Location: L'ville, GA
Hi Paul-Jay - welcome to the forum!
There is a long and detailed answer to your question.
The shortest answer is ... "it's proprietary" ... that is, Nirvana has never officially revealed all the formulae and logic to duplicate what they do, and they have been asked a bunch of times so don't hold your breath.
Intermediate answer ... pivots are formed when the price retraces at least "X" ATR's away from an extreme (H or L) ... you don't know that a Pivot IS a pivot till some # of bars afterwards, so Pivots are "hindsight" events. The X multiplier is input using the word "Sensitivity" in many of the N products.
(potentially) Long answer ... bottom line ... it is possible using OLang to exactly duplicate N's pivots from scratch, but it is NOT SIMPLE ... the ATR rule I just mentioned actually has several nuances to it that only can be discovered by cut and try.
RE short med and long ... AS I RECALL, these are nominally calibrated to produce pivots on very rough average intervals of about 32 bars, 64 bars, and 128 bars. But my memory could be wrong about the #'s ... and it doesn't matter anyways because it's just a way of thinking about it. In reality, it is MOST LIKELY (but proprietary so who knows?) that the short med and long distinction is simply increasing values of "X" that I mentioned earlier. That is, a BIG mult of ATR retracement-required will only find a few pivots (Long), while a SMALL mult will find a lot of them (Short).
Hopefully this explanation can be "stickied" by Barry since lots of people wonder about this. If you do a search for "Pivots" on the forum you will probably get LOTS of hits.
I hope this answers your questions adequately.
[Edited by Jim Dean on 1/18/2014 2:51 PM]