EA TRADES OUTSIDE THE BAR in backtesting.

[[Deleted]]

Hello,

I have seen similar behaviour in articles posted here, but not the same behaviour (please see attached picture).

The ea produces one trade per bar, and the quality of the backtest results are about 89% to 89.7%. I am using a demo account in CROWNFOREX, loaded with HISTORICAL M1 data from ALPARI.

The ea is producing trades outside the bar, in the "ether".

Can you please let me know if this is as a result of:

1) an error in the ea's algorithm?
2) or faulty/incomplete/inappropriate historical data?
3) or a "feature" of the backtester?
4) or a "feature" of the broker?
5) all of the above
6) none of the above

The more detail you can provide, the better...


Thanks!!

NH

[Jimmy]

nothere:

Hello,

I have seen similar behaviour in articles posted here, but not the same behaviour (please see attached picture).

The ea produces one trade per bar, and the quality of the backtest results are about 89% to 89.7%. I am using a demo account in CROWNFOREX, loaded with HISTORICAL M1 data from ALPARI.

The backtest uses data from Alpari, then the result is shown in chart from CrownForex.
The data are never the same as so most probably their timezones.

[[Deleted]]

fireflies:

nothere:

Hello,

I have seen similar behaviour in articles posted here, but not the same behaviour (please see attached picture).

The ea produces one trade per bar, and the quality of the backtest results are about 89% to 89.7%. I am using a demo account in CROWNFOREX, loaded with HISTORICAL M1 data from ALPARI.

The backtest uses data from Alpari, then the result is shown in chart from CrownForex.
The data are never the same as so most probably their timezones.

Hi Fireflies:

Yes, I understand your comment, but I still do not think the difference between brokers is that much, in such case the difference in size and data between brokers would be about 40-60 pips (or more, as per the picture)....I think that would be too much of a difference.

Perhaps you are right, perhaps the ALPARI data can and should only be used on ALPARI MT4 client/connection.

Any other comments from the guys at MQL??

Thanks!!

[Rashid Umarov]

Alpari and CrownForex have different time shift in rates. It is obviously.

[[Deleted]]

Rosh:
Alpari and CrownForex have different time shift in rates. It is obviously.

Hi Rosh,

Thanks.

If the only difference is the time shift in teh data, how can I compensate for this? How can I use ALPARI's data in CROWNFOREX?

It may be that the only thing to do is to capture CROWNFOREX's data for use later. Is this the case?

Thanks to all.

[zap]

Anyways, this is a typical example of another issue too:
Opening and closing on the same candles, which probably won't be the same in realtime trading.
This pattern I see would indicate to me, that this EA is not suitable for M1 data testing.

If you run the backtester in visual mode, it will draw the appropriate chart. Backtester visual mode gets the data from and shows charts upon its .fxt files, the non-visual mode with show chart after testing gets the data from .fxt files and shows charts upon the active broker's .hst file charts.

[Rashid Umarov]

Thaks, Zap. My English don't allow me to give same quite answer.

[Slava]

Described in the article 'Testing of Expert Advisors in the MetaTrader 4 Client Terminal: An Outward Glance'

[[Deleted]]

stringo:

Described in the article 'Testing of Expert Advisors in the MetaTrader 4 Client Terminal: An Outward Glance'

Hi:

Thanks, but.....not necessarily. As I mentioned in my first post, there are articles in this forum that describe similar behaviours, but not the same behaviour.

It seems to me that the only way to test an EA in CRONWFOREX, is to use CROWNFOREX historical (LIVE) data, and since they do not provide it, I will have to start saving it via tools provided at this forum and others.

Too bad some projects in other forums to gather the historical data form specific brokers has not started yet.

At least we have the METATRADER data which we can download now (starting I think on version 200), but where does the data come from?

Thanks to all.

[Slava]

Third, and this is the most important point, one user can have accounts with different brokers. And different brokers provide not only different trade conditions (please be reminded that the FXT-file header contains very important information to simulate the trade server work), but also historical data volume and quality. In this case, after switching to another account at testing, it is highly recommended to recalculate data.

[zap]

Too bad some projects in other forums to gather the historical data form specific brokers has not started yet.

Contact me at forexzap<at>gmail<dot>com if you are interested in historical tick data, with correct settings, I guess you may need it.