Automatic Pattern Search

[qwertzuiop]

Thanks Alex I am looking at the demo version currently and will probably buy the software next week.

Regards

Manmohan

Did anyone of you try out VectorBull ? :innocent:
Trading one step ahead!

They find new patterns, that (yet) nobody knows. :idea:
They also have back test functions (Stocks version) to verify trading results. :clap:

 

[alexander]

Did anyone of you try out VectorBull ? :innocent:
Trading one step ahead!

They find new patterns, that (yet) nobody knows. :idea:
They also have back test functions (Stocks version) to verify trading results. :clap:

Hi,

Does vectorbull output code for Wealth-lab or Tradestation? From the demo on the website it appears it generates entry and exit orders based on a proprietary algorithm that is not disclosed. Maybe I missed that function.

Alex

 

[qwertzuiop]

Hi Alex,

no, VectorBull doesn't output Easy Language or similar stuff. It simply outputs trading signals, like a neural net.

I use it to trade a portfolio of DOW 30 stocks.

Regards, Frank

 

[alexander]

Hi Alex,

no, VectorBull doesn't output Easy Language or similar stuff. It simply outputs trading signals, like a neural net.

I use it to trade a portfolio of DOW 30 stocks.

Regards, Frank

Well, I do not feel comfortable trading a black box even if its performance is very good. But this is personal preference based on a desire to have control and responsibility over own actions. I do not think any serious trader would consider a black box for many reasons. Good luck to you.

Alex

 

[trader3cnd]

Anyone with a used copy of APS for sale please email or send PM.

Martin

 

[intradaybill]

Anyone with a used copy of APS for sale please email or send PM.

Martin

I was also looking for a used (legal) copy of this program but then someone told me in a forum that buyers of used software risk losing everything because malicious sellers can hide a backdoor in the code and steal all your trading and bank codes.

It's pretty scary isn't it?

Bill

 

[byzantium]

APS Sale--Expires Monday, April 14

I was also looking for a used (legal) copy of this program but then someone told me in a forum that buyers of used software risk losing everything because malicious sellers can hide a backdoor in the code and steal all your trading and bank codes.

It's pretty scary isn't it?

Bill

APS Automatic Pattern Search - License puchase

 

[alexander]

I was also looking for a used (legal) copy of this program but then someone told me in a forum that buyers of used software risk losing everything because malicious sellers can hide a backdoor in the code and steal all your trading and bank codes.

It's pretty scary isn't it?

Bill

It's pretty scare these days even with legal software. Look for instance at adobe, java and flash. They are loading a bunch of tasks slowing down the computer, keep constantly checking for updates, etc.

IMO, the best solution is to have a PC only for trading without any personal files and data in it and to change passwords quite often.

Alex

 

[intradaybill]

APS Automatic Pattern Search - License puchase

After doing some market research it appears APS is one of two or maybe three white box pattern discovery programs.

I guess the special offer web page is no longer there but I have decided to buy this software.

Any suggestions about proper use/application of results will be appreciated.

I plan to use it for US stock and FX short term position pattern trading.

Bill

 

[Chorlton]

Hello All,

With APS my understanding is that the software finds "new" patterns based on ones risk/reward parameters, etc. However, if I understand the software correctly, this can only be achieved on one secuity at a time. Consequently, I'd be interested in knowing how reliable a new pattern is when traded over a PORTFOLIO of stocks?

All comments welcome,

Chorlton

 

[ronblack]

Hello All,

With APS my understanding is that the software finds "new" patterns based on ones risk/reward parameters, etc. However, if I understand the software correctly, this can only be achieved on one secuity at a time. Consequently, I'd be interested in knowing how reliable a new pattern is when traded over a PORTFOLIO of stocks?

All comments welcome,

Chorlton

Hello Chorlton,

APS patterns are security specific. You can find patterns for a portfolio of securities using a single workspace but you are correct, each pattern in the results is always associated with a specific security.

This is something I thought about since I got APS almost two years ago. There is an answer: find patterns in a security and then backtest on a portfolio. Select only those securities that work with the specific pattern and form a new portfolio. And so on...

