ForumSections

Limit order slippage statistics on the exchange - page 3

New comment

Start making a demo mt5 account with bx. You will receive an email with a link to the distribution. At the server selection stage you choose not a demo server, but for real trading. Create an account with arbitrary data. Create a certificate. All you have a real account with zero balance with real quotes and history.

pivomoe:

Start making a demo mt5 account with bx. You will receive an email with a link to the distribution. At the server selection stage you choose not a demo server, but for real trading. Create an account with arbitrary data. Create a certificate. All you have a real account with zero balance with real quotes and history.

Cool - it worked, thank you! Heaven and earth - tick data of demo and real. Thanks again!
fxsaber:

In the "based on real ticks" mode the positive slippage of limit orders is about 50% higher than in the "based on generated ticks" mode.

This results in a mishap - we introduced real ticks to increase the accuracy of the tester, but introduced positive slippage of limit orders, artificially hovering over the backtest results.

On the exchange, limit orders do not slip, but are executed exactly at the order price. But this is not the case in the tester.

This bug can be avoided using the library described in the link above. But this is a crutch solution. It makes sense when the tester itself works accurately.

I'm asking forum members for their opinions on this subject. Since, for obvious reasons, the opinion of one member of the community is poorly taken seriously by the developers.

It's a pity no one has spoken out.
[Deleted]
fxsaber:
It's a pity that no one had a say in it.

This topic has already been discussed here on the forum somewhere and the developers themselves seem to have said they'll fix it in the new builds. Try to look for it, I didn't really get into it...

P.s. Here's a topic that was also unanswered https://www.mql5.com/ru/forum/86591/page4.

В билде 1340 MT5 очень странное исполнение отложенных ордеров на FOREX в тестере стратегий
В билде 1340 MT5 очень странное исполнение отложенных ордеров на FOREX в тестере стратегий
  • reviews: 1
  • www.mql5.com
Если тестировать на «OHLC на M1», то ВСЕ ордера исполняются с проскальзыванием в 30-50 пятизначных пунктов...
Maxim Dmitrievsky:

This topic has already been discussed here on the forum somewhere and the developers themselves seem to have said they'll fix it in new builds. Try to look for it, I didn't really get into it...

P.s. Here was a topic, also unanswered https://www.mql5.com/ru/forum/86591/page4

They didn't fix it.
fxsaber:
It's a shame no one spoke up.

Because you have no evidence base whatsoever.

It's much easier to save the report and zip it up. Show me an example of one transaction with the calculations.

Renat Fatkhullin:

Because you have no evidence base whatsoever.

It's much easier to save the report and zip it up. Show an example of a single transaction with a statement.

Thought that would be enoughhttps://www.mql5.com/ru/code/16134.

Got it. I'll prepare it.

SlipPage
SlipPage
  • votes: 10
  • 2016.08.25
  • fxsaber
  • www.mql5.com
Расчет проскальзываний совершенных сделок в валюте счета.

Adviser

#define OP_BUY ORDER_TYPE_BUY
#define OP_SELL ORDER_TYPE_SELL
#define OP_BUYLIMIT ORDER_TYPE_BUY_LIMIT
#define OP_SELLLIMIT ORDER_TYPE_SELL_LIMIT

// Кусок из https://www.mql5.com/ru/code/16006
class MT4ORDERS
{  
public:   
  static int MT4OrderSend( const string Symb, const int Type, const double dVolume, const double Price, const int SlipPage, const double SL, const double TP,
                            const string comment = NULL, const int magic = 0, const datetime dExpiration = 0, color arrow_color = clrNONE )
  {
    MqlTradeRequest Request = {0};

    Request.action = (((Type == OP_BUY) || (Type == OP_SELL)) ? TRADE_ACTION_DEAL : TRADE_ACTION_PENDING);
    Request.magic = magic;

    Request.symbol = ((Symb == NULL) ? ::Symbol() : Symb);
    Request.volume = dVolume;
    Request.price = Price;

