Discussion of article "How to Secure Your Expert Advisor While Trading on the Moscow Exchange" - page 4

 
Konstantin Seredkin:

The article is great, I have implemented in my robot the inputs by limits taking into account the possible slippage of the stack calculated by the analysis.

I also replaced all stop losses and take profits set by the terminal with stop limit orders with the same consideration of slippage.

Tests have shown that this method is not competitive on the Moscow stock exchange, the system performance has improved by about 5-8%.

Thanks to the author for the article and examples.

I don't believe it, unless your system is so-called "high-frequency trading".

 
Vasiliy Sokolov:

:)

Be careful with the tests though. MT5 is not able to calculate slippage on stack tests after all (like no other public tester).

It is highly advisable that you ALWAYS use limit orders. Even for simulating stop orders it is better to use limit orders.

If you trade illiquid, the recommendations in the article are STRONGLY REQUIRED. Simply put, there should not be a single STOP or MARKET order on illiquid.

Just all tests I conduct with the use of the glass, just the robot trades on the glass catching the density of orders.

All the analysis in the strategy tester for entry point, TP and SL setting is calculated using the data from the stack, the stack data was recorded every day for almost half a year, so there are almost 500 gigabytes of data on different futures, for testing and optimisation of strategies on the stack and tape of deals, supply and demand and open interest.

Using terminal stops and takeouts got non-compliances.

Example.

Trading the RTS index for a week, then on the recorded data I run the same period on the visual, then compare the real and tester trades, so in the tester there were deviations that were caused by this sliding, even 1 lot slid sometimes by 5 ticks in the stack.

Having put all calculations on limit execution, namely, take profit is a limit, and stop loss is a stop limit order in which the limit itself is thrown by the method of calculation of slippage that you offered in your class, + having inserted the account of the spread when opening a position, immediately all the inconsistencies real - tester were cut off, the deals there that there are tick to tick in the history go.

Mikhail Simakov:

I don't believe it, unless your system is so-called "high-frequency trading".


My system tracks densities in the stack at different extrema, finds the right density and enters at the moment of its eating up, the robot takes all these small movements after eating up large volumes, in fact the position in the market hangs for 30-60 seconds no more.

I just lacked this method, I even redesigned the trailing stop, now it is not the terminal stop that is trawled after the price, but the stop limit order, now at the moment of sharp movements when eating and throwing in a large volume, I do not have sliding on the price, maybe I didn't hit it, but at least now I take away the number of ticks by trailing that number of ticks, how many it has already passed initially, with a terminal stop, the stop is 100p of profit, the price jerks to it, I take away 30-60p only, with a stop limit now there is no such thing.

I believe that for scalper systems with quick entry and exit at the moment of sharp market fluctuations, these methods should be 100% used that would not lose profit.

 
the accuracy of limit orders is ambiguous.

for example, one of the most popular brokers has such a picture that limit orders are executed 10-100 pips worse than the set one.

and on asking the broker what is the reason? (because the account history shows at what price the order was set and executed) the broker answers that he does not guarantee the best price....

and having called a broker with a licence of the central bank and asked a question how they are doing with execution of limit orders they said that mistakes happen but they return money for unsuccessful executions without any problems...

and a top foreign broker says that the reality is as follows: what price was set and executed at that price and it doesn't matter that it is much worse than the requested one.
 
Pavel Malyshko #:
the accuracy of limit orders is ambiguous.

for example, one of the most popular brokers has such a picture that limit orders are executed 10-100 pips worse than the set one.

and on asking the broker what is the reason? (because the account history shows at what price the order was set and executed) the broker answers that he does not guarantee the best price....

and having called a broker with a licence of the central bank and asked a question how they are doing with execution of limit orders they said that mistakes happen but they return money for unsuccessful executions without problems...

and a top foreign broker says that the reality is as follows: what was the price and executed at that price and it doesn't matter that it is much worse than the requested one.

Yes. I know this phenomenon with previous Brokers, which for me are in the past forever for this reason. They also had problems with Stop Loss triggering, which moved randomly to the worst side for me by the same 10-100 pips and even more. This is typical for Brokers with NDD.

 
Vladimir Gulakov #:

Yes. I know this phenomenon with previous Brokers, which for me are in the past forever for this reason. They also had problems with triggering Stop Loss, which moved randomly to the worst side for me by the same 10-100 pips and even more. This is typical for Brokers with NDD.

and the way out is brokers with a central bank licence ... or those that are clearly not interested in draining your deposit and have leverages of 1 to 40 like brokers with a central bank licence...

and when I see accounts with such leverages in signals 1 to 40, I wonder why they chose such a broker.




 
Well, gentlemen, it has arrived! Now MOEX is tripling the commission for market makers, while for price makers it is zero. In this regard, switching to limiters has become a necessity.
 
Vasiliy Sokolov MOEX is tripling the commission for market makers, while for price makers it is zero. In this regard, switching to limiters has now become a necessity.

Not a necessity, of course.

In slow strategies the size of the commission is too small compared to the losses received in case of missing a "good" trade, which will definitely escape from your limit.

And in fast strategies there is nothing to do on MT. Or almost nothing.

P.S. You will not be able to adequately test a strategy built on limit trades in MT5 anyway. And in general it is not easy.

 
Dmi3 #:

I didn't, of course.

In slow strategies, the commission is too small compared to the losses incurred if you miss a "good" trade, which is bound to run away from your limit.

And in fast strategies there is nothing to do on MT. Or almost nothing.

P.S. You will not be able to adequately test a strategy built on limit trades in MT5 anyway. And in general it is not easy.

I talked to brokers with a licence and they told me that in case of failure to execute at the best prices, they will compensate everything in case of your request. so if it works in the tester, it will also work on the account

 
Pavel Malyshko #:

I talked to brokers with a licence and they told me that in case of failure to execute at the best prices they will compensate everything in case of your application. so if it works in the tester it will work on the account too

We're talking about Mosbirge.

 
Dmi3 #:

It's about Mosbirge.

its instruments are not tested in MT5?