Which instruments to trade at Otkritie - page 10

 
Dmitriy Skub:
It would be nice to add a function to check the status of ticks uploaded/injected - in the local terminal database. Something like CheckXXX( symbol ). So you don't have to keep pulling the copitic all the time.
What I mean. Or better OnCopyTicks. The internal architecture of ticks handling is obviously cool. But externally it all looks very sad.
 
Sergey Chalyshev:
Will the order in the BuyLimit tester be triggered by Ask or Bid?
MetaQuotes Software Corp.:

Of course by Ask.

Check it yourself, it is not difficult.

Was hoping the man was wrong, but no, checked on real ticks in Opening, he is. Here you go: bid below bylimit. Miracles of the glass. ))

And here's the flipper below the limit and it's still standing. What an unbreakable limiter. )))

But seriously - it's all nonsense. You still cannot completely reproduce the reality in the tester (at least because the orders placed in reality change the glass, which cannot be reproduced in the tester; otherwise it would mean that the tester changes the data on which it is testing), so should we be very meticulous about these nuances? Well, you can get an approximate picture of how the strategy works from the asc and the asc. The test does not guarantee anything.

 
Sergei Vladimirov:

I was hoping the man is mistaken, but no, I checked it on real ticks in the Open and it's true. Here you go: bid below bylimit. Miracles of the glass. ))

It has been said several times before that pricing is not affected by the tester.

I have another question for the developers. If the tester uses only Bids and Asks in "real ticks" mode, then why does it run the Expert Advisor throughCOPY_TICKS_ALL-ticks instead ofCOPY_TICKS_INFO-ticks? That's a waste of money.

 
Anton Zverev:

It has been said several times before that pricing is not affected by the tester.

What does this have to do with my post? I don't think I've ever said otherwise.
 
Sergei Vladimirov:
What does this have to do with my post?
Because BuyLimit in the tester, is just an order to the tester that when the ask reaches BuyLimit, to make BUY. No more and no less.
 
Anton Zverev:
Because BuyLimit in the tester is just an order to the tester that when the ask reaches BuyLimit, to make BUY. No more and no less.
Strictly speaking, this should not be the case. Bid should not be activated by BidLimit either - the fact that in my first screenshot the bid is lower than the limit, is not a big deal: in reality the bid would just be a bit higher, at the level of my limit, that's all. You can consider that in the tester this bid is just not conventionally shown. But when the flipper falls below the bylimit, it's already a mistake. It turns out that there was a trade below the limit, while it was not executed, which can not be. The flipper should be an order to the tester, not an ask.
 
Sergei Vladimirov:
But when the flipper falls below the bylimit it is already a mistake. It turns out that the deal was executed below the limit, while it was not executed, which cannot be. The flipper should be an order to the tester, not an ask.

You have to understand the nature of flipper emergence. A flipper on the real appears when someone decides to make a market, looking at the current trading environment. It is the trading environment (history + current state (stakes)) that affects someone's desire to make a market there. This means that if you are in the real world putting in and removing limiters, it may affect someone's decision whether or not to make a market. It may be that if that person saw your limit, they wouldn't or wouldn't send a market to it. Alternatively, if you put a big limit somewhere away from the best price, that person might not be able to market someone else's limit that was a best price.

All that said, the fins have nothing to do with the tester performance. Moreover, you have to be very careful to apply fin information at all in any trading logic. The tester should in no way take fins into account, nor should it influence pricing.

The tester is correct with execution.

In this sense, very strange is the construction of bars (source information for indicators) by fins. Historically, this has been the case, but it is, to put it mildly, questionable. Not mildly - it is erroneous. Much more informative would be bars built on the average price of the same spread.

 
Anton Zverev:

You have to understand the nature of flipper emergence. A flipper on the real appears when someone decides to make a market, looking at the current trading environment. It is the trading environment (history + current state (stakes)) that affects someone's desire to make a market there. This means that if you are in the real world putting in and removing limiters, it may affect someone's decision whether or not to make a market. It may be that if that person saw your limit, they wouldn't or wouldn't send a market to it. Furthermore, if you put a big limit somewhere away from the best priced limit, that someone wouldn't file their market with someone else's limit that was a best priced limit.

You're trying to justify that it's impossible to simulate the behaviour of other participants in the tester. Is anyone arguing with that?

All that said, the flippers have nothing to do with the tester performance. Moreover, one has to be very careful to apply fin information at all in any trading logic. The tester should in no way take fins into account, nor should it influence pricing.

Everything is correct in the Strategy Tester with respect to execution.
The flippers are transaction prices. Real trades that took place in the real world. If there was a trade below the bylimit or above the selimit, then that limit should definitely have been executed. If the price of the flipper changed and fell to bylimit or below, it means that in reality someone gave a market order to buy at or below the price of your order. The order must be executed and this will not affect the pricing in the tester in any way. Execution by ask or bid is wrong, the exchange is not a DC. And about the usefulness of such meticulous tests I have said.

In this sense it is very strange to build bars (input information for indicators) using fins. Historically it has been so, but it is, to put it mildly, questionable. Not mildly - it is erroneous. Much more informative would be the bars built on the average price of the same spread.
Horror. Flipper prices - I repeat - this is the real history of deals, and it is displayed on the charts. This is the price that activates stop orders and in the tester should activate limit orders as well. And the average spread price does not contain any information at all

 
Sergei Vladimirov:
Horror. The flipper prices - I repeat - are the real deal history and are displayed on the charts. This is the price that activates stop orders, and in the tester it should activate limit orders as well. The average spread price doesn't contain any information.

This is the history of deals. It has nothing to do with the backtest. If there was an asc, the exchange was obliged to fill in, send you to the market. But if there was a flipper, it was for that environment where your limiters weren't.

In short, showing stupid stubbornness. You are annoying.

 
Sergei Vladimirov:

I had hoped the man was wrong, but no, I checked on real ticks in Otkritie, and he is. Here you go: the bid is below the bylimit. Miracles of the glass. ))

You are mistaken.

In the above screenshot you can see that Ask (namely, the buy limit is triggered by Ask) is higher than the buy limit.


And here's the flipper below the limit and it's still standing. What an unbreakable limiter. )))

Wrong again as Ask has not broken through the bylimit level downwards. What are you picking on the flipper for when the condition is triggered by the bylimit?


But seriously - it's all nonsense. You still cannot completely reproduce the reality in the tester (at least because the orders placed actually change the glass, which cannot be reproduced in the tester; otherwise it would mean that the tester changes the data on which it is testing), so why take such a scrupulous approach to these nuances? Well, you can get an approximate picture of how the strategy works from the asc and the asc. The test doesn't guarantee anything, anyway.

In all seriousness, you are simply illiterate.

Apparently you have some sketchy knowledge in your head and are not trying to understand what you are being told.