Automatic Pattern Search

[mballagan]

APS Version 8

Any alternatives out there? Have you checked?

I bought APS last year version 4.98 and spent a few months last year playing around with it but didnt tap its true potential.

I can upgrade to version 8.0 but its an extra $500 and although retail sales will stop I will get 2 years email support and a 5 year license.

Would you say version 8 is a significant improvement and has boosted your trading?

Also they are excluding the p-indicator from the upgrade but I understand this can be constructed using info from the books.

Anyone who wants to buy the software has to act quick as they close shop today at 5pm ET. Its currently 1pm ET!

 

[ZEPPO]

Also they are excluding the p-indicator from the upgrade but I understand this can be constructed using info from the books.

About everything in APS can be constructed from books, I guess - the theory is there.
The best way would be to look at this as an opportunity and build one's own software.
APS is not that flexible, which is where the opportunity lies; I guess I'll have to learn to program!
Also, there is no need to make the patterns as complex; say, 3 or 4 bars could be enough to make a profitable system following the theory in M.Harris books.
Anybody cares to comment on this?

Eduardo.

 

[NetTecture]

following the theory in M.Harris books

Which books ? Titles please

 

[mballagan]

About everything in APS can be constructed from books, I guess - the theory is there.
The best way would be to look at this as an opportunity and build one's own software.
APS is not that flexible, which is where the opportunity lies; I guess I'll have to learn to program!
Also, there is no need to make the patterns as complex; say, 3 or 4 bars could be enough to make a profitable system following the theory in M.Harris books.
Anybody cares to comment on this?

Eduardo.

I think it would be tough to recreate the complex data mining engine in the APS software. I am a programmer but still would struggle - I think you misunderstand what lies under the hood.

 

[mballagan]

Which books ? Titles please

Stock Trading Techniques based on price patterns.

Profitability and Systematic Trading.

Both by Michael Harris.

 

[ZEPPO]

I think it would be tough to recreate the complex data mining engine in the APS software. I am a programmer but still would struggle - I think you misunderstand what lies under the hood.

It does not need to be as complex as APS, which is the product of more than 10 years of work.
But it is possible to build something on the same lines, simpler, with say, 3 or 4 bars per pattern, and perhaps a back testing utility.
I'm thinking about something which is not to be commercialized, of course - unlike APS; then that would be a nightmare!

Eduardo.

 

[intradaybill]

I think it would be tough to recreate the complex data mining engine in the APS software. I am a programmer but still would struggle - I think you misunderstand what lies under the hood.

You are probably correct. When I first heard about APS a few years ago I tried to program something along these lines from the information in Harris' book Stock Trading Techniques. I quickly realized that there is much more to it than the methods Harris described in his books. I talked to a few APS users and some have not even realized a few important features of the program. For example, patterns with entry at the close can be saved in System Tracking and the program will extract and calculate the conditions that must be valid on the next bar to get a signal. For example it will tell you that Close of next bar must be > X and High of next bar must be < Y, etc.

I also think there is much more to the p-Indicator than it is described in Harris' books.

 

[ZEPPO]

I also think there is much more to the p-Indicator than it is described in Harris' books.

I agree with you on that one, otherwise an institution would not have bought the program!
Trying to program APS from scratch is something I do not even dare to contemplate; for a start, M.Harris must have had a team of programmers working for him - let's remember, he offers custom work!
I don't think he does all of this by himself, so unless you want to build a business by renting a similar program on the same line as APS, forget it, you won't be able to do it on your own.
But something simpler, like the search and scan feature, should not be beyond the bounds of a competent programmer.
Then it is just question of building your system and monitoring the signals - I can do that by using my MetaStock scanning facility, for example, in addition to APS system tracker.

Eduardo.

 

[trader3cnd]

Michael Harris is light years ahead of the rest of the pack. Just read any of his online articles or published books.

 

[ZEPPO]

Michael Harris is light years ahead of the rest of the pack. Just read any of his online articles or published books.

