Quote from Good1:
I'm reminded of Nicolas Darvis' method. It was set up to operate somewhat unattended, while he was out dancing around the world. If i recall, its ability to operate relatively unattended was part of his key to success. For many of us, our attendance can easily translate into meddling and interference. In my experience, what i've been able to automate did better than when i was not attending (meddling with) it. The weakest link with unattended, imo, is one's internet connection, whether it is sturdy (always on). This would be the prime reason for attending closely to any robot that is already programmed and tested well. It seems to me any robot with a robust connection ought to be able to go on it's own for at least 24 hours. I'm thrilled by those that can go a long time without intervention, and will be watching the MQ5 ATC very closely this year. Another time for close attendance would be the when a successfully demo'd robot is making the transition from demo to the unknowns of a live mode. Another time for closer attendance would be when raising the level of leverage (volume of lots) toward a higher risk% of equity per trade. Watching the screen hurts my eyes anymore, so i wouldn't be considering any of this unless there was the possibility of going a week at a time without looking at a screen. All the better if each week was likely to yield a gain, even if a small %.
I would never advise full live automation from a home computer, unless you plan to have multiple internet connections and power sources. Use a VPS. you can get a reliable one from CNS for about $30. When auto-trading live, do your best to have your computer physically as close to the broker's server as possible (latency).
Manual trading, or testing could be done, but I suggest final testing be done on the very platform that you will be using. VPS are very common and cost effective for this approach.
Quote from Good1:
You seem to suggest that MT4 robots are not as robust as other types of robots, or that MT4 brokers are able and willing to pull more than average shenanigans. Please elaborate.
I understand there can be a big difference between a demo account that seems to do well, and a live account where a broker is taking the other side of your trade, and has ways to make sure they don't lose in the long run. As noted, this is where i would attend closely.
I suggest opening three accounts when going live, run the same trades through all of them, and monitor the bottom line. Stay with the one that least handicaps your results, cut the two handicaps, and add two more for monitoring. Keep this process going, and let the robots go unattended as they (and your internet connection) prove relatively reliable. You could have the robot send emails on a regular basis. If it misses an appointed mail time, it could mean that it has lost connection. If it's connected and still trading, it shouldn't be difficult to monitor account equity through automated email, allowing for a degree of unattended automation.
Reputable brokers typically do not have this issue. As long as the robot itself is configured properly, then you can work on other issues like broker fills. And you begin to tweak the robot to handle the different exceptions that come up (there's not that many)
When I was testing strategies, I had 20-30 demos running at a time at one point, forward testing different settings. This is impossible to do with manual trading (you can only click so much and watch a screen for so long). Scalability is the biggest advantage with automation. I cannot overemphasize the importance of a systems approach to profitability. You can always tweak the system along the way.