Automatic Pattern Search

[mballagan]

There are roughly 10 x 4 x 5 x 6 = 1200 4H bars in that period.

This is a number that is very small for statistical analysis purposes. You need to consider a time span where there is at least one primary uptrend and a primary downtrend and extract your patterns from there. In this way you cover both longs and shorts.

I hope the 10K was virtual money because a 90% loss in 10 days/4H bars is a clear indication of overtrading and non-existent risk management. Basically, in 60 bars you lost 90% of your account. Based on your pip stops I will assume that each trade was closed in maybe 2 to 3 bars minimum. I will also assume that you stayed flat for at least 1 bar so this sums up to about 4 bars.

Summary: I estimate that in about 15 trades or less you lost 90% of your account. This is not related to trading technique but to risk management. Your loss shouldn't have exceeded 15 - 20% max drop from high mark.

I risk 0.5% of my account on each trade and in your case that would account to maybe a 10% drop and you would have plenty of time to change/modify system and maybe end up with making a profit.

Risk management is more important that trading system. System may be good but in the process of experiencing a drawdown and if one is overtrading often it is given no chance of recovering.

The period was 02.01.2008- 18.09.2008 and contained 1105 4HR bars (I got the date period wrong in my last post).

Looking at a weekly chart of EURAUD it had been trading in a range for the last 3 years I would have to go back to 2004 to catch a significant downtrend. It has moved very suddenly up recently though most probably as a result of huge volatility recently in the financial markets.

The 10K was virtual money and I was using excessive leverage in order to boost returns for the period of the competition which lasts 3 months but it has worked against me in this case as the trading is fully automated and there is no option to change the parameters of a system once the contest has started. I will bear this in mind for real trading and set a 1% risk limit.

Looking at the trade hsitory of the system, under Account History:

Participant warwickquant - Automated Trading Championship 2008

It looks like some trades were held for 2 or 3 bars but many were opened and closed on the same bar. There were 33 trades with 22 resulting in a loss.

I will extract patterns on a greater timespan, put in proper risk management and forward test the system.

 

[intradaybill]

The 10K was virtual money and I was using excessive leverage in order to boost returns for the period of the competition ...

I think you shouldn't put the blame on APS but on your wrong use of the program, as well as your bad risk management:

Data FAQ

You sound like several other posters in these and other threads who constantly blame indicators or chart patterns for their losses. Most of these traders start with a small account and huge leverage and they hope they will make it big in a few months...

 

[jiacopel]

Another question:

Has anyone ever found success in using range bars are volume bars with APS rather than time based OHLC?

I can't run the tests myself because my charting platform doesn't offer such data, but I was curious.

 

[equtrader]

jiacopel...the problem is you cannot define exits in APS based on range or volume bars. The program operates based on OHLC only:

Tradingpatterns.com Michael Harris article

mballagan...what you have done has nothing to do with trading but it's pure gambling. Save the wrong application of APS. Maybe you should not trade a real money account before you study risk management seriously and understand that even the best of systems can fail under conditions of high leverage.

 

[mballagan]

I think you shouldn't put the blame on APS but on your wrong use of the program, as well as your bad risk management:

Data FAQ

You sound like several other posters in these and other threads who constantly blame indicators or chart patterns for their losses. Most of these traders start with a small account and huge leverage and they hope they will make it big in a few months...

Thanks for the link - useful info on number of bars required for statistically significant patterns. Will go back to the drawing board and test systems using proper risk management e.g 1% of trading capital on each trade.

 

[mballagan]

jiacopel...the problem is you cannot define exits in APS based on range or volume bars. The program operates based on OHLC only:

Tradingpatterns.com Michael Harris article

mballagan...what you have done has nothing to do with trading but it's pure gambling. Save the wrong application of APS. Maybe you should not trade a real money account before you study risk management seriously and understand that even the best of systems can fail under conditions of high leverage.

The competition was a virtual account of USD10,000 so I havent lost any real money this time but I am aiming to trade a live account next year. You are right its gambling and I learned my lesson. I will see if I can create profitable systems using APS patterns with data sampled properly according to the timeframe being traded and with proper risk management. The objective is to build systems for fully automated forex trading on the Metatrader platform.

 

[trader3cnd]

Virtual trading has nothing to do with real money trading. No real counterparty involve in the virtual game. You are wasting your time. Better try mini account with real money.

 

[ozzymandius]

To any and all using APS:

A couple of questions.

I'm evaluating the demo right now and it appears to generate rules going 3-5 bars back. In the demo version only partial code is visible, but it looks like 3-5 bars. e.g. Buy when Close is less than Close of two days ago, and Low of one day ago higher than Open of five days ago, etc.

(1) Generally how many bars back do the actually patterns go? Is it always five bars and less? Or does some model logic go back even further, for longer term patterns?

(2) For anyone using APS for a year or more, have the results remained consistent for you in real world trading? Or has your portfolio profit/loss swung wildly?

(3) For experienced users - has the program really paid for itself, like some claim?

I'm currently not finding great results testing the demo on Forex daily and 15 min data over two years (2004-2005) and then backtesting the patterns on year 2006. But I continue to investigate.

Thanks for any input from experienced users.

 

[intradaybill]

Hi,

I do not consider myself an experienced APS user but I will offer some comments to your questions.

(1) Generally how many bars back do the actually patterns go? Is it always five bars and less? Or does some model logic go back even further, for longer term patterns?

I believe they go back 6 bars max. If you try the patterns with the delayed entry that can increase significantly.

(2) For anyone using APS for a year or more, have the results remained consistent for you in real world trading? Or has your portfolio profit/loss swung wildly?

It depends what you mean by results. This program only finds patterns that meet some criteria based on historical performance etc. I use those as basic elements to develop more complex trading startegies that include indicators and a banch of heuristics. If you are asking about the performance of individual patterns I think the examples APS developers display on their website about QQQQ are impressive but I don't think they can be easily generalized to other markets:

Tradingpatterns.com Michael Harris article

(I'm currently not finding great results testing the demo on Forex daily and 15 min data over two years (2004-2005) and then backtesting the patterns on year 2006. But I continue to investigate.

Try 1H and 4H data. I am suspicious of forward testing though because it has cost me a lot. A couple of years ago I rejected a system I developed based on data from 1996 to 2003 because it failed during 2004 testing. Then it exhibited a stellar performance during 2005 and 2006 but I was not trading it.

There is no theoretical justification for forward testing unless you are looking for systems that make money each and every year. In that case I recommend investing in T-Bills. I prefer a return of 200% during one year and then giving back 25% of the profits next year and then making again 100% and then giving back 30% of that and so on...At the end of the day you can become very rich by realizing that giving back profit is part of the game.

 

[ozzymandius]

Intradaybill - thanks for the input. I appreciate the feedback.

 

[intradaybill]

Intradaybill - thanks for the input. I appreciate the feedback.

You are welcome. Also, make sure the values you specify for profit target and stop loss are the correct ones according to your Trade Exit selection (inc or %). I use inc for forex because my exists always correspond to pips. This link may be of some help:

Using APS FAQ

 

[market_student]

Hello all:
I recently got APS and am impressed with the "concept", but having spent so many fruitless hours searching for ANY meaningful patterns in Intraday Forex data, I have come up with ***NOTHING***. It just seems that there ought to be a better, logical, step by step development process for this tool. My experience on using this is that you have to "Guess" what the right levels of Stop and Profit Target need to be.... 100/100, 50/50, 67/67 pip etc.. there are unlimited combos and with the intraday data (50K+ bars), it becomes a crap shoot whether after an hour search the program might turn up with some patterns or the dreaded "No Patterns Found" (basically, you just wasted 1 hour of your precious research time).

I have tried 5,10,30,60 min data and have come up with nothing so far. By no means am I complaining about the capability of the software, but perhaps my ability to use it properly.

So, having said that, I am looking for some practical tips from some experienced users here who can guide us to a better, logical and productive approach. I am willing to put in the effort, but get frustrated when I waste valuable time.

Any help would be very much appreciated.

Best

 

[ZEPPO]

Hello all:
I recently got APS and am impressed with the "concept", but having spent so many fruitless hours searching for ANY meaningful patterns in Intraday Forex data, I have come up with ***NOTHING***. It just seems that there ought to be a better, logical, step by step development process for this tool. My experience on using this is that you have to "Guess" what the right levels of Stop and Profit Target need to be.... 100/100, 50/50, 67/67 pip etc.. there are unlimited combos and with the intraday data (50K+ bars), it becomes a crap shoot whether after an hour search the program might turn up with some patterns or the dreaded "No Patterns Found" (basically, you just wasted 1 hour of your precious research time).

