I too tend prefer the fundamental analysis and I certainly understand the discounted cash flow approach adopted by many analysts, even though this has flaws.
But technical analysis, particularly the application of artificial intelligence to specific stocks may prove a surprise to us all. We have some idea of how SPL is held and it is not by those interested in franked dividends. We know there is shorting and day trading activity aside from the underlying long term holders both institutional and otherwise.
An AI algorithm would need to focus on the identification of trade patterns for a specific stock. The patterns for SPL would I expect be very different from the like of BHP or ANZ as would the influencing factors like currency, market movement, sector movement and announcements.
My point is that technical algorithms of a general nature are probably not going to get us very far. Here are a couple of papers:
This first one is a thesis from Stanford University and is NOT what I have in mind. The data sets are far to restrictive and the predictive algorithms are input (not output).
This one is more along the lines suggested and uses learning techniques to identify patterns from a lot of data (such as the tick feed for a stock over time). The results should be different for different stocks.
