is this EA back test results any good? - page 4

New comment
 
@zzuegg - thanks for the code. I think the problem with that code that it is monolithic. Maybe it needs to be dynamic in the way of looking at the previous bars and analyzing the the statistics from it. not only that maybe it needs to look at the candle type as well - with those 2 variables combined maybe we might have a chance. I will code something up and see how it goes....
 
mbirrell:
I bet you would.... the problem is I wouldn't need help to code it if I figured it out.
Didnt think so :))
 
There are a few articles I haven't gotten the chance to read yet about statistical-occurrence of candles. Tho I like zzuegg's approach to simplifying indicators, I agree with mbirrell assessment. I just created my first approach with candles the other day and it looks very promising but only in the back-tester. I used my over-all approach to statistical-occurrence, for me this means looking for Extremes. Looking for un-usual activity on the last couple of candles was a nice place to start.
 
Sheesh! this indicator is very long winded. I have a look at how many candles go in the same direction consecutively tally the statistics up, then compare how many other formations have the same pattern but have the next candle the same as the last - express that as a percentage against the prediction being correct. That should be a good start. Other things that come into play would be trend direction change, resistance points, retracement points and the list goes on. So really we are in for a guessing game any further that 1 candle and even that has no guarantee - remember roulette... red or black?
 

Just an update - last week I went live with this EA with some modifications since I last posted. I made $6,217 for the week - which I know isn't that great but it is something.

I have more tweaking to do to get it to rake in the big bucks.

 
ah, 6k/week times gives you 300k / year- Sounds good
 
@mbirrell... 6,217 per week. I'm in the wrong line of EA. If that's real money, I'm really happy for ya. I'm still working on my double account everyweek EA. Wish me luck.
 

@ubsen - you are an intelligent guy you will get there. Those winnings did add around 127% to my account.

I am just wondering if last week was a bumper one as it hasn't traded yet this week - but then again it didn't trade every day last week either. Also folks I am breaking a few trading rules risk wise - you know what I am talking about ubzen. EAs' seem to be impossible until you come to realise the one thing that is killing your profits. Also let me say that careful money management is the way to go.

 
Not bad, but a careful moneymanagement giving you 127% a week sounds to good to be true, may i ask what percentage was at risk?
 

Very interesting discussion, indeed...

I am new in trading + new in programming (seems to be the worst combination here :-))

At least a little strong I feel in statistic (it was main focus of my study 15 years ago + some years of experience in R&D areas).

After some months of manual trading on demo account I found standard indicators (at least those built in meta trader platforms) not enough powerful and not very useful for decisions of the moment.

I had problems with time lags and also, I somehow wanted to implement my previous findings about data clusters in short time periods.

This was just the reason why I decided to try to code my own indicators and bring more data properties into consideration (tick based indicators).

I often ask myself, if I am just underestimating the use of widely-used indicators and, if you guys, who deal with coding, also develop your own and uniqe measures, data testings, indicators...

12345
New comment
Reason: