Right. Most of the things you (the OP) have listed: <i>Neural Network+ Genetic Algorithm + Simulated Annealing value-at-risk calculations ... Detailed charts with numerous indicators.</i> are totally <b>useless</b> for trading and to make $$$. (They can be useful in other fields) Let me point you to an example of statistical <a href="http://www.datatime.eu/public/gbot/Strats%20G-BOT/Strategy_CT_T_LS_E/Strategy_CT_T_LS_E.htm" > performances</a> (improvable) obtained with a pure "mechanical" approach, just to debunk the illusion that "prediction" is meaningful in an automated context. Tom
really? what are doing for those 10000 hours? Just sitting there? I expect that at least you would gain basic programming and data analysis skills.
It's certainly possible one could gain basic -- or even quite advanced -- programming skills through the process. In addition, if someone takes the time to learn about e.g. statistics (by reading a real book on the topic -- NOT one designed for traders) or linear algebra or analysis, then yes, that would result in bona fide learning/skills. But I strongly doubt that's typical of those who attempt trading on their own. Looking at "charts" and learning about "TA" is in my opinion completely wasted time, unless it leads you to real, consistent, nonrandom results.
Take a look at tradelink, JSystemTrader, marketCetera, tickzoom etc. Why spend a year or more building a platform when you could be building strategies.