I'd like to find out which stocks are likely long and which ones are likely to be short.
I don't think this can be implemented yet as it depends on ta_lib.
ta_lib has a suite of candlestick chart analyzers. Both candlestick charts and candlestick patterns originated in Japan. Here's what I've got so far in pseudo-code:
s[1] = sid(1)
s[2] = sid(2)
s[3] = sid(3)
:
s[n] = sid(n)
# for each sid
for theSid = 1:n
mode = "unknown"
# Bullish Candlestick indicators
if(CDL3WHITESOLDIERS(s[theSid]))
match = match + "CDL3WHITESOLDIERS/"
mode = "bull"
end
if (CDLMORNINGDOJISTAR(s[theSid]))
match = match + "CDLMORNINGDOJISTAR/"
mode = "bull"
end
:
# Bearish Candlestick indicators
if(CDLSEPARATINGLINES(s[theSid]))
match = match + "CDLSEPARATINGLINES/"
mode = "bear"
end
if (CDLKICKING(s[theSid]))
match = match + "CDLKICKING/"
mode = "bear"
end
:
end From there, a list could be built of any stocks showing bull or bear signs. The algo could then short the bulls and go long on the bears.
