Why don't I read the articles? - page 15

[---]

hrenfx:

Everything was told in narrative form. No arguments or proving anything. No aim to convince anyone of anything. Just a lingering conversation, giving some insight into the interlocutors and their experiences.

I did.

Don Quixotes :)

[Valerii Mazurenko]

S# is just a wrapper over dll or COM-objects of trade terminal developers. And one for quickq, another for plaza, and a third for something else (meaning implementation of classes). Maybe, from the plaza you can obtain all the changes of rates without gaps (and the plaza costs money), but it will not succeed on the brokerage platform, in spite of S# or self-made wrappers. + the sluggishness of the .Net platform + some generality (and hence inefficiency) of the architecture - not the best option for hft (the best is to write in C++ yourself, for example).

I doubt that S# can be easily extended to connect to a Brazilian exchange, for example.

[Dmitry Fedoseev]

hrenfx:

...

1. Unsynchronized bars in different symbols. For example, it shows that OPEN(EURUSD) coincides in time with OPEN(GBPUSD). But in fact, it is not.

It is not a problem at all if the analysis is performed by closed bars.

[hrenfx]

This is a complex problem:

For bars formed by the потивокой model, things are a little more complicated. The tester must already be "at closing prices". In doing so, the strategy should have control over the opening, not theclosing bar. I.e. at the moment of bar closing the strategy should perform the corresponding trade actions (analysis). In MT4 and MT5 it is achieved by creating the event of bar closing (for example, through locking of the Expert Advisor in MT4).

[Dmitry Fedoseev]

hrenfx:

For bars formed on a contrary model, things are a little more complicated. The tester should already be "at closing prices". In this case the strategy should have control not of the bar opening, but of the bar closing. I.e. at the moment of bar closing the strategy should perform appropriate trade actions (analysis). In MT4 and MT5 this can be achieved by creating a bar close event (e.g., by looping the Expert Advisor in MT4).

What is this heresy about the bar close event?

You know, who likes to live in problems, they successfully find them in anything.

[hrenfx]

Instead of wiggling your smoothed-out meanderings, you choose the most effective way to shut your interlocutor up.

The main objective of the tester is to achieve results as close to the real thing as possible. It is on the basis of this postulate that that branch was created.

[Dmitry Fedoseev]

hrenfx:

Instead of wiggling your smoothed-out meanderings, you choose the most effective way to shut your interlocutor up.

The main objective of the tester is to achieve results as close to the real thing as possible. It is on the basis of this postulate that that branch was created.

Tell me, how can you tell when a bar has closed?

[hrenfx]

In MT4 I do it through a looped Expert Advisor, which checks the time change of the trading server.

[Dmitry Fedoseev]

hrenfx:

In MT4 I do through a looped EA, which has a time change check on the trade server.

An EA that is impossible to test while doing so, just for the sake of winning one tick in an unknown direction.

[hrenfx]

It is not possible to test in the default MT4 tester. Although no one prevents you from changing the history before testing in this way: Open[i] = Close[i + 1].

I prefer the maximum coincidence of the tester's results with real ones. This is largely the reason for writing my own calculator.

And it's not about one tick. This is especially important for multicurrency, where there is a lot of arbitrage on opening prices due to unsynchronization.