Ha, Big Blue Marble has a project like this.
The web front end is a really simple database-backed site that does none of this interesting processing. When someone requests information on a stock that's not already in the database, the site inserts a stub entry for that stock.
Every few minutes, a cron job on the server looks for stub entries and runs the gamut of historical and daily analysis on those stubs. The live site updates what it displays for that stock as new information comes in. So far no one has objected to waiting up to five minutes for intrinsic value calculations. (We discussed running the cron job more frequently, but I haven't made that happen yet.)
I ported a couple of formulas from Excel, but if you already understand things like a present value calculation, or could do very basic calculus on the back of an envelope (did that too!), you can reproduce those formulas easily in Perl.
With that all said...
I use excel functions to operate on the historical information to predict the next trading day's behavior...
... my best advice is to pray a lot. (I'm a value investor.)
In reply to Re: Scrape Yahoo Financial Historical- Process Dataset - format and create dynamic page