Discussion of article "Create Your Own Trading Robot in 6 Steps!" - page 5
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
Another mistake (yours):
I'll see how I can guard against this in the Library.
I got to the tester (build 619, win xp)
1) Settings tab: date selection by calendar. On the button to select, the calendar appears and disappears, I do not have time to select.
2) optimised the Expert Advisor:
period all history, only euro/dollar hourly (in the review also only eur/dollar), production delay, OHLC on M1, 3 thousand dollars, 1:100, genetic Balance+Sharpe, no forward;
parameters SL 20-50 step5, TP 30-50 step 5, PFast 13-17 step 1, MFast Simple - Linear, PSlow 21-24 step 1, MSlowt Simple - Linear.
all this was done in 25 minutes, 11200 runs.
Optimisation result ranked max result,
I try to run a single test with the right button, but I got an error of Expert Advisor initialisation (although the Expert Advisor itself is attached to the chart normally with these parameters).
log messages see text file
2012.03.28 14:27:18 Core 1 1971.02.04 00:00:01 MA_Cross::CreateFastMA: error of initialisation of fast MA object
I randomly poked a few other lines as a result of optimisation, similar error.
It's the turn of the tester (build 619, win xp)
It is roughly clear. There are problems (with Standard Library indicators) with testing on the "All History" period.
We will sort it out. And you are on another period for now.
It is roughly clear. There are problems (for indicators of the Standard Library) with testing on the "All History" period.
We will sort it out. For now, you are on another period.
1) It may be useful: I have experimentally found out that regardless of the settings of fast MA or slow MA, the single test is triggered at the value of the test start from 18 January 1972 inclusive.
2) By the way, does the calendar selection list button work like this for everyone or is it just me who has a calendar panel that just appears and disappears immediately?
3) and one more thing, manually set those parameters of better optimisation (SL35 TP 50, 17, 24 both smoothed), set from 18 01 1972.
As a result, the graph was drawn only from 1995, and by the end of 2006 there was 97$ left of 3k. (I ran it several times).
Although at optimisation the profit is more than 600$. How so?
4) here is more: the period remained selected from 18 01 1972, in the results of optimisation on the same best clicked to build a single, it worked, similar to the manual run (also with a similar unprofitable result), but also noticed that the machine set the periods fastMA and slowMA correctly (17 and 24) and the method both reset to simple although before it was smoothed and in the results of optimisation of this line also smoothed
attached test result file testergraph.report.2012.03.29.csv with this name is not attached for some reason...
1. May be useful: I found out experimentally, regardless of fast MA or slow MA settings, single testing is triggered at test start value from 18 January 1972 inclusive.
2. By the way, does the calendar selection list button work like that for everyone or is it only me who has the calendar panel that just appears and immediately disappears?
3. As a result, the chart was drawn only from 1995.
4., and by the end of 2006 from 3 thousand remained 97 $. (several times ran) Although the optimisation profit of more than $ 600. How so?
1. It is clear by the date. When creating the indicator class, it tries to fill the buffer with previous values, and when running from the "beginning of time" it has no chance. (Thanks for your concern.)
2. I have the calendar working fine.
3. In the terminal settings "Service-Settings-Graphics" the parameter "Max.bars in the window" is a bit small.
4. If the difference between optimisation and testing remains, request to Service Desk.
I found out that the data is from June 93, so I ran the same test from that date:
1) indeed, and the time of the test has increased significantly (compared to the full history), and most importantly, I got completely different results, which are confirmed in a single test.
2) as I approached the test for the first time, sorry, reflections:
firstly, I was very surprised that the best result was $1700, with an initial 3k, whereas about 10 years ago I accidentally wandered into a free forex course, the only thing I remembered was the average.
Secondly, do I understand correctly that the purpose of testing is the following: now you can look at the chart, select the period where the balance goes up, and try to find out why this particular period, such an algorithm worked (for example, here, clearly the period from April 09 to December 11 - growth), then chemise some more algorithm, test, again select the conditions of a satisfactory result, etc. and accumulate a collection of "rather winning" signals?
3) in the tester . on the page"Optimisation results". is there any way to copy a line to the clipboard? If not, I would think it would be a convenient option, of course, there is export to excel, you can do it that way.
Sorry, am I the only one who can't attach files with the extension:
ReportOptimizer-1024008.xml
testergraph.report.2012.03.29.csv
Or is it undesirable to attach them?
Sorry, am I the only one who can't attach files with the extension:
ReportOptimizer-1024008.xml
testergraph.report.2012.03.29.csv
Or is it undesirable to attach them?
Don't you have a line below the review button?
Max upload size 16 MB, max 10 attachments (.gif .png .jpg .jpg .jpeg .zip .txt .log .mqh .ex5 .mq5 .mq4 .ex4 .mt5 .set .tpl .flv)
in your case into a .zip archive
and don't you have a line below the review button:
Max upload size 16 MB, max 10 attachments (.gif .png .jpg .jpg .jpeg .zip .txt .log .mqh .ex5 .mq5 .mq4 .ex4 .mt5 .set .tpl .flv)
in your case in a .zip archive