Requests & Ideas - page 207
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Hi Mladen,
Sorry to trouble you again... It's me & my investigation of statistical properties again = Danggg. Is it possible if you could code a Trimmed Absolute Deviation from the Mean & a Trimmed True Range using the below attached "SQ 2.01" indicator format? I know that the trimmed method isnt efficient as compared to the other methods but at least it makes for less computational power and easier coding. The exact implementation format (for instance in % trim format or instead in data sample size number trim format) etc etc as you see fit.
Thanks thanks!
Truncated mean - Wikipedia, the free encyclopedia
Yours sincerely,
WinterskyA question : if I replace the mean in the absolute deviation formula with truncated mean - is that what you are looking after? I mean the absolute deviation formula as described here : Absolute deviation - Wikipedia, the free encyclopedia
A question : if I replace the mean in the absolute deviation formula with truncated mean - is that what you are looking after? I mean the absolute deviation formula as described here : Absolute deviation - Wikipedia, the free encyclopedia
Hi Mladen,
No, im not looking for that. Sorry i probably didn't put myself clearly. I meant to look for the trimmed absolute deviation (AD), for which the AD is just made up of the basic simple mean (no trimming of mean here).
In further explanation, i've ranked the below on effects of data outlier effect/kurtosis effect, for which i'm looking for the 4th one. Hope this makes things clearer:
1st: Standard Deviation
2nd: Absolute Deviation
3rd: Absolute Deviation of Trimmed Mean
4th: Trimmed Absolute Deviation of Mean
5th: Trimmed Absolute Deviation of Trimmed Mean
Although in reality, the 3rd version and the 4th version above are quite similar using trimming once, but the fact that there is a location estimator (mean) and a scale estimator (deviation) combined in each of them with the final output being scale estimation, makes the 4rd version more single-outlier resistant than the 3rd version, though at the cost of increased overall error in variance level estimation. The usual Type 1 and Type 2 error statistical tradeoffs showing it's presence here that is.
Thanks for clarifying Mladen
Yours sincerely,
Wintersky
Raymond You are correct : in mtf mode it did not work correctly even though it worked correctly in regular mode. Here is the correct one (see the example of the 2 compared with same parameters - 1 hour time frame both)
Dear Mladen, sorry to have to trouble you again.
1. I would like to have a similar indicator of the one you did with Blau TSI candles, but this time for the Blau TVI indicator as attached. I would like to have the option of coloring on zero cross, slope change, and signal cross as well.
2. Regarding the Blau TVI indicator, can you also update it so that it would also alert based on the same 3 options; zero cross, slope change, and signal cross.
Thank you very much Mladen. Again, sorry for the trouble.
Re:tvi tsi
Dear Mladen, sorry to have to trouble you again.
1. I would like to have a similar indicator of the one you did with Blau TSI candles, but this time for the Blau TVI indicator as attached. I would like to have the option of coloring on zero cross, slope change, and signal cross as well.
2. Regarding the Blau TVI indicator, can you also update it so that it would also alert based on the same 3 options; zero cross, slope change, and signal cross.
Thank you very much Mladen. Again, sorry for the trouble.Raymond did the Blau TVI candles and added the multi alerts to both, alertsOnColor0(zero cross), alertsOnColor1(slope) , and alertsOnColor2(signal cross).
Raymond did the Blau TVI candles and added the multi alerts to both, alertsOnColor0(zero cross), alertsOnColor1(slope) , and alertsOnColor2(signal cross).
Thank you very much Mr. Tools!!! There is still the TVI indi request I made to have those same 3 options as alerts, would you be so kind to assist on that? Thank you!!!
Thank you very much Mr. Tools!!! There is still the TVI indi request I made to have those same 3 options as alerts, would you be so kind to assist on that? Thank you!!!
The Tvi I posted also has the 3 different alerts as well, also something I forgot to mention you can use any option of alert regardless of what color option you use, if you use color = 2 you can use color 2 alert along with color 1 alert, etc.
The Tvi I posted also has the 3 different alerts as well, also something I forgot to mention you can use any option of alert regardless of what color option you use, if you use color = 2 you can use color 2 alert along with color 1 alert, etc.
Yes, currently have it on the chart and realized that this was what you did. Ingenious work Mr. Tools! I was just thinking that if we could have a stand-alone TVI that has the same properties without relying on candle alerts (in case of whatever reason the TVI candle indicator isn't used), it would be better.
Hi Mladen,
No, im not looking for that. Sorry i probably didn't put myself clearly. I meant to look for the trimmed absolute deviation (AD), for which the AD is just made up of the basic simple mean (no trimming of mean here).
In further explanation, i've ranked the below on effects of data outlier effect/kurtosis effect, for which i'm looking for the 4th one. Hope this makes things clearer:
1st: Standard Deviation
2nd: Absolute Deviation
3rd: Absolute Deviation of Trimmed Mean
4th: Trimmed Absolute Deviation of Mean
5th: Trimmed Absolute Deviation of Trimmed Mean
Although in reality, the 3rd version and the 4th version above are quite similar using trimming once, but the fact that there is a location estimator (mean) and a scale estimator (deviation) combined in each of them with the final output being scale estimation, makes the 4rd version more single-outlier resistant than the 3rd version, though at the cost of increased overall error in variance level estimation. The usual Type 1 and Type 2 error statistical tradeoffs showing it's presence here that is.
Thanks for clarifying Mladen
Yours sincerely,
WinterskyWintersky
This is maybe a a possible solution for the last 3 casesTry it out
regards
I need an automatic EA developed from the attached template with indicators. Basically I want a buy or sell order to be placed on the next bar following a bar where the superfast and the chuvashov indicators gives an alert together (do a buy order when both indicators signal buy on the previous bar or do a sell when both indicators signal sell on the previous bar). Once an order is place, the order should be closed when both signals occur again in the opposite direction. This is made specially for the M5 chart and no need to code any SL or that kind of stuff, all I need is fo the EA to reapidly open and close trades on the double signal. Also, when a trade is clodes on a double signal then another should automatically open as its also a double signal for a trade to start. Please see the linked zip with the indicators, template and also image showing 2 trade examples.
The files are attached
I really appreciate the help.
Thanks.
Wintersky This is maybe a a possible solution for the last 3 cases
Try it out
regardsMladen,
It's awesome! There's no error and the indicator works as it should in cases of discard deviation or discard mean settings from a visual checking of the output higher and lower values (their statistical errors too) that should theoretically happen in mean-reverting and non-normal situations. I was scared that the calculation codings would be a data hog and slow down the comp but it proved to be less computationally intensive than i thought even. It works great!!! You're making me greedy to see the final results when you'll make it with the stationary characteristic with Trimmed True Range in the final product.
& yes, indeed more experimentation indeed will be needed to find out the right settings to apply, as it's utility is still primarily "geared towards the extreme spectrum somewhat". & not forgetting the issue of the influence of finite samples on estimators. Crossing my fingers in great dizzy excitement. Thanks alot alot alot Mladen. Hehehhehee
Yours sincerely,
Wintersky
Ps: just in the rare case you might be interested in the research side of this indicator, i posted the most recent published paper with the efficiency ratings on pg 14 for a double trim, though personally i partially doubt the accuracy figures here even with extensive testing. Some of the older papers i've have seen, erroneously report the efficiency to be at 100% for Gaussian situations, which has now been revised downwards.