I have - which is why it is useless to try to copy APS, unless you raise a team of programers; but something simpler should be possible.
After all, you don't need that many patterns to trade - some of my searches gave me 300+ patterns, there is no way I'll be able to trade them all!
With 20 or so I have enough for a system.

Eduardo.

 

[ronblack]

I have - which is why it is useless to try to copy APS, unless you raise a team of programers; but something simpler should be possible.
After all, you don't need that many patterns to trade - some of my searches gave me 300+ patterns, there is no way I'll be able to trade them all!
With 20 or so I have enough for a system.

Eduardo.

I see that the APS code is for sale. I think that a few things can greatly improve the value of this product and make a lots of money for the code buyer:

1 - Multi-core support to increase search speed significantly
2 - Basic charting and improved backtesitng results, like Sharpe ratio calculations.
3 - Enabling the scan mode in real-time
4 - A basic API for the scan mode signals
5 - ATR stops in search and scan modes
6 - Easy way to get data from trading programs like Tradestation, eSignal, Metastock, etc.

 

[intradaybill]

I see that the APS code is for sale. I think that a few things can greatly improve the value of this product and make a lots of money for the code buyer:

1 - Multi-core support to increase search speed significantly
2 - Basic charting and improved backtesitng results, like Sharpe ratio calculations.
3 - Enabling the scan mode in real-time
4 - A basic API for the scan mode signals
5 - ATR stops in search and scan modes
6 - Easy way to get data from trading programs like Tradestation, eSignal, Metastock, etc.

I would add

7. TS and NInja add-on

 

[rajatheroyal]

hi guys anybody able to design a intraday scalping system on 5 minute charts with this APS?

 

[ZEPPO]

hi guys anybody able to design a intraday scalping system on 5 minute charts with this APS?

I think that if you knew the theory behind APS you wouldn't ask this question!
This is about building systematic trading systems, scalping is pretty much discretionary.
I wouldn't use APS for that - wouldn't have time, anyway, to scan and backtest, prepare the "P-indicator", etc..., takes time.
Hope this helps.
Eduardo.

 

[rajatheroyal]

I think that if you knew the theory behind APS you wouldn't ask this question!

I know the theory sir.

This is about building systematic trading systems, scalping is pretty much discretionary.

Scalping not necessarily be discretionary.There are system scalpers who make lot of money scalping the market.
Infact scalping is the most consistent method of all done systematically.

 

[trader3cnd]

You could use APS for scalping but it is difficult to filter out the noise patterns in fast bars like 5 minute charts. So I think I will agree with Eduardo as far as 5 minute charts although for slower bars like hourly charts it is possible to find good patterns.

I noticed today Michael Harris has done some quite revealing analysis about backtesting: Hmmm...very interesting...

Backtesting flaws in popular programs

 

[ZEPPO]

I know the theory sir.



Scalping not necessarily be discretionary.There are system scalpers who make lot of money scalping the market.
Infact scalping is the most consistent method of all done systematically.

And how many of them use APS?
APS wasn't build for scalping, it is too slow for a start. You ned 7000+ trades per pattern just to build a system; I'd go mad doing something like that!
Why go hard on yourself? Use hourly data, as suggested, though I would recommend daily data - but that's me.
I am not a scalper, but everything I heard about it is that it is a pretty judgemental style of trading, where gut feeling is the trigger.
Then again maybe I am wrong, as stated, I am not a scalper.

Eduardo.

 

[rajatheroyal]

Thanks Zeppo i got my answer.

 

[ZEPPO]

Thanks Zeppo i got my answer.

 

[intradaybill]

You could use APS for scalping but it is difficult to filter out the noise patterns in fast bars like 5 minute charts. So I think I will agree with Eduardo as far as 5 minute charts although for slower bars like hourly charts it is possible to find good patterns.

I noticed today Michael Harris has done some quite revealing analysis about backtesting: Hmmm...very interesting...

Backtesting flaws in popular programs

Interesting paper. I hope he posts the full version...Does anyone have any idea about which programs he is talking about?