Current List of Brokers:
Broker,Server Type, Data Type, Pairs Covered
IBFX Live, Demo Tick, M1 EURUSD, GBPUSD, USDCHF, USDJPY, AUDUSD, USDCAD
CFG Live, Demo Tick EURUSD, GBPUSD, USDCHF, USDJPY, AUDUSD, USDCAD
FastBrokers Demo Tick, M1 EURUSD, GBPUSD, USDCHF, USDJPY, AUDUSD, USDCAD
TradeX Demo Tick,M1 EURUSD, GBPUSD, USDCHF, USDJPY, AUDUSD, USDCAD
MIG Demo Tick GBPUSD
For full details see: http://www.cubesteak.net/mt4-multi-broker-tick-and-m1-data-project/covered-brokers-and-pairs/
I want to announce a new project that was born out of the Yahoo MT_E_and_I group that I think all of you back testers will love.
Anyone who has back tested EA's for a while has probably used or at least heard of the Alpari databank, where M1 data from that broker is available all the way back to 2004. It has been very useful in back testing to achieve 90% modeling quality for back testing of EA’s.
The problem with this is that many people don’t use Alpari as their broker, nor do they wish too. And, unfortunately, it makes a difference in accuracy if you test with their data and trade on a different broker.
The vision of this new project is to create a multi broker data bank that not only has M1 data, but also real TICKdata to allow 99% quality back testing!
The purpose of my announcement is two-fold:
1) To let everyone know that back testing with MT4 is about to get a whole lot better
2) To ask for some recruits to help in the gathering of this data. So far we have just a few folks participating, as the project is VERY new. We don't ask a lot and anyone can help - its very easy to participate.
If the idea of being able to have 99% quality back tests for YOUR broker at all appeals to you, please consider lending a helping hand to make that a reality.
Please go to:
to read more. How to join us is on the site, but you can also PM me or email at cubesteak at gmail.com
No offence, but what is the point of doing that? I mean, you do now that from next Build onward (Build 199, should be introduced this week, if all goes as planned), MetaQuotes is going to launch a service called MetaTrader History Center, using which you will be able to get quality tick data off a website, I believe off the: hXXp://history.metaquotes.net/.
I mean, it is always great to have as many alternatives as possible. Well, I basically just wanted to make sure you know about this History Center thing. Again, in general its a great initiative!
I am not promoting this project.
Just want to say about the data.
Every broker is having its own data. Usually we are backtesting using M1 from Alpari broker. I will allow us to backtest EA back to 2001 with 90% modeling quality. But it may be very good to have data from other brokers (M1) for atleast 3 years. It will help a lot.
to understand the problem probable you need to read Codersguru article http://www.metatrader.info/node/67
It is real problem: if we need to backtest EA we want to do it back for at least 3 year. Right? Because market is changing ... Just to see how it works in different market. But we have alpari broker data only. And alpari data and for example ibfx data are very different. Very different.
Systems and EAs developed using alpary data (i am talking about 3 or 4 years data allowing to backtest with 90% modeling quality) will be very different for ibfx broker. Sometimes different concerning pre-set files only (settings of EAs). Sometimes it is necessary to re-code the system or re-evaluate it.
It may be the same for traders trading on 1 timeframe for example. But it is different for intra-day traders and EAs.
So if somebody will collect M1 data for main brokes so it will very good.
Thanks for this info - Where are you getting the info about bld 199? I wasn't aware that MQ shared info about next releases... And, does it make the data available from different brokers?
I'd love a URL with info about this service...
Thanks for the info!
PS - thanks too newdigital for the above reply...
1. The info on the upcoming service center was released on the official MQL4 forum. Again, we are expecting the 199 to be released some time this week. Hopefully.
2. No, the data won't be from any particular broker, they are saying that the data was taken from many BANKs and was filtered out properly. The data will of of highest quality. They mentioned 90%+ quality of modelling for periods from 1999+. Now the interesting thing is that as soon as you recieve the data off that service, MT will automatically built all higher timeframes data, meaning that you need to download tick data once and then MT will create data for all other time-frames (no need to download data for every timeframe >>> say oohhh weee!). Sounds really exiting, to say the least.
And again, as I mentioned your idea/ project is really great. I just wanted to let you know about the MetaQuotes initiative so you think twice before putting TOO much effort into your project. Again, great work mate! : )))
Ah, thanks for the explanation! It will be interesting to see how the 2 sets of data compare... I wish they would do it for each of their licensees! I still think that broker specific data will be the most accurate way to tell how your EA will work when trading with THAT broker...
I agree with you about broker specific data. That is the best way to back test. Combining multiple sources usually removes or softens spikes, that are all too common and is a fact of life that needs to be considered in backtesting. Those who are skeptical, I am requesting them to watch the feeds on this coming NFP Friday, during the release. Keep the 1M charts for GBPUSD for two or three brokers open in front of you and see how they differ. Market movements are mesmerizing and the difference is eye poping.
While ANY tick data will give over 90% accuracy in the modeling quality, what good does that do you if the tick data isn't representative of what you can actually trade against?
I think this is a great move forward for MQ to make (providing tick data) - as it will allow a lot of folks to get a much better sense of their EA's performance. Regardless of where it came from, it HAS to be better than fractal interpretation.
But, I still think that broker specific data is the real goldmine. My hope is that brokers will follow suit and make their own tick data available as well now that MQ is going in that direction. Now THAT would be a win...
I just wonder if they really want the data to be published....
If you want to backtest scalper EAs with any confidence, then ONLY THE TICK DATA FROM YOUR LIVE SERVER will be good enough. This was brought home to me even more when I trialed some tick data from disktrading.com. I'm sure their data is very accurate, but their EURUSD record for 1 week contained, no kidding, 25 times the number of ticks that I have recorded off FXDD.
When I backtested CyberiaTrader with 99% modelling on a EURUSD1_0.fxt (1M timeframe) file derived from the diskrading file I saw a gold mine: continuous profits, 1.5% drawdown. The FXDD equivalent didn't even make a profit.
Recording tick data from your live server opens up a whole new world of optimisation that previously took months of forward testing to complete. Build 199 of MT4 will not match that.
Interesting post. Some time ago I was curious enough about the quality of the M1 Alpari data to write a script that checks the continunity of the data - ie. looks for missing bars. Scary result. Running it on 763714 bars of USDCHF M1 data shows that there are 384388 missing bars - that's over 50% for the time period involved!!! Checking M5 data that has been generated from the M1 data reveals 3156 bars missing from the original 173532 - about 1.8%. In case anyone is wondering, I accounted for the weekends but not for Forex holidays when the market is closed (are there any?). Note also that this is only checking the price data continunity, it says nothing about the price data itself.