From the idea to the actual bill. - page 4

 
Then spare a week or two to make sure that the results online match the trades in the tester.
 
Rosh:
Then spare a week or two to make sure that the results online match the trades in the tester.

The question is that the online testing is the most accurate, but the question is how to get as close to reality as possible offline.
In this version of Expert Advisor trades are made approximately 1 time per day, and to test it more or less adequately 1-2 weeks may not be enough (IMHO), because it's only 5 - 10 trades, to my mind it is not enough to draw conclusions, and I don't want to wait some months and then realize there is some error in Expert Advisor.
 
Enough is enough, in this thread Some thoughts aloud I did it for six months, but in fact it could have been limited to a month, as the trades were matched. The only thing is that I added global variables to reliably disallow a second order in the same day (in the online trading version).
 
HIDDEN:
Rosh:

Expert stands on a demo account for 3 days, going uphill.
It's like Zoshchenko in "Lemonade". Quit drinking. I don't drink for an hour, I don't drink for another hour. To test the strategy on a short interval, the best way is to put Expert Advisors on the symbol portfolio. Then some of them will slowly drain, while others will accumulate equity and release it into the balance. And the total result of the portfolio will show what can be expected from the trading strategy.
 
WATCHING AND SMILING, NO I'M NOT AN IDIOT, IT'S JUST INTERESTING TO WATCH PEOPLE GO DOWN THE SAME ROAD AS YOU, BUT YOU'RE FLATTERED THAT YOU WENT DOWN IT MUCH EARLIER......
2HIDDEN - YOU'RE IN FOR A DISAPPOINTMENT,.... IF YOUR EA USES DATA FROM HIGHER TIMEFRAMES WHEN TESTING AND WORKING, I WOULD ADVISE NOT TO FLATTER ONESELF WITH BEAUTIFUL CHARTS, DRAWN BY METAQUOTEV'S TESTER. I THINK THE RESULTS OF THE TEST AND THE RESULTS IN REAL LIFE ARE LIKE HEAVEN AND EARTH.
 
teleworker:
IF YOUR EXPERT ADVISOR USES DATA FROM OLDER TIMEFRAMES WHEN TESTING AND WORKING, THEN I WOULD ADVISE NOT TO FLATTER ONESELF ABOUT THE BEAUTIFUL GRAPHS DRAWN BY THE METAKVOTE TESTER. I DON'T KNOW IF HE'S SPYING ON THE HISTORY OR SOMETHING ELSE, BUT THE RESULTS IN THE TEST AND THE RESULTS IN REAL LIFE ARE LIKE HEAVEN AND EARTH
Strangely enough, my results match up to a point.
 

Older timeframes are not used in this Expert Advisor.

Moreover, it is not working on the zero bar, or rather calculations are performed on the OPEN bar only, the algorithm calculates it based on the history which is already a fact.


It's not the first day that I have been writing Expert Advisors and inductors and reading the forum, so I have not encountered any simple tester's ambiguities.

 
You mentioned:
"The history on which the Expert Advisor was tested was downloaded from viac.ru, supplemented with history from Alpari and an alternative version from History Center by MetaQuotes Software Corp." That is, you ran the Expert Advisor on data collected partly from viac.ru and partly from Alpari. You also ran it for the same period as an alternative on data from History Center. Did the profit chart differ and how much did it differ in this case when running it on different data? It is also possible to make statistics on starts. That is, start the Expert Advisor at various points of time during the testing period and see if it fails at the initial stage. Then test it on a longer period of time, where the market trend changes and the Expert Advisor starts to fail. If the profit curve of the EA will be the same as that of the tester, it may be sent to the real account.
I like your profit curve too. I wonder what kind of indicators you have. :)
 

I transferred the test trades to the chart, looked it up, very cool.... Not very cool I would say. I got worried about one thing, Expert Advisor is very good at predicting abrupt news movements that seem to follow from nowhere. Knowing Hidden's qualification it is hard to believe that any of his indicators has peeked into the future, but no one is safe from mistakes. I hope it is not so, that the firstgrail known to me has finally been created and wish good luck in trading.

I advise to use not a demo, but at least a micro-real, the Expert Advisor certainly deserves it, because a couple of quid won't make a hole in your pocket and the value of such a test is much higher and the results are more informative. Good luck again.

 
Also on each transaction introduce a margin of error, which will be similar to requoting and also add swaps for the selected currency pairs