originally, my hypothesis was going to be: program trading, as reported to the NYSE, regularly runs around 50% of the total volume - the majority of program trading is statistical driven. can one conclude then, that generally half of the volume is created by non-programs (i.e. "humans", discretionary, events etc.) and this "noise" is fairly symmetrically countered / pared / absorbed by the programs. i am wondering if this conclusion would be far off from reality. however, i have just looked up the recent program trading stats. http://www.nyse.com/marketinfo/datalib/1152267398806.html turns out they have recently changed the calculation methodology and there is a very significant difference: the newly calculated prog. trading share of total volume is only just over half of the old one, i.e. used to be mostly in the 50% area, now around 30% of the total volume. so in this case my hypothesis would be that programs absorb about half of the "noise" volume (albeit the program's share is ever-increasing). would anybody have additional thoughts or comments? p.s. it seems easy to guess how they ended up changing the methodology - under the old one, the last reported week (june 26) showed a surge in prog. volume to a whopping 92% of total volume - somebody must have finally woken up and said: geez, that looks a bit high (wouldn't 92% imply the market is dominated by machines trading against each other).