Recent content by NYDreamer

Yes, education in the US is another problem, but I was talking about this phenomenon: http://www.npr.org/blogs/alltechconsidered/2013/04/03/176134694/Whos-Hiring-H1-B-Visa-Workers-Its-Not-Who-You-Might-Think I didn't deny that there is a lack of skilled workers, and you are right that the...

Yes, the visas are being abused on a large scale by using them for importing cheap labor (Indian outsourcing companies received over half of the available visas in 2012). It's unfair competition for US engineers, and it will hurt the US in the long run. A few simple additional conditions on the...

"(...) bitterness, instead of a form of disillusionment, is really the refusal to give up your childhood illusions of importance. Ignored instead of welcomed by the world, you fault the world as blind and evil in order not to fault yourself as naïve. Bitterness is a child’s coddling...

Detroit needs a superhero to clean up the filth. Maybe something like a half robot, half cop?

Ask yourself: the Nth percentile of which distribution?

eoddata.com is not expensive.

You should be making an 8-dimensional matrix (that's what vectorization means, building vectors and matrices) and apply your function to that. This operation will be a 1000 times faster that way. http://www.mathworks.com/help/matlab/matlab_prog/techniques-for-improving-performance.html#br8fs0d-1

For accurate backtesting, one needs to define his/her historical universe. Suppose I want to trade small and micro cap stocks only, so that I define my universe below a certain market cap limit. For accurate backtesting, I would need either historical market caps, or historical components of...

Python is even easier.

Pascal is just a fad. Abacus and sun dial are my preferred tools for trading!

You don't need a DB layer at all, really, but I might use tick data someday, so it can't hurt to plan ahead. I suppose HDF5 is less complex compared to SQL, which is a plus for a DB neophyte like myself. BTW, another ET thread on HDF5...

Thanks, interesting info. However, we are talking EOD data here.

After more reading, it seems that HDF5 is more suitable for time series. Well-supported by both Python (PyTables) and Matlab.

Currently planning a similar project (EOD stock database for Python and Matlab access). I have little practical experience with SQL, but I'm under the impression that it is the best solution. Can't think of any reason why you would need monster queries such as you describe. Speed shouldn't be...