darthtrader2.0

Well, I'm not really sure what you would want as far as an example. It basically gives a primer in bias/heuristics, frequentist statistical analysis, monte carlo vs the bootstrap as far as generating a random system to judge your system against..Its all theoretic though and only scratches the surface..your not really going to be able to do anything with the stuff in this book without alot of heavy lifting on the ideas presented. Its really just a bridge between something like John Murphy's books and the quant world.

I think you need to take the stance that if you want to automate things, then commit fully to it and it will take as long as it takes. Just like if your going to trade discretionary, you need to put in the massive hours of screen time, commit to it 100% and there is no way around that. To me what you need to focus on is what the brain is not good at. Your not going to beat a brain watching one market for 5 years trading simple patterns. There is no algorithm that can match the brain as far as pattern recognition in that sense. However, the brain can't watch a 100 markets at once and scan each market for a 100 statistically valid patterns then excute on those patterns like a machine can. I've just started down this path also and if it takes 25 years of research and a phd in stochasic modelling then that is what it takes.

There is a great quote from John von Neumann that applies to this. "Precisely describe to me what it is a machine can not do and I will make it do exactly that". To me the issue with retail auto trading is people are taking what discretionary traders are using(fibs, MP, patterns, candles, whatever), automate the simple patterns, curve fit the data, run it live and pray you get lucky that these simple patterns match up with the present market conditions. All the time not thinking about that what makes a successfull discretionary trader is the catalog of patterns in their brain for making decissions under uncertainty while the simple indicators are just a way of filter out some of the noise. Sure you don't need many conditionals/lines of code for the simple indicators but there is no way to emulate human decission making without getting extremely complex.

As its been mentioned a few times the only way your going to keep something secret is to not show anyone and run your own business. I mean if your making 10-20% a month, do you really expect that guys sitting next to you at the prop to not take your ideas? To not beg you day and night to explain what it is your doing? If you put up those kind of numbers and we worked at the same prop I would drill your brain day and night until you either told me what you were doing or quit because you were sick of me bugging you about it.