over the last 3 years i've tried to use - Ninja Trader - Strategy Runner - Trading Technologies - CQG - MetaTrader - MultiCharts for automated trading. None of these were even able to execute and manage orders correctly. Only Tradestation was able to do things right. Of course, i was always told by various people, how TS is bad. And how CQG/ TT are "professional". F#ck others opinions. Try everything for yourself.
I will always choose simple moving average based system over any NN. I'm happy to see so many other traders using "professional" overcomplicated and usually full of sh!t methods, so they could lose their money for me.
Maybe you should come down to earth (Mars is cold this time of year) and stop talking 300% trash. The only way you will make anything approaching that is using crazy leverage, or getting very lucky. If you weren't so willing to believe numbers like that, you would have found the bug in your code a lot faster.
You are underestimating his capabilities - he is using forward-looking neural networks. Even more is possible!
Ha, couldn't agree more! Keep it simple guys! Money is made through understanding another traders behavior AND THEN quantifying that behavior. FWIW, take this from a guy who dropped out of a PhD program in engineering/fuzzy logic: Any NN is only as good as its initial rule set! Unsupervised learning algos with no initial rule sets are only pratical in very specific cases. While NN's have applications in finance, the time and effort in constructing and maintaining an NN usually outweighs the potential reward.
Figured I would get beat up some here. It was a big mistake.. however, being 24 hours past the issue, several networks are now trained and the results are looking good. It will not be the insane results I showed previously, but I would not outright dismiss hitting the 300% goal I set forth in this journal.
Do the math. Thousands of Folks 1000x smarter than you doing the same analysis These folks have 10000000x the bankroll you do If there was anything to be found, they would have already found it, and reduced its profitability to much less than 20%. Be glad if you find anything at all that works reliably, no less 300% return. Especially since you're mining where the big boys play. If you said you were trading Canadian Warrants I might give you a shot.
Those equity curves you posted earlier looked way too good to be true. Regardless, good luck on your patched up code. It will be interesting to see the results.
That goes with the territory. It`s ET, you know I for one will be following this journal, so please don`t be discouraged by the expected negativity and trolling. The ignore function can be very useful.