[performance question] How accurate is the volume data offered by brokers?

Mt4 volume is tick-volume. As such it can range wildly from broker-to-broker and from data-to-data. Brokers will usually have accurate tick-count (aka volume) in real-time for their tick-sequence.

Hello Uzben, thank you for answering again.

I think I understand. So I should look for a big broker, with lots of clients, so it's volume is a better statistical representation of the real market?

Very, very, very bad is better than extraordinarily bad.

Volume is really a bad name for what "volume" in MT4 means . . . TickCount is what it should really be called. If you can make the TickCount mean something for you then that is great . . . just don't think you are looking at anything that can reliably mean volume.

Yes, I mean ticks! Price change / divided by volume (aka tick count) indicates the willingness of the market to move the price. I use that heavily on my trading decisions and it is proving to be extremely useful :-) That is, how much is the price affected by each transaction on the market. It is extraordinarily useful but only works if the broker has enough market share to make that data representative of the whole market sentiment.

That is why I need a huge broker.... :-) Which one would be the best? (Besides direct market access)

statistical representation of the real market. Imo, there's no such thing. It's a nice way to think about it but broker prices serve their own purposes. Unfortunately, we cannot talk about brokers (this forum rule could have changed), also I don't want to open a jar-of-worms which floods your post with spammers and promoters.

So I should look for a big broker. Perhaps, but not because you think their price-data closely relates to real-markets. It does-not. One reason I cannot recommend large brokers is because (in my experience) allot of them have variable-spreads.

However, nice that you like volumes and it's working for you, I'll just caution you to-be-careful when using volumes within the back-tester until you fully understand all the nuances effect real-markets vs back-testing. Some of the things which effect this particular example are

a) Real-trading can miss ticks. example u can miss volume[1] but that never happens in the tester.

b) Brokers can go back and modify the historical data by merging their data servers and going with the highest volume on a particular server....something which would have been different when you traded.

c) The historical-data you download usually may not come from your particular broker you use in the future. It'll be impractical to download new broker's data and go through the system-testing process every time you switch brokers.

d) Variance in the Spreads. Tho most would consider variable spreads as the norm, most of us retail traders systems cannot survive spreads above single digits (example 9). Doesn't matter how accurate/in-accurate this broker is reflecting the real-deal we just cant trade. This does-not show up in back-testing.

In Conclusion. IMO, if you're going to employ tick-volumes then perhaps you should stick to testing your systems on Live-Data. Along the lines of "Volume" within the traditional sense, I'll recommend doing a google search for "Market Depth" indicators for mt4. <-- Of course like tick-volume testing, it's limited to Live data.