Thoughts for setting a One Click "envirnoment"

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.

Re: Thoughts for setting a One Click "envirnoment"

Postby Overload » Tue Apr 14, 2009 4:06 pm

Good post. Thanks for sharing your thoughts and methods.

Pete
Overload
 
Posts: 2248
Joined: Wed Nov 30, 2005 12:14 pm

A Simple Bear Market Filter

Postby cme4pif » Thu Apr 30, 2009 12:35 am

Hi Gang;
I was reading this forum again because I recalled it had some of Steve's methods for using SS.

There is an article on a simple "in a bear market" indicator on page 38 of the April 09 issue of Active Trader.
It mentions that all G7 markets have had a drop of 75% at least once in the last 100 years.
So of course an automated system that could prevent that drawdown would very valuable.
Once Capital is drawndown it is much harder to replace.
The article sights a paper
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=962461
but the heart of it, if you time the market on the basis of "Go long when the S & P 500 average is above the 10Month simple Moving Average, and sell when it is below."

The result is in years 1900 - 2008:
S & P buy and hold = 9.79%
S & P with cross over signal 11.33%

So if you simple added that signal to a system based on longs it should improve the results.

Cheers
Blaine
cme4pif
 
Posts: 35
Joined: Thu Jan 22, 2009 2:08 pm

Re: Thoughts for setting a One Click "envirnoment"

Postby Overload » Thu Apr 30, 2009 2:38 pm

The S&P crossing over its 200 day moving average may be better than nothing at all, but I'm pretty confident there are better ways. Moving averages are known to be "lagging" indicators, and a 200-day average is going to have quite a bit of lag. Even further, there can often be a number of "false signals" with such simple crossovers. Looking at a chart of this method, I counted roughly 20 false signals between 1990 and 2008. That might not seem like many, but there were only about 25 signals altogether during that time. So, again, the approach might be better than nothing at all, but I believe there's better things that can be done.

Regardless, any idea is worth rolling around, so thanks for passing it on.

Pete
Overload
 
Posts: 2248
Joined: Wed Nov 30, 2005 12:14 pm

Re: Thoughts for setting a One Click "envirnoment"

Postby TomHam » Thu Apr 30, 2009 9:27 pm

Some good & interesting posts on "envirnoment settings" (a.k.a. time period) for a StrataSearch One Click runs.
While I'm still learning the entire program (it's not hard to use, just so full of useful tools that I want to try everything :) ) At this point I'm staying with my orginal concept that the selection of the envirnoment that you give SS to run in (as well as the members of the sector), has a big effect on your results.
I'm seeing that it can find some high performance, robust systems in one market envirnoment, that do not do well in another. That seems to hold for Long onlys, and Long & Short systems. When I try an "all weather" approach, performance drops significantly (tho still O.K.). I may change my mind later, but a very good all waether system is elluding me (so far).

But that brings up the item of "how do you know when your system breaks ?". The good news is that SS has a great system performance report, plus performance charts to give us the information when a system starts to "stray" from the anticipated results. My person favorite is the MAE chart and the max trade drawndown information. That may be a method to determine when a system / multi-strategy that was effective in one envirnoment is in the process of "breaking". Either the envirnoment is changing or the actions of the secotr members have, in relation to the formulas.

I like SS because it allows you to focus on the definition of exactly what you are looking for, and gives you the results analysis to point you toward making the changes that might be effective to improve future results. I'm learning a lot about system development . . . it's NOT as easy as others would have you to believe (with other software . . . just run an optimization . . . yeah, right !). Good stuff guys. I just wish I had an octi-core processor. :o

Thoughts & comments are most welcome. Good Trading. ............. Tom .............
TomHam
 
Posts: 18
Joined: Sun Jan 18, 2009 1:41 pm
Location: near Detroit, Michigan

Re: Thoughts for setting a One Click "envirnoment"

Postby Overload » Fri May 01, 2009 9:01 am

But that brings up the item of "how do you know when your system breaks ?".

First I think there needs to be a distinction between a system not following its expected performance in the first place, and a system deviating from its expected performance after a certain period of time.

When beginning to trade a new system, I think it's important to monitor the trades very closely and make sure your actual results are in line with what the StrataSearch results have predicted. You can do this by simply including your live trading period within a back test of the system. Primarily you're looking to see if your net profit per trade is matching the net profit per trade listed in the back test. There are several reasons why they might differ. For example, StrataSearch assumes you'll be purchasing at the opening price, but there might be some slippage actually getting that price. Or there might be a larger spread than you had configured in the back test. Or you might find you're not able enter limit orders at the expected rate. If there's any difference between what your the back test expects, and what you're getting during live trading, you need to adjust your Trade Settings and rerun that back test to see how the system will perform with these new parameters. Again, I believe it's very important to do this when starting to trade a new system, just to be sure your system is following the expected performance in the first place.

Once you're sure the system is performing properly, it's also important to monitor it to make sure it doesn't deviate from the back test projections. To do this, I think one of the best ways is to compare many of the current performance values with those of your back test evaluation period. For example, suppose your back test showed that your system experienced a couple 10% drawdowns, but nothing greater than 15%. If you then see a drawdown of 20% in your live trading, then your system has moved outside of normal operating parameters and it's time to seriously examine your system and perhaps abort. There are numerous values that can be looked at in this same way, including:

Equity Drawdowns
Worst Trade Return
Number of Consecutive Losses
Worst Year or Month Return

Over the longer term, values like Average Annual Return, Percent Profitable, Percent in Market, Days Held, etc., should be monitored as well to ensure that the system continues to follow the path expected by the back test. But I think the above values are important to watch on a more regular basis, as these can give a more immediate indication that a system has deviated from its expected performance.

I suppose one could also evaluate whether their system is operating outside of normal parameters on the upside (better than expected returns, etc), but this is probably a "failure" that few would complain about. And it just might be because your back test projections were simply too conservative.

Pete
Overload
 
Posts: 2248
Joined: Wed Nov 30, 2005 12:14 pm

Re: A Simple Bear Market Filter

Postby Dacamic » Wed May 06, 2009 11:16 am

cme4pif wrote: ...There is an article on a simple "in a bear market" indicator on page 38 of the April 09 issue of Active Trader ...

Thank you for posting the link to the paper referenced in the article, Blaine. I found it to have very good insights.

Without benefit of dividends or interest, my back tests on the S&P 500 showed the subject indicator had a compounded annual return of about 7% from January 1901 to April 2009. It is difficult to accurately include effects of dividends and interest; nonetheless, it's reasonable to believe they would boost returns to a level close to that stated in the paper: 10.45%. Also as stated, the indicator suffered a 50% drawdown (in the early 1930's), but it was recovered in only three years. Overall, the performance of this simple indicator was consistent during the past 109 years. It seems to have potential as a stand-alone system -- as well as a supporting indicator -- so, I plan to further ponder how to include it in my trading.
Steve
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Re: Thoughts for setting a One Click "envirnoment"

Postby averdo » Fri May 27, 2011 6:49 am

Hi,

I was wondering if anybody has made any further progress in creating a bear market indicator that performs better than e.g. a MA cross over.

I currently use a EMA200 over EMA50 cross over but it is rather laggy and can create false signals, especially if used in multiple indexes/sectors. I use it to switch from a bull market multi system (incl short strategies to a bear market multi system (incl. long strategies). My bull market multisystem even shows positive returns in periods 200-2001 and 2008 but the bear market multisystem shows much better results.

Using the EMA crossing for switching sytems already gives a better performance than just always using the same multi-system but I'm sure further improvements will be possible.

Looking forward to your ideas.
averdo
 
Posts: 11
Joined: Mon May 09, 2011 4:48 am

Re: Thoughts for setting a One Click "envirnoment"

Postby averdo » Fri May 27, 2011 8:56 am

Hi Peter,

Bye the way, I tested the above mentioned suggestions of the last day of the month's prices crossing the 10 months simple moving average and this is a much better indicator then my MA 200/50 crossing. I don't know how to programm this in SS however. Would you have any tips for this?

regards,
Alfred
averdo
 
Posts: 11
Joined: Mon May 09, 2011 4:48 am

Re: Thoughts for setting a One Click "envirnoment"

Postby Overload » Fri May 27, 2011 4:40 pm

I haven't worked with this indicator in awhile, but you'll want to use the period() formula to create the monthly prices. So you can add this filter to your Entry String:

period(MONTHLY, close) > period(MONTHLY, mov(close, 10, simple))

and this filter to your Exit String:

period(MONTHLY, close) < period(MONTHLY, mov(close, 10, simple))

Hope this helps.

Pete
Overload
 
Posts: 2248
Joined: Wed Nov 30, 2005 12:14 pm

Re: Thoughts for setting a One Click "envirnoment"

Postby averdo » Mon Jun 06, 2011 6:11 am

Hi Pete,
thanks, that worked just fine. I want to use this long term trend change indicator for selecting the trading system for several indexes. The 10 months period leads to different timings for different indexes. As it will only be used as a long term trend indicator that should apply for multiple indexes I would like to use this indicator only on the SP500 prices and use the outcome also for other indexes like the DAX.

Is it possible to select the SP500 in this formula so it uses the SP500 prices even though the strategy is used for other indexes?
averdo
 
Posts: 11
Joined: Mon May 09, 2011 4:48 am

Re: Thoughts for setting a One Click "envirnoment"

Postby Dacamic » Mon Jun 06, 2011 10:10 am

averdo wrote:Is it possible to select the SP500 in this formula so it uses the SP500 prices even though the strategy is used for other indexes?

The Symbol reference formula will allow you to do what you described:

    Entry String -- symbol(!SPX, period(MONTHLY, close)) > symbol(!SPX, period(MONTHLY, mov(close, 10, simple)))
    Exit String -- symbol (!SPX, period(MONTHLY, close)) < symbol(!SPX, period(MONTHLY, mov(close, 10, simple)))
In QuotesPlus, !SPX is the symbol for the S&P 500. You should use whatever the appropriate symbol is in the price service you are using.
Steve
Dacamic
 
Posts: 457
Joined: Wed Nov 30, 2005 12:40 pm

Previous

Return to OneClick Searches and Scoring

cron