Hello folks! Your help is really appreciated!I need to graduate to publish in a review and my group (5 guys) is hesitating between 2 topics:1) study neural network prediction stocks using technical analysis --> there are plenty of articles about that so my group could find solutions more easily 2) try to divide the autocorrelation function into a pseudoperiodic function and a pseudorandom function; when the pseudorandom fuction would be too high, that would mean that the usual models might be wrong for the moment--> this is more academic and then the chance to be published would be higher but its much more complicated and the deadline is in FebruaryWhat do you think about that 2 topics? Does someone has an experience about how to publish? What kind of review are willing to publish students? O.oThanks for your help! Ben

Hi,Though I am not having a great deal of knowledge but would like to give some ideas:Nueral Network has been nearly exhausted for basic research. You have to go to down a lot as at abstraction level you cannot have anything that has an impact. It has been explored a lot in recent years.Psuedoperiodic function seems a better option. I think you can go deeper in it as you said and it is academic as well. This would be a nice idea to explore into. here is a lot of scope for any origional Mathematical & Computational model. I am reading and exploring idea for publishing research on copula on Credit risk on MATLAB / R. Hope to get something good in near future.Shivgan