Programming Candlesticks - page 2

 
elihayun:
PM me with your e-mail address

I appreciate you sharing your EA with me but alas it doesn't help me much.

 

i have been running this in the custom indicators and it doesn't appear to be accurate identification of the patterns at least not as I understand them. It's close but incomplete. I guess this is what I get to start with and see if I can make it into something. I would appreciate help making it from an indicator into an EA that executes trades. I'm still iffy on that part. I think I can slog thru the pattern recognition problems. For example for a hammer to be most significant it needs to be at the bottom of a downtrend. this presently marks it as a down arrow regardless of it's place in a trend of any kind. I guess with some work the patterns can be correctly used. It's just a streach for me programming wise.

Files:
 

Ok I have a question about the logics....

The current candlestick EA uses this kind of test to identify that there is an uptrend...

High[n3]<High[n2] && Low[n3]<Low[n2][/PHP]

I have wanted to use this instead....

[PHP]Close[n3]<Close[n2]

yet in looking at the charts I see instances where using the high/low seems to work better than using close/close. Both sometimes fall short of identifying an uptrend.

 

why dont use both high - low and close?

 
Aaragorn:
I appreciate you sharing your EA with me but alas it doesn't help me much.

i tried helping him with it, but got snubbed. oh well.

 
phoenix:
why dont use both high - low and close?

it's a situation where using one way eliminates a percentage of recognitions and doing it the other way eliminates a different percentage of recognitions. Doing it both ways would therefore eliminate the sum of those percentages...the goal is to recognize more not less.

 

here is what i've done so far....

with the intent of perhaps making this indicator into an EA eventually I started to include a bool variable which would signal a livebear or livebull candle signal...right now it's not a really well developed code so it's not really doing much.

I reworked the logics of the bearish signals, some are using hi/lo and others are using close/close to establish the preceeding trend.

One other thing I thought of...

I expanded the number of bars available to the EA from 4 to 12. My idea was to make some of the bars available for confirming a signal...this raises some questions for me...

If I push the signal recognition back from the first two or three bars won't that defeat the purpose of it being a leading indicator and make it lag rather than lead? if so how much can I push it back and still have it be useful?

this is the age old question of how much confirmation can I tolerate and still be agressive? how much agressiveness serves a purpose and where is the point of diminishing returns?

Also this is placing arrows correctly but the text identifier is chaotic. I have no idea why it is messing up the text but doing the arrows right. Coding is so weird. anyway...I think i like my candle pattern definitions better than the previous definitions, some of them just seemed like lame approximations of the patterns.

I have yet to do the bullish signals.

Any suggestions are appreciated.

Files: