Is there anything factual about these two points on this chart -- or any two arbitrary points on a chart?
The same line can be been drawn any number of times in the past five years, over any given time frame, connecting any two or even three or more points in "alignment," but what's the proven statistical significance of these connections?
On a price chart, it's defined by its statistical reliability or it's just a decoration.
I've studied various ways to annotate price charts over the years. I found ways to quantify the statistical reliability of all kinds of things (mostly, below a "coin toss" 50:50 expectation) and all kinds of things are taken quite seriously (50 MA, 200 MA, "golden" cross, etc.) but they each turn out to be demonstrably unreliable.
I wonder if there's some recent work to prove a trend line has any statistically reliable significance. It would be great to have a set of rules that says something like "a line connecting two points of 60 bar highs, separated by one 60 bar low precedes a period of 60 subsequent bars with an 80% likelihood of any new 60 bar high being at or below the projection of the line within 1%" ... or something along those lines.
The same line can be been drawn any number of times in the past five years, over any given time frame, connecting any two or even three or more points in "alignment," but what's the proven statistical significance of these connections?
On a price chart, it's defined by its statistical reliability or it's just a decoration.
I've studied various ways to annotate price charts over the years. I found ways to quantify the statistical reliability of all kinds of things (mostly, below a "coin toss" 50:50 expectation) and all kinds of things are taken quite seriously (50 MA, 200 MA, "golden" cross, etc.) but they each turn out to be demonstrably unreliable.
I wonder if there's some recent work to prove a trend line has any statistically reliable significance. It would be great to have a set of rules that says something like "a line connecting two points of 60 bar highs, separated by one 60 bar low precedes a period of 60 subsequent bars with an 80% likelihood of any new 60 bar high being at or below the projection of the line within 1%" ... or something along those lines.
Anybody done that work? 8/18/12