Automatic Pattern Search

Dear Listmembers,
as I understand with APS you can choose to use a % exit (stop).
Based on what input would you choose an appropriate value ?

Would a normalised ATR (average true range) be a good choice ?
(i.e. ATR/Close)
 
Question to ronblack:

Ron, I have two questions to you, if you do not mind.

1) Did you have any positive results applying genetic selection to APS patterns?
2) I have no doubts that APS, if used well, can give a trader a winning bias. But how many trades did you have in a single year?

Alex.


Hello,

1. I'm in the process of modifying a genetic programming algorithm I developed while ago as part of a graduate course project to take as input the APS patterns and the operators AND and OR. I don't have any results yet because my progress is slow due to other projects and trading.

2. If you properly select the patterns and use them in ways to maximize the probability of win you have a winning bias after slippage, commissions, etc. I like the APS patterns because they reflect price action and nothing more. But keep in mind to to profit you need to go beyond patterns in your strategy. As far as trades, I use the following numbers as minimum sample size (sum of all patterns signals):

Position trading: 300 trades/yr minimum
Intraday trading: 1000-5000 trades/yr depending on timeframe

I doubt anything less than that can be classified as significant in terms of sample size.

As far as the number of bars in the historical data file I more or less agree with the developer suggestions:

Data FAQ

Ron
 
Last edited:
I think adding the profit factor to the search parameters and results in the next version will make the program more useful. Good move:

APS Automatic Pattern Search Version Releases


I too think the changes in the latest version are significant. Especially the increase in patterns and the forward testing capability. I have v4.9 and I will upgrade right away. By the way they just released the new version and I am testing the demo. The search results have a different look as the profit factor is included, which makes it much easier to identify good patterns. You can see an example of the updated results screens here:

Discover trading systems automatically with APS Automatic Pattern Search

I wish good health, happiness and a prosperous New Year to all of you!

Alex
 
I too think the changes in the latest version are significant. Especially the increase in patterns and the forward testing capability. I have v4.9 and I will upgrade right away. By the way they just released the new version and I am testing the demo. The search results have a different look as the profit factor is included, which makes it much easier to identify good patterns. You can see an example of the updated results screens here:

Discover trading systems automatically with APS Automatic Pattern Search

I wish good health, happiness and a prosperous New Year to all of you!

Alex

Alex, I wish to you health and prosperity throughout the New Year.

I just upgraded to APS 7. This version is improved a lot. Definitively, the addition of the profit factor in the search criteria and results is a major one. I wish they have added Amibroker code generation.
 
Alex, I wish to you health and prosperity throughout the New Year.

I just upgraded to APS 7. This version is improved a lot. Definitively, the addition of the profit factor in the search criteria and results is a major one. I wish they have added Amibroker code generation.

I upgraded to v7 a couple of days ago. The program is much improved and I like both the test function and the calculation of the 1-bar success rate in scan results.

I get more patterns with this new version and it appears the search speed is slightly improved.

I agree that the addition of the profit factor resulted in a major improvement.

Ron
 
Has anyone able to install the older 4.9 version on Vista machine?
Also I am getting only Short patterns and I even tried 19 yrs of data for IBM Dialy with Pattern success rate of only 25% and still no results.
I get only short patterns and they are one day patterns it shows enter at Open next day and Exit EOD.
What am I doing wrong?
 
Check your data and T/S file. I get many good patterns IBM using 10 yrs of historical data. What are your profit target and stop loss values?

I don't know about Vista since since I use XP for my trading because I believe it is more stable.

Check this out:

Installation FAQ
 
Alex, Thks I noticed data file had extra 1500 every line. I got only 1 Long Pattern with a very small (uncomfortable) to trade Profit 4 and Stop 2 on expanded search. I have 19 Yrs of IBM data uptill 2007 I thought I will use 2008 & 2009 to test the old patterns on it .. What can I be doing wrong. What parametres you recommend. I had 66% profit less than 4 cons losses,std stuff ..
 
Alex, Thks I noticed data file had extra 1500 every line. I got only 1 Long Pattern with a very small (uncomfortable) to trade Profit 4 and Stop 2 on expanded search. I have 19 Yrs of IBM data uptill 2007 I thought I will use 2008 & 2009 to test the old patterns on it .. What can I be doing wrong. What parametres you recommend. I had 66% profit less than 4 cons losses,std stuff ..

I think with a reward:risk ratio of about 2 and 66% success rate you are asking too much from the program. With 4% target and stoploss I get 23 patterns with normal search and high profit factors using about 16 years of historical data. Note that I have version 7.0. I think the addition of the profit factor in the search criteria and also in the results made this program much more useful.

