Looking for high quality tick data for MT4 (broker:IBFX) | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Tick Size, Tick Rate, Tick Spread, and Tick Logger indicators 36 replies

Indicator Tick Chart Vs. Broker Tick Chart 0 replies

Tick By Tick M1 MT4 History 2 replies

Need Tick data for EURUSD (99% quality on MT4 build 409) 1 reply

Quality data feed & Tick chart Questions 2 replies

Attachments: Looking for high quality tick data for MT4 (broker:IBFX)
Exit Attachments

Looking for high quality tick data for MT4 (broker:IBFX)

  • #1
  • First Post: May 11, 2009 4:57pm May 11, 2009 4:57pm
I'm looking for high quality tick data to thoroughly back test the EA I created. I've done a forward test since April 1st and so far it looks awesome. No losing trades... ever! Although some people may find it risky since there is no stop loss involved and it's determining trades off the 5min chart for GJ.

Trading results attached... account started at $10,000 and since april 1st has gone to about $16,000

I'm looking for high quality data only because the backtesting I've done has a modeling quality of 41% and one trade (a losing one at that) was executed when I think it shouldn't have. Anyway, I need better data.

Once I get a good back test done I'll edit this thread to tell you guys about my system.
Attached Image
  • #2
  • May 11, 2009 4:59pm May 11, 2009 4:59pm
Hard to get free clean tick data. Be prepared to pay a pretty penny for that data.
  • #3
  • May 11, 2009 5:39pm May 11, 2009 5:39pm
I got a couple requests for a long term graph... here's a back test since November 25, 2008 with no stop loss. Although there are no losing trades, some of the potential downdraws are huge, but the data is not accurate and many of those trades should not have occured.

Start value: $10,000
November 25, 2008 ----> May 7, 2009
End Value: $34,813.35

41.78% Modeling Quality
Total trades103

also note... I only let one trade happen at a time (for all data shown so far)
Attached Image
  • #4
  • Edited 9:11pm May 11, 2009 8:29pm | Edited 9:11pm
NEVER!! trust less than 90% Modeling Quality - Download a copy of the http://Alpari-idc.com and their Databank info and create a 2nd EA testing environment that you can trust. Here are a couple way to do it. I prefer method #1 though.

http://alansforexblog.com/2008/02/27...pert-advisors/
http://alansforexblog.com/2008/04/16...rategy-tester/

NICE PDF of Method #1 - http://tinyurl.com/mt490percent

Also holding trades till they pay off - WILL GET YOUR ACCOUNT KILLED!!!
  • #5
  • May 11, 2009 8:49pm May 11, 2009 8:49pm
Quoting Fulltime247
NEVER!! trust less than 90% Modeling Quality - Download a copy of the http://Alpari-idc.com and their Databank info and create a 2nd EA testing environment that you can trust. Here are a couple way to do it. I prefer method #1 though.

http://alansforexblog.com/2008/02/27...pert-advisors/
http://alansforexblog.com/2008/04/16...rategy-tester/

[b]NICE PFD of Method #1 - http://tinyurl.com/mt490percent

Also...
sweeeet! Just what I needed Thanks!

Quoting Fulltime247
NEVER!! trust less than 90% Modeling Quality
...
Also holding trades till they pay off - WILL GET YOUR ACCOUNT KILLED!!!
I've done backtests since March with about 91% modeling quality with the same results as my live test. No S/L, very good risk %, and no losing trades. The maximum -pips from all the back tests done that's ever been reached so far has been -300 with the average around -100, but they always turn around. Oddly enough those only happen around the oldest 1M data that my charts have which is around November/December of 2008 otherwise all trades go green. I think you can trade fine with no S/L as long as your risk management is good.
  • #6
  • May 11, 2009 9:04pm May 11, 2009 9:04pm
Quoting MoForce
I've done backtests since March with about 91% modeling quality...
91%!!!! - How the Heck did you do that!!! - The MAX that MT4 will report is 90%. - Also I get that the Hold & Hope strategy works most of the time but during the financial collapse there were periods of 1000+ pip surges. That would wipe out all gains. - I've tried those methods but usually turns greater than 1xx pips means the market has turned so I usually hop off and take a fresh look at the market. But whatever works...
  • #7
  • May 11, 2009 9:16pm May 11, 2009 9:16pm
If you can't get the Databank data (site was down last time I tried), try this source. Its .csv but works the same.