    Request.tp = TP;
    Request.sl = SL;
    Request.deviation = SlipPage;
    Request.type = (ENUM_ORDER_TYPE)Type;

    Request.type_filling = ORDER_FILLING_RETURN;

    if (dExpiration > 0)
    {
      Request.type_time = ORDER_TIME_SPECIFIED;
      Request.expiration = dExpiration;
    }

    Request.comment = comment;

    MqlTradeResult Result;

    return(::OrderSend(Request, Result) ? ((Request.action == TRADE_ACTION_DEAL) ? (int)Result.deal : (int)Result.order) : -1);
  }

  // Такая перегрузка позволяет использоваться совместно и MT5-вариант OrderSend
  static bool MT4OrderSend( const MqlTradeRequest &Request, MqlTradeResult &Result )
  {
    return(::OrderSend(Request, Result));
  }
};

// Обязательно ПОСЛЕ #include <Trade/Trade.mqh>: CTrade::OrderSend
#define OrderSend MT4ORDERS::MT4OrderSend

void OnTick( void )
{
  static int TicketBuyLimit = 0;
  static int TicketSellLimit = 0;
  
  const datetime time = TimeCurrent();
  
  if ((time == D'2016.08.11 19:12:33') && (TicketBuyLimit == 0))
    TicketBuyLimit = OrderSend(Symbol(), OP_BUYLIMIT, 1, 95090, 0, 0, 0);
  else if ((time == D'2016.08.16 18:44:02') && (TicketSellLimit == 0))
    TicketSellLimit = OrderSend(Symbol(), OP_SELLLIMIT, 1, 97070, 0, 0, 0);

