I guess it really depends on how the EA trades. If it scalps (as I assume it does), then it goes withotu saying, having a good broker that allows it and has low spreads to boot will be a must. It's not so bad when scalping the EURUSD as most brokers have low spreads on it... but when you start talking about the likes of the EURCHF, and USDCAD scalping, then low spreads are a must. I can't think of a good broker for scalping, but I hear that Alpari, and perhaps FXCM could be worthwhile. FXDD is not good for scalping at all, spreads are too high and are variable - you're better off trading directionally with them.
...and then there are all the bucket shops. From what I hear, and if you google around the net FXOpen has earned the rep of bucket shop. As soon as you're doing too well - somethign bad happens.
For sure, an EA that is there for the long-haul is the goal. We mainly want trading bots that can handle and adjust to market conditions - which is why I'm impressed with the FXShocker's backtests. Not only does it survive on it, but it thrives on it. You can pretty much optimize any free EA to survive 10 years in the market, but it won't be very profitable over those 10 years... so to have one that can not only survive but hit millions from a small deposit is certainly closing in on holy-grail status. The only question is, did they optimize the bot to simply survive the 10 years? you never know how it will survive the next 10.....
For sure, you're right about withdrawing your initial deposit as soon as possible. Need to trade for free in case the worst happens and the account blows up.
In any case, if I were to get this EA, I would only deposit $1000 on a live mini account and then just let the EA run all the way with it. If it can make 13 million in ten years on one pair. 1 million from a $1000 start up should be good enough over the long haul