The unfortunate truth is brokers are only required to return you money after clearing within 3 days and usually they purposefully keep it for that 3 days. The reason is that they can trade with your money for those 3 days free and clear of interest, then give it back to you afterwards. Oh and ignore cold. I am not sure if he was born an arrogant asshat or if he just thinks acting that way will make his e-peen look bigger.
Error was so stupid (or so classic) that I'm too ashamed to tell you. Oh, what the heck: I used bid in place of ask and ask in place of bid But if I use prices of real trades instead of bid/ask to project my buy/sell price, it makes only about 0.1% a day now
***EDIT: I see you found a different issue, but I'll leave my comment here just for good measure. **** Just be warned that if you're system is going for 1 or two ticks in a demo mode it's very likely the fill is "at touched" as most of the demosystems are which means... instead of getting at the end of the line, you're at the front of the line. Just be careful about knowing where that is. Because, if you are "at touched" you need to calculate based on passing to the next price up (or down) to calculate your % success and then run that % success number against your profit ratio for a realistic estimate.
No, it's not outstanding. First of all, this is max number. Min is close to 0. Second, 0.1% for $30K start capital is only $30 a day. Of course, it's better than nothing, but not that good to count on it either. But the funniest thing is this: I really was only testing ordering subsystem of my program (just finished that part last week), and I needed just some test algo to issue buy/sell commands to the ordering part. So I briefly coded in Mersenne-Twister randomizer plus some very basic logic (like if it goes up, keep it, if it goes down, sell it immediately, plus some other checks). I was shocked to see the results. After I found an error I'm actually still shocked. In my understanding random logic should (on average) lose money (because of fees), not gain it.
One of the problems is assuming execution prices. In real-time execution will not be as accurate as your model. I have been there/done that. Start small and make sure your exceution platform has small slippage as possible. Order routing is much more different in real-tie that in simulation mode.