  return;
}

Tester Log

MR      0       16:47:50.960    Tester  RTS-9.16: ticks data begins from 2016.08.01 00:00
LE      0       16:47:50.963    Core 1  agent process started
CE      0       16:47:51.473    Core 1  connecting to 127.0.0.1:3000
NR      0       16:47:52.736    Core 1  connected
DJ      0       16:47:52.741    Core 1  authorized (agent build 1401)
RR      0       16:47:52.743    Tester  RTS-9.16,M1 (BCS-MetaTrader5): testing of Experts\LimitsFill.ex5 from 2016.08.11 00:00 to 2016.08.17 00:00
GP      0       16:47:52.763    Core 1  common synchronization completed
DI      0       16:47:52.780    Core 1  RTS-9.16: ticks synchronized already [47 bytes]
IL      0       16:47:53.493    Core 1  1482 bytes of tester parameters loaded
PH      0       16:47:53.493    Core 1  188 bytes of input parameters loaded
OR      0       16:47:53.493    Core 1  8562 bytes of symbols list loaded
MI      0       16:47:53.493    Core 1  expert file added: Experts\LimitsFill.ex5. 8164 bytes loaded
FR      0       16:47:53.493    Core 1  initial deposit 100000.00 RUR, leverage 1:0
EI      0       16:47:53.493    Core 1  successfully initialized
IS      0       16:47:53.493    Core 1  35 Kb of total initialization data received
QJ      0       16:47:53.493    Core 1  Intel Core i7-2700 K  @ 3.50 GHz, 16301 MB
LR      0       16:47:53.493    Core 1  RTS-9.16: symbol to be synchronized
PF      0       16:47:53.493    Core 1  RTS-9.16: symbol synchronized, 3224 bytes of symbol info received
RJ      0       16:47:53.493    Core 1  RTS-9.16: load 31 bytes of history data to synchronize in 0:00:00.000
PM      0       16:47:53.493    Core 1  RTS-9.16: history synchronized from 2015.06.22 to 2016.09.01
IS      0       16:47:53.493    Core 1  RTS-9.16: ticks synchronization started
JD      0       16:47:53.493    Core 1  RTS-9.16: load 38 bytes of tick data to synchronize in 0:00:00.000
NO      0       16:47:53.493    Core 1  RTS-9.16: history ticks synchronized from 2016.08.01 to 2016.09.01
RI      0       16:47:53.493    Core 1  RTS-9.16,M1: history cache allocated for 610971 bars and contains 43890 bars from 2015.06.22 10:02 to 2016.08.10 23:49
CM      0       16:47:53.493    Core 1  RTS-9.16,M1: history begins from 2015.06.22 10:02
DD      0       16:47:53.493    Core 1  RTS-9.16,M1 (BCS-MetaTrader5): generating based on real ticks
ML      0       16:47:53.493    Core 1  RTS-9.16,M1: testing of Experts\LimitsFill.ex5 from 2016.08.11 00:00 to 2016.08.17 00:00 started
LQ      3       16:47:53.493    Core 1  RTS-9.16 : real ticks begin from 2016.08.01 00:00:00
GK      0       16:47:53.493    Core 1  2016.08.11 19:12:33   buy limit 1.00 RTS-9.16 at 95090 (95260 / 95270 / 95270)
EK      0       16:47:53.493    Core 1  2016.08.11 19:40:17   order [#2  buy limit 1.00 RTS-9.16 at 95090] triggered
GJ      0       16:47:53.493    Core 1  2016.08.11 19:40:17   deal #2  buy 1.00 RTS-9.16 at 95050 done (based on order #2)
 GR      0       16:47:53.493    Core 1  2016.08.11 19:40:17   deal performed [#2  buy 1.00 RTS-9.16 at 95050]
GP      0       16:47:53.493    Core 1  2016.08.11 19:40:17   order performed buy 1.00 at 95050 [#2  buy limit 1.00 RTS-9.16 at 95090]
QR      0       16:47:53.493    Core 1  2016.08.16 18:44:02   sell limit 1.00 RTS-9.16 at 97070 (97020 / 97030 / 97020)
GF      0       16:47:53.493    Core 1  2016.08.16 19:00:00   order [#3  sell limit 1.00 RTS-9.16 at 97070] triggered
CG      0       16:47:53.493    Core 1  2016.08.16 19:00:00   deal #3  sell 1.00 RTS-9.16 at 97170 done (based on order #3)
 FJ      0       16:47:53.493    Core 1  2016.08.16 19:00:00   deal performed [#3  sell 1.00 RTS-9.16 at 97170]
DO      0       16:47:53.493    Core 1  2016.08.16 19:00:00   order performed sell 1.00 at 97170 [#3  sell limit 1.00 RTS-9.16 at 97070]
KR      0       16:47:53.493    Core 1  final balance 102788.71 RUR
IF      0       16:47:53.493    Core 1  RTS-9.16,M1: 1122105 ticks, 3240 bars generated. Test passed in 0:00:00.717 (including ticks preprocessing 0:00:00.124).
JE      0       16:47:53.493    Core 1  252 Mb memory used including 35 Mb of history data, 64 Mb of tick data
KK      0       16:47:53.493    Core 1  log file "C:\Program Files\BCS Broker MetaTrader 5 Terminal\Tester\Agent-127.0.0.1-3000\logs\20160902.log" written
DJ      0       16:47:53.507    Core 1  connection closed

Slippage of limit orders is marked in bold. The slippage in the tester is worse when a Limit order slips through the session - at opening. But I didn't take these cases as an example. I took the usual market.

Is it reproducible?

Unfortunately, the debugging doesn't work, so it was not convenient to create the example.

Forum on trading, automated trading systems and strategy testing

Bugs, bugs, questions

fxsaber, 2016.09.01 20:18

Unable on RTS-9.16 BCS-MetaTrader5 to do debugging of EA by CTRL+F5. The tester writes
Tester  Leverage 1:1 set error


I have the same problem. Slips on both stocks and futures.
fxsaber:

Adviser

Tester Log

Slippage of limit orders is marked in bold. The slippage in the tester is the highest when a Limit order slips through the session - at opening. But I didn't take these cases as an example. I took the usual market.

Is it reproducible?

Unfortunately, the debugging does not work, so it was not convenient to create an example


Try it not in the tester, but in demo (better Opening, it has higher speed).
12345678910
New comment