http://forextester.com/datasources.html
  • #8
  • May 11, 2009 9:27pm May 11, 2009 9:27pm
Quoting Fulltime247
91%!!!! - How the Heck did you do that!!! - The MAX that MT4 will report is 90%. - Also I get that the Hold & Hope strategy works most of the time but during the financial collapse there were periods of 1000+ pip surges. That would wipe out all gains. - I've tried those methods but usually turns greater than 1xx pips means the market has turned so I usually hop off and take a fresh look at the market. But whatever works...
lol I ran the test again and yeah... 90.00% is what I got. For some reason I remembered 91

I'll definitely check up on it often... especially during big movements like a 1000pip surge. I've noticed a nice pattern with about 100pip S/L. For the balance graph you get a step type pattern where the general balance direction is up with some small (100pip) drops every now and then. Wouldn't that mean it should be easy to find what's in common with trades during those times and modify the EA to avoid those conditions?
Attached Image
  • #9
  • May 11, 2009 9:28pm May 11, 2009 9:28pm
Quoting Fulltime247
If you can't get the Databank data (site was down last time I tried), try this source. Its .csv but works the same.

http://forextester.com/datasources.html
since 2001! amazing! Great find Thanks again... and yeah Databank wasn't working
  • #10
  • May 12, 2009 11:08pm May 12, 2009 11:08pm
Quoting Fulltime247
If you can't get the Databank data (site was down last time I tried), try this source. Its .csv but works the same.

http://forextester.com/datasources.html
I've tried multiple times with this data and Method #1 and I keep getting OrderSend error 131 which is a volume error. Are there any ways around that? A back test since 2001 would be so awesome! The error happens immediately after start.
  • #11
  • May 12, 2009 11:16pm May 12, 2009 11:16pm
Quoting MoForce
I've tried multiple times with this data and Method #1 and I keep getting OrderSend error 131 which is a volume error. Are there any ways around that? A back test since 2001 would be so awesome! The error happens immediately after start.
boy do I feel dumb =P it was just a simple # of lots error... back test since 2001 looks pretty good btw
  • #12
  • May 29, 2009 10:21am May 29, 2009 10:21am
Does M1 history data from Alpari data-center contains tick data?
i thought M1 history only saves 1 prices data per minute
  • #13
  • May 29, 2009 12:38pm May 29, 2009 12:38pm
Quoting AdamN
Does M1 history data from Alpari data-center contains tick data?
i thought M1 history only saves 1 prices data per minute
No.

When you test with tick data in MT4 it first runs through the M1 data.
Since it knows the O/H/L/C of each 1 min period and the number of ticks (volume) that ocurred in that minute, it then creates 'volume' number of pseudo random (modeled) ticks that lie within the confines of the O/H/L/C.

I say pseudo random beacause they are not simply random values within the confines. They have developed an algo to create a convincing progress from open to close. These ticks may be nothing like the realtime price action during that minute but since they adhere to the actual O/H/L/C, and since the TF is so low, most EAs wont know the difference.
  • #14
  • May 29, 2009 9:41pm May 29, 2009 9:41pm
Quoting rangebound
No.

When you test with tick data in MT4 it first runs through the M1 data.
Since it knows the O/H/L/C of each 1 min period and the number of ticks (volume) that ocurred in that minute, it then creates 'volume' number of pseudo random (modeled) ticks that lie within the confines of the O/H/L/C.

I say pseudo random beacause they are not simply random values within the confines. They have developed an algo to create a convincing progress from open to close. These ticks may be nothing like the realtime price action during that minute but since...
I see, thanx for the explaination
So we can still backtest Tick-based EAs but with less accuracy
  • #15
  • Last Post: May 31, 2009 10:10pm May 31, 2009 10:10pm
Hmm, well the easy answer is : yes and no

If you use the strategy tester settings

Model : Every tick
Period : M1

Then you get what i described above, however due to the limitations of MT4 you are stuck creating EAs that are based on, and calculate solely from the M1 TF. If thats the TF that your trying to create an EA for then I'd be confident that it wouldnt notice the difference between real and modeled ticks. AS LONG AS your EA has no interest in synthetically created TFs lower than M1.

If you up the 'Period' setting to say H1 then its been my experience that the ticks are not modeled in the same manner. MT4 creates a cut down version of ticks and hence acurracy is lost.

Unfortunately the strategy tester does not enable users to setup and create modeled ticks from M1 data, but test the EA on a H1 TF. Nor does it allow you to easily use multiple TFs in your EA code.
Its a real shame the user cant supply their own FXT (tick data) file either but c'est la vie.

These limitations can be circumvented but require extensive coding to do so, and I wouldnt like to try to do it all in native MQL, rather with the use of a custom built DLL to help out here and there.

Maybe MT5 will address these issues, I here talk that you'll be able to use MTFs in the strategy tester so we will have to wait and see.
0 traders viewing now