Very low quality historical data issue of broker!

 
Hi,
On live account with MT5 while running EA tests I should be able to access all historical data on the server with "99% quality" as soon as the broker connects to the trading server. (At least I know it is.)

But, in backtesting with major parities, I encounter "very low and unusable historical data quality of 10-30%" !  
Below you can see the "extremely low historical data rates" provided by the xxx MT5 live server in the backtests of the "EURUSD" and "GBPUSD" major parities !

I have reported the problem to my broker and have yet to receive a response. I suspect I missed an important detail about this that I should know or do in MT5.
I would be glad if you share your experience and suggestions about the quality ratios of historical data obtained from your live servers while performing backtests (?)
Best.



 
Edviao:
Hi,
On live account with MT5 while running EA tests I should be able to access all historical data on the server with "99% quality" as soon as the broker connects to the trading server. (At least I know it is.)

But, in backtesting with major parities, I encounter "very low and unusable historical data quality of 10-30%" !  
Below you can see the "extremely low historical data rates" provided by the xxx MT5 live server in the backtests of the "EURUSD" and "GBPUSD" major parities !

I have reported the problem to my broker and have yet to receive a response. I suspect I missed an important detail about this that I should know or do in MT5.
I would be glad if you share your experience and suggestions about the quality ratios of historical data obtained from your live servers while performing backtests (?)
Best.



Don't overestimate this percentage. In MT4 it can be written by a program providing test quotes. In MT5 (I think and don't know for sure) it depends on what you have selected: every tick, 1 min OHLC, only open,....

If you are not sure about the quotes check for holes. On the other hand if your EA is able to manage holes or interruptions it is more robust.

 
you can find a better broker or buy data of your own
 
Pak Hong Poon #:
you can find a better broker or buy data of your own

Unfortunately this is a serious problem that is hard to solve. 
I've been looking for a reliable broker for about three years but haven't discovered it yet!

 
Carl Schreiber #:

Don't overestimate this percentage. In MT4 it can be written by a program providing test quotes. In MT5 (I think and don't know for sure) it depends on what you have selected: every tick, 1 min OHLC, only open,....

If you are not sure about the quotes check for holes. On the other hand if your EA is able to manage holes or interruptions it is more robust.

Don't underestimate it !

It doesn't depends of the selected mode (EDIT: it should not) and only means a very poor history data which should not be used at all for serious backtests.

Testing and debugging, or having a robust EA is something else entirely.

 
Alain Verleyen #:

Don't underestimate it !

It doesn't depends of the selected mode and only means a very poor history data which should not be used at all for serious backtests.

Testing and debugging, or having a robust EA is something else entirely.

I'm of the same opinion.

 
When I ran the test with "Every tick based on real ticks" I got 99% quality data.
But when I do the same test with "1 minute OHCL", the quality drops to 19%.

I think the historical data presented in the broker is in tick data format, not "M1" format.



 

Never use broker data.

Import data with Tickstory or similar. (Google for it)

 
Bernhard Schweigert #:

Never use broker data.

Import data with Tickstory or similar. (Google for it)

Frankly, this is the first time I've heard from you that google provides data for mt5 (!) 

Where can I access google based history datas?

PS; I have read some comments that there is unreliable datas for Tickstory. If you have alternative 3rd party application suggestions, I would appreciate if you share them.
 

Import data with Tickstory or similar. (Google for it)


I think @Bernhard Schweigert meant to find in Google more information about this topic



Reason: