| Welcome Guest, please sign in to participate in a discussion. | Search | Active Topics | |
Registered User
Joined: 3/15/2006
Posts: 15
|
I've created two percent true customs indicators. I use one to scan for 4 bullish candlestick patterns, and the other for 4 bearish candlestick patterns. Given the fact that each one has 4 individual boolean formulas, that makes for one very long formula for each.
I thought that might work, but now I am finding that it is taking extremely long to do real-time sorts. Moreover, even switching between different time frames on the charts or calling up different charts is extremely slow. Is that because the charts contain indicators that use long formulas requiring many calculations?
If this is the case, what is my solution? Breaking each indicator down to numerous ones wouldn't do the trick because all of them would still be on the same chart. And then I would have to sort for each pattern which would take long in itself. Is there any way to effectively do real-time sorts for candlestick patterns in a quick manner that doesn't tie down the system? Because right now i'm not being able to trade- it's just too slow to scan and sort.
|
|
Worden Trainer
Joined: 10/1/2004
Posts: 18,819
|
Post your formulas for me. I might be able to clean them up a bit which should speed things a tad.
- Craig
Here to Help!
|
|
Registered User
Joined: 3/15/2006
Posts: 15
|
If decreasing formula length makes the difference, then there is one definite place I could work on. Basically, I am putting conditions to quantify a long candles. To do this I've used ur "long hand" suggestion below. These long candle qualifications are the bulk of my formulas. I know you had suggested a shorter one based on trading range: AVGH13-AVGL13. But a long candle in my opinion is a condition that relates to the previous candle bodies, whithout taking the shadows into account. That's why I was using this long hand. But if there is a way to compare Candle body length with a previous range with a much shorter formula that would be great.
100 * (ABS(C - O) / ((ABS(C - O) + ABS(C1 - O1) + ABS(C2 - O2) + ABS(C3 - O3) + ABS(C4 - O4) + ABS(C5 - O5) + ABS(C6 - O6) + ABS(C7 - O7) + ABS(C8 - O8) + ABS(C9 - O9) + ABS(C10 - O10) + ABS(C11 - O11) + ABS(C12 - O12)) / 13)) > 200
|
|
Worden Trainer
Joined: 10/1/2004
Posts: 18,819
|
There is not.
For kicks and giggles you should try my suggested formula. You will find if you sort by yours then sort by mine you will see little or no difference in the order the stocks were ranked.
- Craig
Here to Help!
|
|
Registered User
Joined: 3/15/2006
Posts: 15
|
Well i'm definitely gonna try urs if it makes the scans faster. Thanks a lot for ur help again.
I might me asking frequent questions but that's just because I'm still setting up my whole system. Hopefully when things are under way i'll take up a lot less of ur time :)
|
|
Worden Trainer
Joined: 10/1/2004
Posts: 18,819
|
I am here to answer questions. NEVER hesitate to ask... it is my job and pleasure to answer.
- Craig
Here to Help!
|
|
|
Guest-1 |