I have tried 5,10,30,60 min data and have come up with nothing so far. By no means am I complaining about the capability of the software, but perhaps my ability to use it properly.

So, having said that, I am looking for some practical tips from some experienced users here who can guide us to a better, logical and productive approach. I am willing to put in the effort, but get frustrated when I waste valuable time.

Any help would be very much appreciated.

Best

The way I do it is by copying the data in a spreadsheet (I use the daily range from an indicator in my MetaStock software) and then I arrange it from the smaller to the higher, call up a bar chart (from the spreadsheet) and select the range with contain the most data.
For example, to trade the FTSE100 I select that way the 4T4S range, that is,4% target and 4% stop loss.
I suggest that you read Michael Harris literature as he covers this topic in his books and his articles.
Hope this helps.

Eduardo.

 

[intradaybill]

...or the dreaded "No Patterns Found" (basically, you just wasted 1 hour of your precious research time).

The computer can do the search while you are doing something else. It is not a good idea to sit and watch the progress bars because as you said if nothing comes up you are just wasting your precious research time.

I run the program during the weekends when I am not trading. I create and save a T/S file with the target/stop pairs I determine with a method similar to Eduardo's. Actually I plot the 14-day ATR and I determine the max and min values. Then I make sure my target and stop pairs are withing 2 x ATR, 2.5 x TR, 3 x ATR, etc. so trades don't get stopped out during the first bar.

More importantly your target/stop should have the correct values and also select the extended search option to increase the possibilities. I suppose you have reviewed the FAQ webpage in the product website:

Using APS FAQ

 

[trader3cnd]

It just seems that there ought to be a better, logical, step by step development process for this tool.

Read the manual that comes with the program. FYI, using the 30 minute EURUSD data that came with the demo the program found 72 patterns after doing a search with the following parameters:

- target = stop = 50 pips (You MUST input this as 0.0050 in the T/S file)
- Trade Parameters: inc, Open, Delay Off
- % Prof for long > 66
- % Prof for short > 66
- Trades > 30
- Max cons. losers < 4
- Search Range = 500 bars
- Extended search

It took less than 45 minutes to run the search and display the results.

You must be doing something fundamentally wrong. This program can find tons of patterns. First step read the manual. Then try to replicate the results above. If you cannot get those 72 patterns you are doing something wrong, probably a basic mistake.

 

[market_student]

Read the manual that comes with the program. FYI, using the 30 minute EURUSD data that came with the demo the program found 72 patterns after doing a search with the following parameters:

- target = stop = 50 pips (You MUST input this as 0.0050 in the T/S file)

Thank you.... Found the issue.
You were spot on... it was a fundamental BASIC issue.

Had the wrong format in the T/S file...

Many Thanx

 

[alexander]

Thank you.... Found the issue.
You were spot on... it was a fundamental BASIC issue.

Had the wrong format in the T/S file...

Many Thanx

market_student, before you input the values in a T/S file you should open your data file with an editor and look at the values. In the case of point target and stop, the exit prices in APS are calculated as follows:

Profit target price = entry price +/- T
Stop loss price = entry price -/+ S

depending on whether the pattern is long or short.

If your EURUSD entry price is in the usual format, like for example 1.2754, and your target is 30 pips then it should be inputted as 0.0030 in the T/S file. Then, for long patterns

Profit target price = 1.2754 + 0.0030 = 1.2784
Stop loss price = 1.2754 - 0.0030 = 1.2724

In the case there are 5 significant digits, like in 1.27544, the value for 30 pips should be still 0.0030.

I hope this helps.

Alex

 

[intradaybill]

In the case there are 5 significant digits, like in 1.27544, the value for 30 pips should be still 0.0030.

I hope this helps.

Alex

Alex thanks a lot, this is very useful information.

 

[Albert Carey]

You are welcome. Also, make sure the values you specify for profit target and stop loss are the correct ones according to your Trade Exit selection (inc or %). I use inc for forex because my exists always correspond to pips. This link may be of some help:

Using APS FAQ

Well, % exits also correspond to pips but it is not a constant amount. I decided to use % exist because they account better for the increased volatility as the value of a currency pair makes new highs. for example when EURUSD reached the 1.5000 level.

I would like to hear from others what they think about constant pip vs. percent stops in the case of forex.

 

[fobos]

I'm currently investigating the possibility of using a genetic algorithm with the operators AND, OR to automatically design systems for me using as input the APS patterns.
Ron

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.