I mean there is no justification as to why a pattern should be profitable across the board. Remember that investors often sell one security to buy another or do sector rotation type of strategies. Also pair trading involves long/short position in similar securities.

I think APS has a lot of power and if used smartly the results can be excellent.

Ron

 

[intradaybill]

Hello Chorlton,

APS patterns are security specific. You can find patterns for a portfolio of securities using a single workspace but you are correct, each pattern in the results is always associated with a specific security.

This is something I thought about since I got APS almost two years ago. There is an answer: find patterns in a security and then backtest on a portfolio. Select only those securities that work with the specific pattern and form a new portfolio. And so on...

I mean there is no justification as to why a pattern should be profitable across the board. Remember that investors often sell one security to buy another or do sector rotation type of strategies. Also pair trading involves long/short position in similar securities.

I think APS has a lot of power and if used smartly the results can be excellent.

Ron

Ron, your strategy is smart. It makes sense too!

Bill

 

[Chorlton]

Hello Chorlton,

APS patterns are security specific. You can find patterns for a portfolio of securities using a single workspace but you are correct, each pattern in the results is always associated with a specific security.

This is something I thought about since I got APS almost two years ago. There is an answer: find patterns in a security and then backtest on a portfolio. Select only those securities that work with the specific pattern and form a new portfolio. And so on...

I mean there is no justification as to why a pattern should be profitable across the board. Remember that investors often sell one security to buy another or do sector rotation type of strategies. Also pair trading involves long/short position in similar securities.

I think APS has a lot of power and if used smartly the results can be excellent.

Ron

Hi Ron,

Thanks for taking the time to reply is such detail.

I mean there is no justification as to why a pattern should be profitable across the board.

mmm.... not sure whether i'd agree 100% with this statement. IMO There is a reasonable likelihood that a particular pattern may actually be more than just a "repeatable random" pattern. In other words I'm suggesting that the price pattern (structure) may follow a logical idea as to why it actually works and as a result has a higher likelihood of successfully repeating across a portfolio.

For me personally, I'd rather trade this type of pattern rather than simply a pattern that has proven reliable in the past yet has no real "market" logic behind why it works.

Just my view though....

PS. I like your approach to how you overcome this....

 

[ronblack]

For me personally, I'd rather trade this type of pattern rather than simply a pattern that has proven reliable in the past yet has no real "market" logic behind why it works.

In the past I also tried to determine the logic behind the patterns I used. I remember that I liked most those that looked like TA chart patterns, like double bottoms, triangles, etc.

Then, through experience I found out that most winners offered no immediate logical explanation. Now, I treat them as mathematical signals (probabilities) and I do not bother with the logic.

However, I appreciate you methodology, what counts in this business is what works for you best.

Ron

 

[Chorlton]

In the past I also tried to determine the logic behind the patterns I used. I remember that I liked most those that looked like TA chart patterns, like double bottoms, triangles, etc.

Then, through experience I found out that most winners offered no immediate logical explanation. Now, I treat them as mathematical signals (probabilities) and I do not bother with the logic.

However, I appreciate you methodology, what counts in this business is what works for you best.

Ron

Hello Ron,

An interesting view and maybe worth considering.

Based on your experience would you say that the more "logically derived" patterns (i) continue to stay profitable more so than the purely "random" ones, (ii) less profitable or (iii) about the same?

At the end of the day, I've always chosen the logically derived ones as I felt they would offer more robustness long-term but if this is not the case then maybe it is time to re-evaluate my mindset. Afterall, at the end of the day our goal is the same...to be consistantly profitable over the long-term.

 

[Albert Carey]

Hello Ron,

An interesting view and maybe worth considering.

Based on your experience would you say that the more "logically derived" patterns (i) continue to stay profitable more so than the purely "random" ones, (ii) less profitable or (iii) about the same?

At the end of the day, I've always chosen the logically derived ones as I felt they would offer more robustness long-term but if this is not the case then maybe it is time to re-evaluate my mindset. Afterall, at the end of the day our goal is the same...to be consistantly profitable over the long-term.

Hi guys. I'm a new APS user and I trade mainly forex and ETFs at this point in time.

This past wednesday after the market close I scanned my ETFs and got three long signals for SPY. Market was down and that pattern didn't seem logical. I decided to reduce risk and take a small long position. Yesterday, just one day after those long signals, the dead cat bounced so hard I just couldn't believe the timing of APS was that precise.

My scan parameters:

In T/S file : 7, 7
Percent exit, open of next day
Percent for long and short: 72%
Trades: > 31
4,000 daily bars of historical SPY data

I need about 100 such signals to retire (just kidding!)

Albert

 

[ronblack]

Hi guys. I'm a new APS user and I trade mainly forex and ETFs at this point in time.

I trade quad Qs and SPY and APS performance has been excellent. There will be drawdowns but if you stick to profit target and stop loss you have assigned you can make a lot of money.

I use the system tracking function to generate my signals. I could use an API for placing orders online but no such option is available at this point.

Ron

 

[mballagan]

I tried using APS on forex data EURUSD & GBPUSD 60 minute bars on the metatrader platform but didnt have a great deal of success. My methodology was as follows:

bool longcondition1 = Open[3] > Close[1] && Close[1] > Open[2] .........
bool longcondition2 = Close[2] > Close[3] && Close[3] > Close[1] ........
.....
.....


bool shortcondition1 = Close[1] > Open[1] && Open[1] > Close[2].........
bool shortcondition2 = High[3] > High[2] && High[2] > Close[3] .........
......
.......


if(longcondition1==true or longcondition2 == true or .........) then open long position

if(shortcondition1 ==true or shortcondition2 == true or .........) then open short position

In other words if any of the long conditions is true on the current bar then open a long position, reverse for short. I was extracting patterns on one year of data then testing out of sample on 3 months data. I did this for data from 2000 to 2008 rolling the window forwards by 3 months each time. The results were mixed, some periods showing good performance and others not.

I typically had 50 pip profit target/stop loss.

I read with interest the comments of Ron and Albert. How are you guys generating your trading signals on the etfs SPY and Qs? So if for example your run a pattern search on SPY such as the following:

roughly 4000 bars daily SPY (1993 - 2008)
5% profit/stop
% profitable = 66%
number of trades > 20

This generated 25 long patterns. How would I then use this to trade going forward? I am not sure how to update data on a daily basis using APS and generate trading signals.

I can generate the wealth lab code and then trade using a similar approach to that I used in Metatrader in Wealth-Lab ie if any of the long patterns is true on a particular bar then open a long position (but only allow a single position at a time so dont open if position already open - this can be adjusted to allow a maximum number of open positions).

Its very encouraging that you both get good results trading ETFs but do you get a high of enough frequency of signals trading just two ETFs - Q's and SPY on daily data? Or is it a good idea to increase the basket of ETFs?

Regards

M Ballagan

I trade quad Qs and SPY and APS performance has been excellent. There will be drawdowns but if you stick to profit target and stop loss you have assigned you can make a lot of money.

I use the system tracking function to generate my signals. I could use an API for placing orders online but no such option is available at this point.

Ron

 

[ronblack]

bool longcondition1 = Open[3] > Close[1] && Close[1] > Open[2] .........
bool longcondition2 = Close[2] > Close[3] && Close[3] > Close[1] ........

if(longcondition1==true or longcondition2 == true or .........) then open long position

The OR method is too simplistic and rarely works. I cannot go into details and reveal my own method but you need to use confirmation and that can be accomplished with AND. For example:

if((longcondition1==true AND longcondition2 == true) or .........) then open long position

This may change your way of looking at APS and patterns. I'm sorry I can't reveal more details but you should be getting the idea now and good luck to you.

Ron

 

[alexander]

Confirming signals is essential and using AND is a way of doing it. It reduces the number of available signals but it increases success rate significantly. I use the same method.

I agree with Ron that using just OR is a very simple approach. One must go beyond simple and obvious ways of doing things. There are too many smart kids out there with powerful computers and fancy algorithms ready to take your money.

Alex