bid and ask price difference

 

is there bid and ask price difference among demo account plateform,real account plateform, mini account plateform,standard account and ECN account plateform ?these plateform belong to same broker.
which one is faster?

is the amount of ticks equal between them in a certain time ?
i test an ea on demo and real account seprately,they open and close orders almost in the same time,but demo account always get profit,and real account always lose ? why ?

 
YALEWANG:

is there bid and ask price difference among demo account plateform,real account plateform, mini account plateform,standard account and ECN account plateform ?these plateform belong to same broker.
which one is faster?

is the amount of ticks equal between them in a certain time ?
i test an ea on demo and real account seprately,they open and close orders almost in the same time,but demo account always get profit,and real account always lose ? why ?

From my own experience I can say that for the majority of brokers the answer is Yes. I did a thorough check on my own brokers a few months ago, cause I had the same feelings and thoughts.

My checks included running tick loggers for more than a week on a real account and on a demo account. I also run 2 identical ea's on both during that time and captured timing stats of opening/closing orders.


My conclusions at the time where that there are differences in prices and incoming ticks. Differences in account properties (Accountxxx() functions) and market information (MarketInfo() functions). There are huge differences in timing, demo servers being 2-4 times faster in processing any orders. Demo servers also have almost no re-quotes and no slippage and generally u r almost always filled instantaneously with no problems. This does not happen in a live account.


On the other hand I found that Demo servers tend to have a lot of downtime and disconnections as opposed to live servers which are usually very reliable in that regard.


But all this is broker specific, and there might be brokers out there with very little differences... I recommend you do your own checks.

 
gordon wrote >>

...there might be brokers out there with very little differences....

Dont hold your breath on that!

You list many of the differences that will be liable to occur, others are discussed here among other threads:-

https://www.mql5.com/en/forum/110605

https://www.mql5.com/en/forum/108676

https://www.mql5.com/en/forum/110709

https://www.mql5.com/en/forum/109565

There is no point hoping or searching for a demo account server or a live account backtest that will be 100% match to live forward trading

There will always be a small difference at least, at worst you will have profit on the demo & loss on live

The better your system, the less it is affected

IMHO, price driven hedging will have the least variation demo-to-live, spread sensitive scalping bots will show the most difference

With 100% indicator-based EA's somewhere in between

Good Luck

-BB-

 
BarrowBoy:

Dont hold your breath on that!

You list many of the differences that will be liable to occur, others are discussed here among other threads:-

https://www.mql5.com/en/forum/110605

https://www.mql5.com/en/forum/108676

https://www.mql5.com/en/forum/110709

https://www.mql5.com/en/forum/109565

There is no point hoping or searching for a demo account server or a live account backtest that will be 100% match to live forward trading

There will always be a small difference at least, at worst you will have profit on the demo & loss on live

The better your system, the less it is affected

IMHO, price driven hedging will have the least variation demo-to-live, spread sensitive scalping bots will show the most difference

With 100% indicator-based EA's somewhere in between

Good Luck

-BB-

I'm breathing normally. Just saying that brokers differ in how close their demo server is to their live one. Some have better demo servers than others... If it's important for YALEWANG then he should search for one of those. Specifically he can verify that:

1. All AccountXXX() functions give same results for live and demo server.

2. All MarketInfo(...) functions give similar market properties for live and demo server.

3. Run tick logger on both and verify similar characteristics (spread should also be logged and verified).


The last thing that can be done in order to have demos closer to live is to simulate requotes and slippage via code. It's not all that difficult to implement and can be very effective in bringing a demo account to behave more like a live one.

 
gordon wrote >>

The last thing that can be done in order to have demos closer to live is to simulate requotes and slippage via code.

It's not all that difficult to implement and can be very effective in bringing a demo account to behave more like a live one.

Thats a very neat idea! Have you simulated spread variations in as well?

I think though you may be being a little modest though with "It's not all that difficult to implement"!

Demo servers do indeed differ but the long-term trend is for the worse as more brokers use a shared/pooled data source to save (their) costs..

With micro accounts available from $200, I still say you cant beat the real thing

The final kicker is that no broker's feed will neccessarily stay the same, as more & more go more STP or (nearly) ECN..

So live will differ again from demo, even with some ingenious inline simulation of variations..

FWIW

-BB-

 
BarrowBoy:

Thats a very neat idea! Have you simulated spread variations in as well?

I think though you may be being a little modest though with "It's not all that difficult to implement"!

Demo servers do indeed differ but the long-term trend is for the worse as more brokers use a shared/pooled data source to save (their) costs..

With micro accounts available from $200, I still say you cant beat the real thing

The final kicker is that no broker's feed will neccessarily stay the same, as more & more go more STP or (nearly) ECN..

So live will differ again from demo, even with some ingenious inline simulation of variations..

FWIW

-BB-

Well, I am constantly on the search to make Tester and Demo runs as close as possible to a Live account. This is crucial since Tester and Demos are the only tools available to check a potential EA's behavior in Live without actually running it Live (which costs money).


In the Tester I have successfully simulated errors, requotes and time delays. I couldn't find a way to do it properly with spread variations, so that's still one limitation of the Tester. My methods are similar to those explained in the article Modelling Requotes in Tester and Expert Advisor Stability Analysis. The same methods can be used to simulate time delays, disconnections and generally any type of error. The basic idea being that skipping a tick will cause a time delay of the number of seconds between this tick and the next one.


In the Demo's there is no need to simulate spread variations since they are already there. But from my experience demos tend to have a lot less requotes or any kind of errors than a Live account, so what I do is simulate errors and requotes, but in the demo I have a lot more control over this than in the tester since I can actually inject mili-second delays to simulate these errors and delays. I usually collect timing and error statistics in a Live account while trading and try to bring the demos to have the same stats.


I agree though that brokers will likely have worse and worse Demo's in the future. Almost all big brokers now have several types of accounts which have big differences between them (sometimes even different data streams and liquidity providers), but still have one demo server for all clients. It sucks.

 
gordon:

Well, I am constantly on the search to make Tester and Demo runs as close as possible to a Live account. This is crucial since Tester and Demos are the only tools available to check a potential EA's behavior in Live without actually running it Live (which costs money).

You could run an EA live without costing you any money. Instead of executing the order send command replace it with a function that records all the market values. This can be as complicated as you like. By letting the code think it has a trade and dumping the exit conditions as well you can get close to real trading because you will see the spread and the entry price etc. I know it won't pick up problems like requotes and connection break but at least it will be using live data and get rid of some of the back tester and demo false trades.
 
Re-quotes is a huge problem that can't be ignored. But there are also other errors that are much more prevalent in Live than Demo. Not opening or closing orders also removed the fill time and general delay problems that occur when trading Live. I think while this method solves some problems, it introduces new problems caused by not actually trading. I guess every solution has it's ups and downs.