there is a book available "algorithmic trading" leshik and cralle;
lots of hardware reqd,there is a cd,and it is all backed by enormous amount of back testing and 12 years of research. and by now it will most likely be outdated,but its a start. i read what i could understand,and i have spoken to people who really know about high end maths the consensus is they will just build a better algo,now its not that you cant make money,
but from my own experience watching live data then stripping it to arrive at what might be raw data i like to calculate the broader strategy that might be in play,from there it is a matter of finding an entry and importantly your exit before you place an order
so to digress, the book i mentioned is for the usa,where the uptick rule is deployed.you will notice a difference on the asx in european or usa public holidays just so you know you are not alone
what i have noticed is those stocks held by a desk automated or not in australia they require human participation throughout the day,that is the first weakness.
setups are just that,its not always the intent to trade so there will be no data forthcoming,i realise its not algo which is your focus.
however if you could develop an algo that simulates a potential plan from a desk it might work very well,all algos have one objective to make money from the volume price spread,it could be just to cover desk costs but none of that concerns you.
lastly for a system such as the one i am proposing,you would have to be able to hide any order or buy or sell at market within seconds to 2minutes dependant on liquidity,and when you get an algo dont tell anyone and keep moving through a range of stocks.
otherwise you might enjoy the fate of jesse livermore and the bucket shops,because they will fight tooth and nail to push you out
