Posted by frtaylor on 30th of Mar 2015 at 09:13 am
So, it gives false positives but not false negatives. Most would
say that's better than false negatives. Saying, "it only ever
produced five false signals" makes it sound incredible, but there
were only 15 signals total. So 1 in 3 is a false positive. Still
pretty good, especially considering that 2013-14 was anomalous in
terms of market strength.
it's interesting, but they lock it down - also notice that they
restrict your time frame, i.e. you can't go back more than about
2006 even though @SP goes back many years, or try it on $INX even
though it goes back to 1961 it forces the chart to only go back to
about 2006 or so. I suspect that it's lock either because it
didn't work very well before that time frame, or they lock it down
with and then later on offer a paid version that is less
restrictive or something. It also seems to work best on @SP,
the large ES contract, and not so well on other things. Also seems
to work best on the weekly vs the daily
Sort of seems like a classic marketing tool: Show an
interesting indicator, then give a version of it way free - which
generates publicity and traffic - then perhaps sell something later
on as well
So, it gives false positives
Stock market crash warning signal has gone off The indicator has ...
Posted by frtaylor on 30th of Mar 2015 at 09:13 am
So, it gives false positives but not false negatives. Most would say that's better than false negatives. Saying, "it only ever produced five false signals" makes it sound incredible, but there were only 15 signals total. So 1 in 3 is a false positive. Still pretty good, especially considering that 2013-14 was anomalous in terms of market strength.
it's interesting, but they lock
Posted by matt on 30th of Mar 2015 at 09:26 am
it's interesting, but they lock it down - also notice that they restrict your time frame, i.e. you can't go back more than about 2006 even though @SP goes back many years, or try it on $INX even though it goes back to 1961 it forces the chart to only go back to about 2006 or so. I suspect that it's lock either because it didn't work very well before that time frame, or they lock it down with and then later on offer a paid version that is less restrictive or something. It also seems to work best on @SP, the large ES contract, and not so well on other things. Also seems to work best on the weekly vs the daily
Sort of seems like a classic marketing tool: Show an interesting indicator, then give a version of it way free - which generates publicity and traffic - then perhaps sell something later on as well
Didn't Art Cashin count about
Posted by a_l_ on 30th of Mar 2015 at 09:22 am
Didn't Art Cashin count about half a dozen false positives in 2013 alone?