I am looking for an algorithm that identifies parabolic PA (similar to Nasdaq in 2000) on the daily time frame in stocks.
Personally, I would use a crossover of two EMAs of price that have been filtered through a parabolic parameter. For example: LONG = emaP3 > emaP12 where P# = Psub3 s.t. Psub0 > [Psub-3 * 1+r] and = Psub12 s.t. Psub0 > [Psub-12 * 1+r] r = whatever your definition of "parabolic" is ... Spitballing, but that should smooth out your data, while insuring you don't trigger a LONG when the thing has just dipped, nor missed a climb b/c the thing had just stalled a second.
Have you tried the "Automated Trading" thread here at Elitetrader.com ? Designing, building, testing, and running Automated Trading Systems, also referred to as Algorithmic Trading, High-Frequency Trading, Black-Box Trading, or Algo Trading. wrbtrader
A stock that went 200% in one month can easily go up more and more till it blow up your account if you short them You will not sleep over night. Check TLRY.