Forecasting the 10-year Philippine Treasury Bond with a system of Neural Networks

Date

11-2007

Degree

Bachelor of Science in Applied Mathematics

College

College of Arts and Sciences (CAS)

Adviser/Committee Chair

Sharon P. Lubag

Abstract

The forecasting model used to predict the weekly average closing price and yield to maturity (YTM) of 10-Year Philippine Treasury Bond Series No. SPT2 is based on a system of Neural Network (NN). Weekly data are assumed by the closing price and yield for a given Friday. Using 4-week past data, sixty four networks with different architectures were trained for both price and YTM prediction. The feasibility of using Feedforward Backpropagation Neural Network in financial forecasting is also analyzed. The results were indeed satisfactory, giving a strong positive correlation, while maintaining the true nature of price and yield-maturity. The performance of neural network presented n this study demonstrate a clearly possibility of learning. Therefore, the accuracy of the network's prediction was determined not merely by the structure, but also by the amount of historical data used in the training set, how the data is pre-processed, the range and number of input vector, and the number of learning iterations.

Language

English

Location

UPLB Main Library Special Collections Section (USCS)

Call Number

Thesis

Document Type

Thesis

This document is currently not available here.

Share

COinS