Bear market and 2008

StrataSearch will find what you tell it to, but what should that be? Will the same criteria work across different Sectors? Different time periods? Here we discuss the ins and outs of OneClick Searches.

Bear market and 2008

Postby Schutten » Tue Jan 18, 2011 3:20 pm

Hi,

I would like to know how people handle the 2008 bear market. It seems that the market in 2008 is a different type of bear market then wat happenend in 2000-2003. When I google 2008 markets I sometimes find the term 'financial meltdown'. When I look for diversification products there is almost nothing available, because for that period even commodities which is normally a good diversification product shows strong signs of correlation with the sp500 in the period 2008.
Any suggestions or opinions?

regards,
Dennis
Schutten
 
Posts: 42
Joined: Thu Oct 25, 2007 6:17 am

Re: Bear market and 2008

Postby Kevin_in_GA » Tue Jan 18, 2011 8:35 pm

I totally screwed up handling this, but have since figured out a relatively simple way to avoid it in the future. If the three month return on the S&P 500 goes negative, go to cash.

proc(close,63) < 0 means go to cash.
Kevin_in_GA
 
Posts: 171
Joined: Wed Aug 04, 2010 10:14 am

Re: Bear market and 2008

Postby Schutten » Wed Jan 19, 2011 9:47 am

This might work ok. But did you optimize this value? If so then this might lead to curve fitting....

Regards,
Dennis
Schutten
 
Posts: 42
Joined: Thu Oct 25, 2007 6:17 am

Re: Bear market and 2008

Postby Kevin_in_GA » Wed Jan 19, 2011 10:05 am

Sort of - it is based on the work of Mebane Faber, who has looked at Tactical Asset Allocation strategies for the major markets. In short, one buys the stock/asset that has shown the strongest perfromance over the preceding 3 months (he looked at 3,6, 9, and 12 months - all worked better than buy & hold, and all steered you clear of the 2008 market collapse).

http://www.mebanefaber.com/2009/02/19/a ... n-updated/

I would love to optimize this, but over the longest period SS can backtest (since 1985). I am still trying to get a simple rank function to work for selecting buy/sell, and Pete and crew have been helpful in getting the pieces on place. Basically rank each asset class by proc(close,@1days) with the signal to buy the top scoring asset class and keep it until it no longer is at the top. Then buy the new one. Could also use weekly values, etc to see if that smoothes the data and gives you fewer whipsaws.

Kevin
Kevin_in_GA
 
Posts: 171
Joined: Wed Aug 04, 2010 10:14 am

Re: Bear market and 2008

Postby Dacamic » Thu Jan 20, 2011 1:31 pm

Kevin_in_GA wrote:I totally screwed up handling this, but have since figured out a relatively simple way to avoid it in the future. If the three month return on the S&P 500 goes negative, go to cash.

proc(close,63) < 0 means go to cash.

What method do you use to decide when to get back in?
Steve
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Re: Bear market and 2008

Postby Kevin_in_GA » Thu Jan 20, 2011 2:27 pm

Basically when proc(close,63) crosses back above 0, but in truth it makes more sense for it to be higher than bonds or other investments. If you read the Mebane Faber paper I cited above, it is very well explained and back-tested.
Kevin_in_GA
 
Posts: 171
Joined: Wed Aug 04, 2010 10:14 am

Re: Bear market and 2008

Postby Dacamic » Fri Jan 21, 2011 11:51 am

Ah, yes ... I remember that paper. Its underlying thesis isn't revolutionary; nonetheless, Faber did a nice job quantifying some important points.

