This project is used to implement various stock pattern recognition mechanisms. Focus is on implementing algorithms to identify the cup and handle pattern as popularised by William O' Neil in his CANSLIM approach to stock picking.

The codes are based on the following papers:

1. http://web.mit.edu/people/wangj/pap/LoMamayskyWang00.pdf
2. ftp://80.240.216.180/Transmission/%D0%A4%D0%B0%D0%B9%D0%BB%D1%8B/S&C%20on%20DVD%2011.26/VOLUMES/V16/C10/077CUP.pdf

Bandwidth for kernel regression = 3.

1. p.a > 30, p.a >= 1.15*p.k
2. uprv1/dprv1 > 1

1. p.b < 0.85p.a, p.b >= 0.65p.a
2. avg.vol < avg.ma.vol
3. Length of left side of cup 20 to 60 days

1. p.c > 0.6p.a + 0.4p.b & p.c <= p.a
2. uprv2/dprv2 > 1
3. Length of right side of cup 3 to 30 days

1. p.d <= p.c
2. uprv2/dprv3 > 1
3. Length of right side of cup 3 to 20 days