a trading strategy based on Elliott Wave Theory - page 272
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
I thought it was me who was trying to talk you into using Hirst. You just can't talk me out of using it. Statistics, though. :о)))
At the moment I've got everything in Matkadec. I am thinking of making a version for MT for testing. The only question is: in testing mode at the moment of start of any function (for example, forecast) which requires long time for calculation (for example, 1-7 hours), the tester will supply quotations during calculation or wait for completion of the function?
I do not know. I always test in standalone terminal.
I don't know. I always test in the offline terminal.
I was referring to the same mode. I meant whether the tester will emulate new arrivals while the forecast is being calculated (1-7 hours)
Не знаю. Я всегда тестирую в автономном терминале.
I was referring to the same mode. I meant whether the tester will emulate new arrivals while the forecast is being calculated (1-7 hours)
The tester does not take the time factor into account. It will first count your start() function to the end and then move on to the next new tick.
Something suspiciously long calculation time.
Thanks, that's what I assumed theoretically. This is just what is needed for "objective" testing. As far as I understand, it will work in testing on a demo account, too? I.e. the Expert Advisor will wait for the start() function to be executed completely and will not be able to realize a parallel check of the process based on incoming quotes until the start() function is calculated?
To Rosh
Something suspiciously long calculation time.
I have an AMD Athlon 64 Processor 3800+ 2.4 GHz, 2 GB RAM. Hearst calculation for a sample of 600 samples takes about 20 minutes but the calculation is executed in MathCAD. No obvious lapses in calculation algorithm, i.e. it calculates in optimal way from MathCAD programming viewpoint. But it is a small part of what is, for example, entropy is considered about 40-50 minutes for the same sample.
Such duration of calculation, due to point one in my strategy: "the sample that is the minimum in terms of the strength of the relationship between the elements, calculated on the basis of the correlation criterion, is determined. We get a total sample on which it makes sense to look for something" and a rather "contrived" model. The number of requested samples varies from 300 to several thousands, depending on analyzed data.
The results in MathCAD are quite satisfying (more than satisfying), but I understand, that it's not enough. So, I've decided to implement simplified version in MT for the time being, just for testing. I'll look at the results.
The tester does not take the time factor into account. First it counts your start() function to the end and then moves on to the next new tick.
Thanks, that's what I assumed theoretically. This is exactly what is needed for "objective" testing. As far as I understand, it will also work when testing on a demo account? I.e., the Expert Advisor will wait for the complete execution of the start() function and will not be able to realize a parallel control based on incoming quotes until the start() function is calculated?
Working on a demo account is working in real time. MT4 will of course end your function start(), no matter how long it takes. But only during this time new quotes will be received. If at the end of your long function start() you want to make a trade, you will need to use function RefreshRates(), which updates the current market data, including the current Ask and Bid. That is, with Bid and Ask values that existed a few hours ago (at the moment of start of calculation) you will not be able to open an order on a demo account (and on a real one, of course). But you can open an order only for the current ones, after you use RefreshRates() function.
Тестер фактор времени не учитывает. Сначала он досчитает Вашу функцию start() до конца, а потом перейдёт к следующему новому тику.
Спасибо, я так теоретически и предполагал. Это как раз то, что нужно для «объективного» тестирования. На сколько я понял, это так же будет работать и при тестировании на демо счете? Т.е. эксперт будет ждать полного выполнения функции start(и всего того, что в ней напихано) и реализовать параллельный контроль процесса на основе поступающих котировок не удастся, пока не завершиться расчет функции start()?
Working in a demo account is working in real time. MT4 will, of course, finish your start() function, no matter how long it takes. But only during this time new quotes will be received. If at the end of your long function start() you want to make a trade, you will need to use function RefreshRates(), which updates the current market data, including the current Ask and Bid. That is, with Bid and Ask values that existed a few hours ago (at the moment of start of calculation) you will not be able to open an order on a demo account (and on a real one, of course). But you can open an order only for the current ones, after you use RefreshRates() function.
Thank you very much for the tip. I will have to pay due attention to controlling the process after the calculation. After all it will be necessary to check correctness of the forecast.
The calculation time is suspiciously long.
I have AMD Athlon 64 Processor 3800+ 2.4 GHz, 2 GB RAM. Hearst calculation for 600 samples takes about 20 minutes, but the calculation is done in MathCAD. No obvious lapses in calculation algorithm, i.e. it calculates in optimal way from MathCAD programming viewpoint. But it is a small part of what is, for example, entropy is considered about 40-50 minutes for the same sample.
Such duration of calculation, due to point one in my strategy: "the sample that is the minimum in terms of the strength of the relationship between the elements, calculated on the basis of the correlation criterion, is determined. We get a total sample on which it makes sense to look for something" and a rather "contrived" model. The number of requested samples varies from 300 to several thousands, depending on analyzed data.
The results in MathCAD are quite satisfying (more than satisfying), but I understand, that it's not enough. So, I've decided to implement simplified version in MT for the time being, just for testing. I'll look at the results.
My Hearst calculation for a 3000-bar sample in MQL4 takes about 40 milliseconds. If I'm not mistaken I will try to use Hirst's code in MQL4 but if I am, I'd rather use MathCad as a substitute.
Something is wrong with the calculations in any case. My email is rosh AT metaquotes DOT ru.