Alex
 
Dear Alex,

I see that you are an experienced user of APS and most probably a very experienced trader too. I'm thinking of getting a copy of this program and I would like to hear from you if it does what it's developers describe on their website.

Jim
 
I do not think this program would help to experienced traders. This program may help you for some period of time, but testing it on "new data", using few alternative aproaches, I see that it may lead you to a disaster in long run. And APS actually understands it, recommending to rerun the search every 6 months for daily bars. But every rerun means invalidation of the work you did before.
Any trading system must be based on some well sounded idea. APS has one, when it teaches how to calculate and use profitability. But that's it. Just searching for random combination of bars, where only boolean logic was under consideration (i.e. true or false, 1 or 100 points difference in price considered as same statement), and offering this statistical result of the past will not give you an edge in thrading business. Everyone should understand clearly, this software does a good job finding what was promised - combination of bars based on your criteria. It is a good use for theorists, but for traders it is under a big question.
 
Dear Alex,

I see that you are an experienced user of APS and most probably a very experienced trader too. I'm thinking of getting a copy of this program and I would like to hear from you if it does what it's developers describe on their website.

Jim

It does what is supposed to do and much, much more if you can figure out ways to use patterns to configure advanced systems.

Alex
 
Last edited:
Just searching for random combination of bars, where only boolean logic was under consideration (i.e. true or false, 1 or 100 points difference in price considered as same statement), and offering this statistical result of the past will not give you an edge in thrading business.

I do not agree that APS searches for random combinations of bars. Actually, it does not do that:

Using APS FAQ

If you have the program just look at the results. Most patterns exhibit some kind of logic and order, for example double bottoms, key reversals, etc. Also, random patterns are unlikely to stay profitable any any market for 6 years as it is shown in this article:

Tradingpatterns.com Michael Harris paper on pattern robustness

I know there are some other programs that do combinatorial searches but this is not what APS does to the best of my knowledge. Maybe some people think that because Michael Harris presented a combinatorial method of finding patterns in one of his books but I was told by his staff that APS is not based on such method.

Alex
 
Michael Harris idea is a simple one (to find statistically profitables price patterns) but the way he presented it is quite deep and somewhat circumvoluted (he's a systems engineer, for goodness's sake!) so he is bound to confuse those people who are less inclined to use maths for finding solutions which can be found in a less exacting and more intuitive way.
The APS is a data mining engine (Harris himself made this clear) and it does what it says - boy, can you find patterns "in them thar hills" when you run it.
Trading successfully is, however, more than selecting a few patterns (harris was also quite clear about this, too) and trading them, which is where the difficult bit starts.

Eduardo.
 
Trading successfully is, however, more than selecting a few patterns...

Eduardo.

True. Trading successfully requires proper capitalization, psychology and above all a dynamic risk and money management method.

Alex
 
"...and above all a dynamic risk and money management method"

example?

I would recommend Van Tharp for psychology and money management. Try 'Trade your way to Financial Freedom' for its commentary on position sizing, expectancy and system development. Van's site - Trading Education.

Van is a coach of traders so his work extends into psychological development and resolving the inherent flaws and disadvantages that each of us has. These flaws prevent us from achieving our best and we can see their effects in the biases that each of us brings to our life/trading, i.e. The need to be right, lotto, representation, etc.

Van was one of the experts that appeared in Jack Schwager's 'Market Wizards'. One of the truths that both Jack and Van identified is that any trading system must fit the psychology of the person using it. Jack interviewed the great and good of the trading world and found they were all very successful because their trading fit their personal beliefs. Their success had nothing to do with the market they chose, timeframe or geographical location. It had everything with their character, trading plan and successful use of money management.

APS looks like an excellent program but if it does not fit with how you live your life or the beliefs that you bring to the market then you will not get the full benefit of it because you and it will clash. Depending on your beliefs you could go your own way and ignore its signals or create new signals because you think it should trade differently.

I attended a conference in London where the famous trader Larry Williams was a speaker. He told a story of how his laptop with all his trading setups and data was copied at another con. He wondered how the person who did that got on with all the different systems he would use from day to day. He looked each year at the ratings in the Robbins world trading cup for someone getting similar results to himself, there never was. He was the secret to his own success. I remember him saying that he could give each of the attendees daily updates detailing what to buy, how much to buy, when to sell, how much to sell and no one would follow it precisely because everyone's personal biases would get in the way.

Mr Williams colourful life story - http://en.wikipedia.org/wiki/Larry_Williams_(trader)

Personal psychology and a sound money management plan are the bedrocks of trading success. I also like the book Way of the Turtle - Way of the Turtle: The Secret Methods that Turned Ordinary People into Legendary Traders: Amazon.co.uk: Curtis Faith: Books
 
Last edited:
Top