The Motley Fool Discussion Boards

Previous Page

Investing/Strategies / Mechanical Investing


Subject:  Re: New screen - votes needed Date:  2/5/2013  3:04 AM
Author:  rgearyiii Number:  241427 of 263912


tpoto's original RadiScript includes:

Create [10DV] :[SI Volume--Average Daily 10d]
Create [3MDV] :([SI Volume--Average Monthly 3m]/21)
Create [26W] :[SI Price Change 26 week]
Create [VR] :([10DV]-[3MDV])/[3MDV]
Create [SORT] :([26W]/[3MDV])*[VR]
Sort Descending [SORT]

It looks like you mistranslated this into "English" when you wrote

VR = [(Average 10-day volume) - (Average 3-mon daily volume)] / (Average 3-mon daily volume)

VR is the normalized recent increase in volume

Final sort, descending:
(26wk change) / [(Average 3-mon daily volume) x VR]

According to the RadiScript, [26W] is first divided by [3MDV] and the result is then multiplied by [VR], meaning the whole thing is proportional to [VR]. Your expression, however, is inversely proportional to [VR]. Consequently, the screen has been inaccurately characterized as rewarding drops in volume, when really it rewards rises in volume.

And by the way, this will make RayVT smile: The CAGR for the 5-stock 20-day hold variant collapses from 51 to 30 (and GSD stays at about 37) when you replace SI Pro data with VL data where possible (using the Expanded Edition). VL+'s coverage of the market isn't quite as thorough as SI Pro's, but the 5-stock screen still picks an average of 4 stocks, versus 4.6 for the SI Pro version. (I tested in "shrink" mode to keep the comparison as fair as possible.)
Create [P2FCF]: [[[daily VL+-adjusted Price; share_lag=1 days; quote_lag=1 days]*[VL+ Common Shares Outstanding; lag=1 da