There is always something new to someone with many gaps in knowledge like mine. And the minimum spanning tree which I saw in an astronomy article "bootstrap, data permuting and extreme value distributions" by Suketo Bhavsar (which I couldn't understand but sent to Dr. Z  as he could and Mr. Grain who could also understand it and it seems like a very good thing for technical analysis. But I will have to study it when not losing in market (all too rare). 

Kim Zussman adds: 

The article is unfortunately significantly beyond my boundaries. Observationally I would worry also about what we can see vs what is obscured by interstellar dust, but presumably they adjust by using infared (which penetrates dust better).

But I see your point vis traded price path:

"The Minimal Spanning Tree or MST (Zahn 1971) is a remarkably successful filament tracing algorithm that identifies filaments in galaxy maps (BBS; Ung 1987; BL I). The technique, derived from graph theory, constructs N - 1 straight lines (edges) to connect the N points of a distribution so as to minimize the sum total length of the edges. This network of edges uniquely connects all (spans) N data paints, in a minimal way, without forming any closed circuits (a tree). In order to distil the dominant features of the MST from the noise, the operation of "pruning" is performed. A tree is pruned to level p when all branches with k galaxies, where k < p, have been removed. Pruning effectively removes noise and keeps prominent features."

Victor Niederhoffer writes: 

But it's descriptive right? I don't see anything about how to use it as predictive?

Kim Zussman replies: 

They seem to be trying to lift signal (filaments) from background noise. Do you think some market moves are signal and others noise?

Victor Niederhoffer responds: 

Yes. Like when bonds and stocks are both up on the day. The green on our chart or a break of a round number. Like an orgasm. But I can't understand the astronomy of the Indian paper so can't unravel it. 





Speak your mind


Resources & Links