Mystery results on the same EA from different brokers!!!

i dunno abt historical data but for demo servers all these mentioned brokers, the results can be differerent. FXCM keeps giving me invalid stops error and refuse to execute orders but the other brokers be they 5-digit or 4-digit has no problems executing them; and the funny thing was i even printed out the order entries, sl, tp and they are fine! ie no invalid

There are lots of differences between brokers. You should not assume that any EA will work the same, let alone at all from one broker to the next.

Look at lot size rules, spreads, stoplevel, and freezelevel, just to name a few. And make sure you really have the history data that you think you have. This is probably only half the list of potential gatchas, but it's a start.

- Tovan

I was just reading an article from my VPS Server company which may shed some light.

LATENCY!

They have a server farm in California and one in NYC. They ran a test on several EAs placing trades simultaneously to the same broker, but using their servers on opposite coasts.

The results were amazing. Due to the difference in latency, and ONLY that difference, the VPS Server which was closest to the Broker Server was 3X GREATER than the VPS server a few thousand miles away.

Bottom line, if you're serious about trading and results, pick a quality VPS as close as possible to your Broker's Server.

Hope that helps. Apparently, milliseconds count in terms of fills.

Regards,

Merlin

Yes, this is another important thing to consider when choosing a broker. Good thing to keep in mind.

In this case, however latency is not a factor since back-tests don't check for that. This is another weakness of the Strategy Tester that must be considered when transitioning from back-testing to live-trading.

- Tovan

@lamass:

https://en.wikipedia.org/wiki/Emphasis_(typography)

CB

I think I know which VPS provider you're talking about, and I argued with him about this article when it was first published. To my eyes his data was saying that signal pick-up and profitability was random across the two sites, and not a feature of latency. He's a very nice guy, but it's worth remembering that he had a brand new datacentre in NYC which he wanted to fill.

Depends on the sort of service you're getting from your broker. Milliseconds only count if (a) your trading system is sensitive to very precise fills, and also (b) your broker is filling trades on a pure FIFO basis. This definitely isn't the case if your orders are being warehoused, and it's not necessarily true under other circumstances.

In turn, this leads to other problems. If your trading system is sensitive to very precise fills, then I don't believe that either MT4's strategy tester or demo-account forward testing is going to give you a true picture of system performance. On all brokers, fills on demo accounts are materially different to fills on live accounts. That's one of the reasons why many of the commercial scalping EAs make pots of money on paper but lose money in live trading.

If your system benefits from ultra-fast fills, then MT4 is probably the wrong platform. You should probably be using something like Hotspot who quote 16ms for order-placement.