Transcript Presentation

Financial Data mining
and Tools
CSCI 4333 Presentation
Group 6
Date10th November 2003
Group Information
Group members
Muralikrishna Pinnaka
[email protected]
Prateek Bali
[email protected]
Azam cheema
[email protected]
Kashif bhatti
[email protected]
2
Financial Data mining
Introduction
Time series analysis
Long term or trend moment
Cyclic moments or cyclic variations
Seasonal moments or seasonal variations
Irregular moments
3
Stock Market Prediction
Stock market data
Programming
Models
Statistical indicators
Genetic programming
Neural networks
Trade simulator
4
Stock Market Prediction(Pictorial view)
Stock market Data
Trade_Creator
Data Element
Data mining model
Trades Data
Trade Simulator
5
What Is Stock Charting?
Technical aspect of the stock market
Identifying buy/sell signals
Dow theory
Primary trend is constant
May be changes in stock market are secondary
Elliot Wave Theory
Prices move in predetermined no of waves using (fibbonacci)
6
Stock Charting
1.Grand Supercycle
2.Supercycle
3.Cycle
4.Primary
5.Intermediate
6.Minor
7.Minute
8.Minuette
9.Sub-Minuette
7
Few Applications in Data mining
Individuals are likely to go bankrupt
Who will be interested in buying certain products
How valuable a particular customer is
Who is a good risk for an auto loan
What tax returns are likely to be fraudulent
The probability that a particular credit card stolen
8
Classification, clustering and Prediction
Two different forms of data analysis
Used to extract models for predicting trends
Decision trees
Trends are forecasted in multiple directions
Ability to model highly complex functions
Ability to use more no of variable in a functions
Cluster Analysis
Collection of patterns which are similar
Kohenen’s SOM (self organizing map)
9
JExpress-clustering
10
Algorithms
 ARIMA( autoregressive integrated moving average)
 Takes time series data as input
 Prepares a model for extrapolating the financial market
 attempts to evaluate the stationarity of a time series
 Ordering the autoregression and moving average components
 estimation of the autoregression and moving average
 Neural Networks
 Able to respond with true(1) or false(0) for a input vector
 Highly complex and more processing power required
 It consists of
 Input layer
 Output layer
 Hiddern layer
11
Genetic programming
Genetic programming
Automated method
Writing a computer program which know how to program computer
Genetic algorithms
 Adaptive
Search and optimization problems
Survival of the fittest
Search starts from population of many points(parallell)
Dealing with broader class of functions
Rules are probabilistic but not deterministic
12
Genetic programming
Parameter used
Fitness function and value
No of individuals(112)
No of generations(max 1000, used 3)
Percentage cross over
Probability of function ( 30%)
R square value( Ex: 1.000 means fittest)
Input
X Y
# @ - / * +
13
Genetic Programming Tool
14
Conclusions
Genetic programming
Useful in game programming
Useful in predicting the future trend of the stock market
Used in financial institutions
Statistical modeling techniques
ARIMA used for extrapolation
Neural networks
Highly complex and more processing power is needed
It is not in great practice
15
References








http://www.5paisa.com/abc/rsrh/Research/BSX/Technicals/SUM/ACCL.BO.html
http://www.stockcharts.com/education/index.html
http://www.stockcharts.com/education/What/IndicatorAnalysis/indic_RSI.html
http://www.stockcharts.com/education/What/IndicatorAnalysis/indic_williamsR.html
http://www.stockcharts.com/education/What/IndicatorAnalysis/indic_Bbands.html
http://www.equis.com/Education/TAAZ/
http://www.lascruces.com/~rfrye/complexica/dm_whatis.htm
http://www.pafis.shh.fi/~ulidau02/SFIS/workshop2.htm
16
Thank you
Questions?
17