Neural Network Stock Index Forecasting

1
Neural NetworkStock Index Forecasting

• 2004/05/20

2
Outline

• The problem of forecasting stock market
• The development of ANN model
• Neural function
• Neural architecture
• Neural learning
• Case study
• Kula Lumpur Composite Index (KLCI)
• Discussion

3
The problem of forecasting stock market

• People tend to invest in equity because of its
  high returns over time.
• Stock markets are affected by many factors, and
  hence it is very difficult to forecast the
  movements of stock markets.
• Prediction in stock market has been a hot topic
  for investors and researchers.

4
The development of ANN model

• Neural function
• Neural net architecture feedforward network
• 5-3-1
• 5-4-1
• 5-3-2-1
• 6-3-1
• 6-5-1
• 6-4-3-1
• Neural learning LMS

5
Case study - KLCI

• The Kuala Lumpur Stock Exchange (KLSE) is
  considered a young and speculative market
• 10 years of history
• 492 companies
• Kuala Lumpur Composite Index (KLCI) is calculated
  on the basis of 86 major Malaysian stocks.

6
Case study -- KLCI

• Major types of indicators
• Stock index (It-1, It, It1)
• Moving average (MA5, MA10, MA50)
• Momentum (M)
• Relative Strength Index (RSI)
• Stochastics (K)
• Moving average of stochastics (D)

7
Case study -- KLCI

• Daily data from Jan 3, 1984 to Oct 16 1991 are
  collected.

8
Case study -- KLCI

• Normalization
• Building neural network model
• Profit strategy
• Seed money is used to buy a certain number of
  indexed stocks when the prediction shows a rise
• The basket of stocks will be held until the next
  turning point that the neural network predicts

9
Case study -- KLCI
10
Discussion

• How do we determine the architecture of neural
  network?
• General rule
• Trail and error
• Can we predict all stock markets?
• Random walk vs predictable