For what it's worth, StrataSearch has many pre-built trading rules intended to provide the macro overview discussed by Faber. If my memory is reasonably reliable, those rules might have parameters of a relatively short-term nature. They can be easily modified, though, to allow exploration of filters similar to those you mentioned. (I don't remember having ever done so ... hmmmm ... that might be a good idea for a new search ... thank you for the tip).
Steve
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Re: Bear market and 2008

Postby Kevin_in_GA » Fri Jan 21, 2011 3:03 pm

I'd love to see which pre-built rules you're thinking of here, and what you find if you do the search. I posted recently a question on how one might buy/sell based on the rank of a particular indicator. So far the proposed custom formula looks good but does not seem to be working as expected. Not sure why, but I'll be posting some recent results and info later today.
Kevin_in_GA
 
Posts: 171
Joined: Wed Aug 04, 2010 10:14 am

Re: Bear market and 2008

Postby Dacamic » Mon Jan 24, 2011 1:44 pm

Pre-built trading rules can be viewed by opening any AutoSearch Setup. The "macro level" rules I mentioned are listed under the Supporting Entry and Exit tab, and they have one of the following labels in their name:

    Sector
    Market
    Index
Steve
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Re: Bear market and 2008

Postby Schutten » Tue Jan 25, 2011 3:48 pm

I can confirm that proc(close,63) indeed seems to works in 2008. Still my wory is there about a optimized value.
In my opinion short term changes in the market can be overcome by stoplosses, long term changes by a filter or even by changing from long to short trading or vise versa.

Place a moving average on the SP500 index like mov(close,150,simple)>ref(close,150,simple),-1) is idea as well, but also this value is heading to a optimized value.

Anyone any more ideas?

Regards,
Dennis
Schutten
 
Posts: 42
Joined: Thu Oct 25, 2007 6:17 am

Re: Bear market and 2008

Postby Dacamic » Thu Jan 27, 2011 7:48 pm

Dennis,

Even assuming those rules are robust, they still have a possible downside: they might cause your system to underperform in secular bull markets.

The Simple Moving Average (SMA) system studied by Faber is intended to reduce volatility without having a significantly adverse effect on returns. It achieves this goal by producing an equity curve that smooths out the peaks and valleys of the S&P 500. By doing so the system tends to lag behind a rising S&P 500, and it catches back up when that index drops. Along those lines, Faber noted in his paper that the SMA system underperformed the S&P 500 in about half of all the years since 1900. I am guessing this is because the SMA system's timing is good for the plunges of bear markets, yet not so good for the comparatively gentler dips of bull markets.

Faber concluded that the benefits of the SMA system accrue over the course of entire business cycles, i.e., 30-40 years. This time frame seems impractically long for most people, and thus makes the SMA system itself vulnerable to poor timing in practical applications.
Steve
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Re: Bear market and 2008

Postby Kevin_in_GA » Fri Jan 28, 2011 7:32 am

Steve:

I agree - the use of any moving average with the PROC(63) system I mentioned results in a lower return. Once again showing the connection between volatility and return. This is why I like the proc approach - it provides a relatively clear and timely signal, but should be used in a TAA approach with other asset classes in the mix to really do well.
Kevin_in_GA
 
Posts: 171
Joined: Wed Aug 04, 2010 10:14 am

Re: Bear market and 2008

Postby Schutten » Fri Jan 28, 2011 3:07 pm

Besides that, profit is nice, but I myself prefer a smoother equity line above max profit. It's always possible to use leverage to maximize gain.
Schutten
 
Posts: 42
Joined: Thu Oct 25, 2007 6:17 am

Re: Bear market and 2008

Postby Dacamic » Fri Jan 28, 2011 3:19 pm

Schutten wrote:Besides that, profit is nice, but I myself prefer a smoother equity line above max profit. It's always possible to use leverage to maximize gain.

But, alas, leverage also makes the equity curve less smooth.
Steve
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Re: Bear market and 2008

Postby Schutten » Fri Jan 28, 2011 3:23 pm

Let's put it in another way, my focus is on reducing drawdown and not on max profit....
Schutten
 
Posts: 42
Joined: Thu Oct 25, 2007 6:17 am

Next

Return to OneClick Searches and Scoring

cron