Topics
Replies
jeex
20 Nov 2013, 19:30
Towel
We just threw the towel... This EA is a lab rat. It will only work in a lab, not in real time. So we stopped developing on this project. It's a complete and utter waste of time. Although the maker has done a nice job in analysing how to skim the momentum, this robot can never win from spread, commission and slippage.
@jeex
jeex
20 Nov 2013, 12:45
Stats
Over the past years, on the H1 timeframe (over 17,000 candles), the average candle length was 0.00221. The average body length was 0.00107 = 44% of the candle, with a standard deviation of 0.00114 (candle) to 0.00084 (body) = 22%. So simply measured a candle where the body is less then 44% - 22% = 22% of the candle can be considered as an abnormal body-candle ratio = a doji.
Or am i missing something here? Maybe also consider the length of the candle in comparison whith the rest? Use the ATR in this?
@jeex
jeex
18 Nov 2013, 22:03
Calculate
All might be nice for people, but we are talking about robots.
A robot acts on the values provided by the function Calculate (in the indicator). Now if Calculate recalculates the former tick or bar, then during the former tick or bar the robot already made a wrong decision. It does not know that the value it gets returned from Calculate is to be corrected (or not) during the next tick or bar. It can not make assumptions, it can only act or not act on data.
This indicator might work nice as an indicator for people who like to work with indicators, i do not oppose to that. It simply does not work for robots.
@jeex
jeex
17 Nov 2013, 22:11
RE: My Email
epps11 said:
I can try to help with optimization London vps with SSD maybe able to help with latency. Here's my email michaeleppsjr@yahoo.com
An optimal connection would help with a cAlgo robot like this one. Still working on it. I'll be in touch. Am working on it together with Hyperloop
@jeex
jeex
17 Nov 2013, 17:08
RE: RE: Make an indicator
/forum/cbot-support/1853?page=2#18
epps11 said:
jeex said:
My findings after filleting the jibberjabber out of the code:
With respect for the maker of the code: the method is child's play. On the M1 timeframe it calculates 4 channels: IMA, BollingerBands, Envelopes and ATR. The placed trade is with the minimal allowed StopLoss and TakeProfit and then it starts trailing. So with high volatility the engine can't keep up with the facts and you start losing money at high rate.
What would be nice if the method of the 4 channels was translated in an indicator.
Was this for Calgo or Mt4? If for Calgo maybe I can take a look at it. I know that in Mt4 the high volatility is too fast and causes the EA to lose, but I'm not sure about Calgo.
@jeex
jeex
17 Nov 2013, 15:17
Make an indicator
My findings after filleting the jibberjabber out of the code:
With respect for the maker of the code: the method is child's play. On the M1 timeframe it calculates 4 channels: IMA, BollingerBands, Envelopes and ATR. The placed trade is with the minimal allowed StopLoss and TakeProfit and then it starts trailing. So with high volatility the engine can't keep up with the facts and you start losing money at high rate.
What would be nice if the method of the 4 channels was translated in an indicator.
@jeex
jeex
16 Nov 2013, 19:49
( Updated at: 21 Dec 2023, 09:20 )
Easy but useless
As often with indicators, they show how beautiful things would have been. This indicator shows a beautiful and predictable market, but calculates this backwards:
UpSeries[index - 1] = _movingAverage3.Result[index - 1];
When things went wrong, it makes an alteration in to the curve in the past - changing history. In trading that is simply impossible. See the results in the 1H timefame. Dito for the 4H timeframe.
Charts show results of the indicator with an extra CCI mechanism. It makes some huge wins and very many small losses. Unfortunately these are not predictable by an indicator. Like with CCI you could use the MacD to limit the bad entry points, then test it on a volatile pair and it might give positive results but not steady enough.
@jeex
jeex
15 Nov 2013, 13:12
Correction
Testing proved me wrong. the value is NaN if not drawn, but somehow, it remains NaN if a line is drawn afterwards, like in this example
public override void Calculate(int index)
{
....
stepsback = index - 4;
for (int x = stepsback; x <= index; x++)
{
drawDisLine(x, value_Y);
}
}
Here the value when tested remains NaN for the first 4 steps, only the step with x=index has a numeric value, while the line is actually visible for all 5 steps.
So drawing a line afterwards, gives an ambiguous result: visible but not valid.
@jeex
jeex
22 Nov 2013, 13:11
Howto link one bot to another
A simple question on this subject: how do i link one robot to another, like we do with an indicator and robt?
I want to use one robot as library, so that the methods in that bot can be used in other robots.
@jeex