I found the same problem, I thought it would be possible to do iterations to identify "rolling" stocks with a sine wave formula. Pulled out the old trig books and jumped into sine, cosin... etc.
There may be some other ways to narrow down all the possibilities, too. Moving averages might be easier to work with. Also, the steps in the loops could be increased, from say 1 to 5.
I think, too, the scans could work their way back through the cup, starting first with the move from the base to the right rim. That's a significant move, and a scan that focused on it would narrow things down a lot. Then the next scan could look for the left rim of the cup, or the handle. The "setup" of the cup, the move from Point K to Point A in Martinelli's article, would be the last thing to look for.
One other thing: The time frames for the components of the cup and handle in Martinelli's article could be reduced. I noticed that the "search results" in Figure 4 seem to fall well below the maximum lengths in the time frames in Figure 3. I'm sure longer frames show up in the results -- Figure 4 just gives a few results -- but scans that would find smaller cups would still be useful.
Anyway, it's looking more and more possible.
Brooke |
