I admittedly did not get NEARLY as much time as I would have liked to on this assignment. I spent much more of the time on the alternate assignment 8 of documenting and answering Roys question on my project prototype.
I had already done this project once in the past, and did not care to do the exact same thing again. I wanted to relate my searches into some real inquiries and curiosities that I have had over the last few weeks of implementation and seeing other projects.
I was very interested and enthused by the project and results of Matt’s Neural Net in MATLAB project. Me and him have talked by email and met outside class and agreed that we may be a good supplemental fit to work on some projects and ideas together. I wanted to find out if his general idea of fundamental data to predict market outcome via Neural Nets was a new idea, or if it had been done before.
From my cursory parusal of the below papers, that I got from Business premiere and IEEE databases via the web and JabRef, I can report that I did find relevant information. There have been a handful of projects that I could find that link fundamental, technical, and neural nets or artificial intelligence, but not many! My cursory view of some of the research is initially promising because fundamental + heterogeneous data + ANN’s looks to be a lightly touched research subject. Please see one of the most clear cut references to the subject quoted below:
Several studies have examined the cross-sectional relationship between stock index and fundamental variables.34,66 Variables such as earnings yield, cash flow yield, book to market ratio and size have been found to have some power in pre- dicting stock indices in these cross-sectional studies. These studies in general find positive relationships between the stock index and earnings yield, and between the cash flow yield and book-to-market ratio, and a negative relationship between stock returns and size. Fundamental variables are not the only type of cross-sectional vari- able that contain information for predictability. Chopra, Lakonishok and Ritter21 documented that a stock’s ranking in terms of its performance relative to the market can contain predictability. Extreme losers have been shown to outperform the mar- ket over subsequent years. In empirical work, several studies have found evidence of nonlinearities in the relationship between stock returns and trading volume, and stock returns and dividends.