and the ones that I read say it is not.... now, if you can find some that show there is benefit to doing it... and not just a one off example... I am all for reading it...

 

[steelyman]

I would enjoy more market timing newsletters over in the Stock Picking forum.

As you know, I have one there which is now in the top-5 most-viewed of non-sticky threads for that sub-forum: http://www.early-retirement.org/forums/f44/lol-s-market-timing-newsletter-57042-11.html

My rule there is that I try to post moves before I submit my orders for them.

I've posted in LOL!'s thread and enjoy it although my nature is more like a turtle when it comes to investing.

 

[eta2020]

I would be interested if OP posted into this thread on the day when jumps back into the market.

It is funny but today (my net Worth) I am exactly where I was when OP sold his stocks.

I did not sell anything and I bought 1000 of KO at 39.50 since then.

And I have no taxes to pay on any trading My invest-able assets were and are few million dollars 100% in equities.

 

[Lsbcal]

I have a backtest spreadsheet for the S&P500 (including re-invested dividends), monthly, Jan 1950 to Jan 2013. 10 month SMA, like Faber. But sell at 3% below the SMA, not at simply under the SMA. That simple modification eliminates almost all the whipsaws. 50 trades total (25 round-trips) in 63 years.
....

Sounds reasonable to me. To clarify just one point, you do a sell at 3% below SMA and then buy back at what point? Maybe 2% below SMA or do you have a wider band, maybe buy back when the price hits the SMA?

I have an out of sample data set that I could throw at this.

---------------------------
I was curious about this so created my own test set. Using SP500 with dividends and 5 year Treasuries from Feb 1950 to July 2014:

CAGR for SP500 only = 10.9%

CAGR for SP500 and 5 yr Treasury = 13.0%
sell when SP500 3% below 10 month moving average
buy back when SP500 1% above moving average
21% of time in Treasuries, 21 trades

 

[target2019]

Sounds reasonable to me. To clarify just one point, you do a sell at 3% below SMA and then buy back at what point? Maybe 2% below SMA or do you have a wider band, maybe buy back when the price hits the SMA?

I have an out of sample data set that I could throw at this.

Can't answer for others, but if you set bands of say 2.5-5% above and below the SMA, you can consider getting out when the SMA pierces the lower band, and get back in when the SMA breaks through the SMA on the way up, or even wait for the SMA to pierce the upper band.

What I found in my own personal study of this with real money, was that reacting to these signals is more complex that it seems. I used a mutual fund. It took more time to track these things with public data than I ever thought.

 

[bobafett]

Just curious but..

Instead of selling in and out of large equity positions wouldn't it be more practical to use things like SMA timing to hedge with puts or e-mini shorts like in that other thread?


You're not risking nearly as much when you do something like this.

 

[bob boag]

It is funny but today (my net Worth) I am exactly where I was when OP sold his stocks.

OP here, after today's close I am also just about exactly where I would have been had I done nothing...

 

[NW-Bound]

You both made me look. Lemme see...

On 7/30 when this thread started, the S&P was 1970. It is now 1972.

I am up $5755. Add to that the $3202 that I spent in that interval of time, and I am up $8957. How's that for one-upmanship?

In terms of percentage of portfolio, it's nothing, but hey, a $K is a $K. For short-term trading of just 2 or 3% of portfolio, if I can make a bit of extra money to help pay for gas for my motorhome, I am happy.

I am leaving in 3 days to embark on that 2-month long and 9,000-mi trip to the Canadian Maritimes. Won't be posting here much when I am on the way.

 

[Lsbcal]

...
I am leaving in 3 days to embark on that 2-month long and 9,000-mi trip to the Canadian Maritimes. Won't be posting here much when I am on the way.

Getting out of town while you are ahead. Sounds like a good poker move. Have a good trip!

We are going to the High Sierras this coming Saturday to beat the Labor Day crowd.

 

[Animorph]

I think the only exception(i hope you agree) is the "blood in the street" opportunities when the market is crashing big-time, a la housing crash. We all know it's coming back eventually, the trick is do have cash or equivalent at those times. Buffet calls it the soft pitch. But timing/opportunity could also be used.


Sent from my iPhone using Early Retirement Forum

Absolutely, been there, done that, posted about it. But still, there's a problem of when. The market goes down 20%, how much of your cash goes in? 30%, 40%, 50% down, how much goes in? If it goes down 20% and you stay out, it could recover and you never get in. If it goes down 20% and you go all in, it could go -50% or more before it recovers.

I split my excess cash into five equal parts and invested in -5% increments from -20% down and lower in 2008-2009. Then I had to find a bit more to add at -45% and -50%. If it had continued down I would have been left to shifting from large caps to small caps or something similar. Almost did that too.

Anyway, all algorithmic, no judgment calls beyond the original setup, executed according to plan. The cash was stockpiled because I was retiring in late 2007 and wanted insurance against just such an event. If the market had continued up I would have just spent the cash before touching equities. Not something I would normally do during accumulation.

 

[Animorph]

I hope you do not think I'm being picky. I read several posts on Bogleheads saying it is easy to get fantastic backtesting results with several variables. But I've not seen a demonstration of this. I would really like to see a great backtest that covers several decades with fantastic results and maybe 5 variables.

Sorry, no examples. Repeated from various readings, and from curve fitting experience.

Say you get a rough match to S&P 500 total return using P/E10. Now maybe you can get a better match by also considering yearly profit margins, so you use that. Maybe using P/E with expected earnings doesn't help, so you can discard that. So if some guy is using five inputs you can assume each added input improved the fit.

So that guy could be claiming his 5 factor market timing scheme generated 1% better gains on average in back testing than a 3 factor baseline scheme or simple buy and hold. The problem is, he is likely just matching the past data more and more closely as he adds inputs to his scheme. He could get so he predicts at least a few of the past market crashes fairly accurately, which is all he has to do to look pretty good. 0.51 * price to book instead of 0.50 * price to book gives him a sell signal right on the peaks in 2007 and 2000. The more inputs he uses, the more he can fine tune the results. You could have perfect 20/20 hindsight with enough complexity.

That's a degree of accuracy that just isn't present in the market as it continues into the future. And why really complex schemes that work really well in back testing can completely fail when used to predict the future.

 

[eta2020]

You both made me look. Lemme see...

On 7/30 when this thread started, the S&P was 1970. It is now 1972.

I am up $5755. Add to that the $3202 that I spent in that interval of time, and I am up $8957. How's that for one-upmanship?

In terms of percentage of portfolio, it's nothing, but hey, a $K is a $K. For short-term trading of just 2 or 3% of portfolio, if I can make a bit of extra money to help pay for gas for my motorhome, I am happy.

I am leaving in 3 days to embark on that 2-month long and 9,000-mi trip to the Canadian Maritimes. Won't be posting here much when I am on the way.

Well when I wrote I am exactly where I was on 7/30 I meant within plus and minus 15k. I don't remember exact sum I had and since we are still working we probably added another 5-10k of income into the pile.

So maybe you beat me But IF market from here goes up OP will not beat me and I did zero work to get here.

 

[rayvt]

Sounds reasonable to me. To clarify just one point, you do a sell at 3% below SMA and then buy back at what point? Maybe 2% below SMA or do you have a wider band, maybe buy back when the price hits the SMA?

The last. Buy when it rises to the SMA.

IMHO, whipsaws are not a big deal, but mainly annoying and costing you trading costs, commissions, etc. I suspect that the bad ones and the good ones cancel out, since whipsaws are just like a relay chattering just as it it on the verge of flipping. Haven't backtested this, though.

 

[bob boag]

OP checking in.

Yeah I regret it down a couple grand