This preview shows page 1. Sign up to view the full content.
Unformatted text preview: in buy/sell
recommendations. The inputs used are the S&P 500 index return for the past 3 days and
the US Treasury rate lagged 2 and 3 months. The authors report that a simple filtering
scheme gives better results than the more complex scheme using directional indicators.
The average annual rate of return obtained using this approach is close to 15% which is
much higher than the return obtained using a buy-hold strategy. This approach is not used
in this project owing to its complexity.
Some of the literature dealing with stock market prediction is described above. Of these
papers, the present project is based on the one week ahead prediction approach of .
The next section describes the assumptions underlying the work done in this project, the
analysis of data and extraction of features for training of the neural network. 3. Data Analysis and Feature extraction
The basis of this project is the assumption that in the long term, interest rates affect the
stock market after some delay. There are many models constructed by economic analys...
View Full Document
- Spring '12