A little surprised :) Thought I'd share and ask a NOT rhetorical question.

[Academic]

I'm at a loss for words. :(

I've been... I've been poking around with an idea and I didn't notice I'd written something like a tester. :)

Of course, it's a simple one, but the balance is counting and shows all ticks since 2006. It's multicurrency, but I'm just working with only one symbol.

What puzzled me - 15000 runs, all ticks since Feb 15, 2011, (direct search of two parameters) takes FOUR minutes. Everything is running on one core, of course.

But now I sit and think - what for do I need MT5 slow tester? For the love of God, I really don't get it - to check the strategy, everything is fast and easy. So why is the MT5 tester so useful?

Name the advantages. I believe there are !!! But I'm at a loss - what am I waiting for? :)

Now I will run the same 15000 since 2007 with ticks at 300 ticks per minute. And I'll get upset. :) Please change your mind.

[Rashid Umarov]

Academic:

Now I will run the same 15000 from 2007 with ticks at 300 ticks per minute. And I'll get upset at all. :) Change your mind plz.

Run MovingAverages in your "tester", let's see :)

You shouldn't call a simple calculator a tester.

[Academic]

Rosh:

Run MovingAverages in your "tester" and see :)

You shouldn't call a simple calculator a tester.

Well actually I have in my "strategy" just a calculation for each bar a retrospective analysis for about 300 bars backwards. So, it is almost there.

You have to do a lot of math there. I'm sure it's faster than the number cruncher. So far, I see only one useful feature - for "buyers" of indicators and strategies to be able to run them in the tester. Not programmers. That's all. :(

[Rashid Umarov]

Academic:

There's a lot of counting to do. The fact that it is faster than my digitizer is for sure. So far I see only one useful thing - for "buyers" of indicators and strategies to be able to run them in the tester. Not programmers. That's all. :(

Put an indicator in your "tester", how easy is it? And then make your "strategy" from the "tester" work online, not in a spherical vacuum of your "tester".

[Academic]

Rosh:
Put in your "tester" indicator, how easy is it? And then make your "strategy" from the "tester" work online, not in the spherical vacuum of your "tester".

Alien indicator? I think it would be a pain in the ass. I won't even try it.

And online, I don't think it would be difficult. It's not a TC, it's an idea. You will still have to adapt it. But the fact that you need a language for programming strategies is not in dispute. :) But the value for me of a slower racer on the historical data, alas, as it seems to me now - not obvious. :(

So let's write - check "other people's" strategies. That is exactly what I wrote. But it is only a tester - not an optimizer after all. :)

[Mykola Demko]

Academic:

Well, actually my "strategy" has just a calculation for each bar with retrospective analysis for about 300 bars ago. So, it is almost there.

There is a lot of calculating to do. It's faster than my digits, that's for sure. So far, I see only one useful feature - for "buyers" of indicators and strategies to be able to run them in the tester. Not programmers. That's all. :(

I agree that it's faster to run through the quotes and calculate the difference between opening and closing signals than to run the Expert Advisor in the tester.

But the tester is a universal trading simulation environment. It does not depend on the algorithm of the Expert Advisor.

For example, try to simulate in your calculator a grid operator that works with pending orders, you'll get tired of adapting.

This universality is the reason of wasting resources. So, the tester is needed first of all for a trader, not for a programmer. The programmer can check his or her strategy by writing his or her own calculator.

---

But do not grasp the word "programmer". In MQL there are no programmers who know nothing about trading. So, the programmer is an extension of the notion that includes trading experience. Just like many traders know the basics of programming in MQL.

The trader, as I understand it, is a person who trades and this is his main occupation. He may or may not know programming.

A programmer is a person writing programs; he may or may not be a successful trader. But he must know the basics of trading, otherwise he has nothing to do in MQL.

Now, tell yourself, who needs a tester - a programmer who may do without it, or a trader who must call the programmer for any minor detail?

[Renat Fatkhullin]

It is useful for everyone to write their own tester.

[Andrey Dik]

Renat:
It's good for everyone to write their own tester.

Golden words.

No, really, they are golden. A tester, moreover, written by your own hands, really opens your eyes to many things.

And to write a tester in which you can run ANY TC without any fuss is actually very difficult. Most likely, if the tester is written with your own hands, it's just a "calculator".

[Mykola Demko]

  Немного удивлен :)

I'm surprised how the author is going to share the knowledge, since there are no codes or new ideas.

And the fact that the calculator is faster than the tester is an old story, I've known it since 2004, since MQL2.

Although I have to admit that now the gap is wider in favour of the calculator. It indicates an advanced language, which is almost at the same speed as SRR.

But it almost has a lot of application benefits.

[Academic]

Renat:
It is useful for everyone to write their own tester.

The paradox is that the main advantage of the optimizer is its performance. All other things being equal, the one that allows you to draw something is better, for example. :)

In a word, the advantages of MT-optimizer, namely, the optimizer, are not evident to me.

But not the main point, I mean, that may be it's worth building a monster for you - anyway a digital crusher will be more useful - you can use it only if it won't lose too much. Which I personally doubt. Not because you're not finished, but simply because IMHO - you can not combine a tester and optimizer - they have different tasks. Think about it - one must check the "ideas" and quickly search for YES/NO. The other is to simulate reality.

[Academic]

joo:

Words of gold.

No, really, they are. A tester, and even written with your own hands, really opens your eyes to many things.

And to write a tester in which you can run ANY TC without any fuss is actually very difficult. Most likely, if the tester is written with your own hands, it's just a "calculator".

I've already written more than one what's called a tester - the most complicated one :) no worse than the MT-ish one.... Well even better - faster. :) But it's a commercial product. :) And now from this experience I can say (share) it is necessary to differentiate tasks - for tester one (simulation) and for the optimizer another. And one must not increase the performance at the expense of that the essence of realism is reduced ( for example only by the opening prices ).