Stock Market Simulator

The way we can be millionaires through CS267!!!

CS267 Assignment 0

Sukun Kim

 

 

1. Introduction

    What if we can predict the future, or even tomorrow of stock market? Yes, we are MILLIONAIRES! Even if this kind of dream may not be realistic, we know that tools helping market prediction will interest dealers at Wall Street, giving us another chance to become millionaires. Here are applications which prophesy of future of market. I found two interesting applications concerning market simulation. A research group at Los Alamos National Laboratory does economic analysis in diverse areas. And CDAC, an Indian company developed a stock market simulator in a product level. I introduce them here.

 

2. Los Alamos National Laboratory

(1) Application

    A research group in LANL does researches in two categories, Macroeconomics and Microeconomics. Macroeconomics covers International Economics and Regional Economics. Microeconomics includes Energy Economics, Finance, Industry Analysis, and Market Analysis. They are doing extensive economic model simulation. There are many links to research materials. However, unfortunately accesses to them are not permitted. It was impossible to find information about algorithms or performance of application. You can get more information about the research group from the link below.

http://www.lanl.gov/orgs/d/d4/econo/economics.html

 

(2) Platform

    They use the 6th powerful supercomputer in the world (http://www.top500.org/list/2002/06/). It is a cluster of Compaq AlphaServer ES45 Systems (http://www.compaq.com/alphaserver/es45/). Its peak performance is 4096GFlops/s, and LINPACK performance is 2916GFlops/s. Since no information about application could be accessed as stated above, the ratio of performance of application compared to peak performance is unknown.

 

3. Strategy Development 2000

 (1) Application

    Centre for Development of Advanced Computing (CDAC) already developed a stock market simulator and sells it. Named as Strategy Development 2000 (SD2000), the simulator uses artificial intelligence techniques such as neural network, genetic algorithm, fuzzy logic based on massive data base to forecast trader behavior. Unfortunately, as you may guess, the algorithm of the application or research material can not be accessed. Information about SD2000 can be found from the link below.

http://www.cdacindia.com/html/finance/sd2000.asp

 

(2) Platform

    SD2000 uses PARAM 10000. It used up to 4 UltraSPARC-II processors. The PARAM systems can be extended to a cluster supercomputer. A clustered system with 1200 processors can deliver a peak performance of up to 1TFlops/s. Even though PARAM 10000 system is not ranked within top 500 supercomputers, it has a possibility of gaining a high rank. It uses a variation of MPI developed in CDAC. No performance data is available, although one would presume that it will not be very different from that of other UltraSPARC-II based systems using MPI. Because SD2000 is a commercial product, it is impossible to gather detailed data about algorithm and performance of the product.

More detailed specification of PARAM 10000 can be found at http://www.cdacindia.com/html/param.asp

www.top500.org also introduces PARAM 10000 at http://www.top500.org/ORSC/2002/cdac.html#cdac

 

4. Remaining Challenges/Difficulties

    Since economy itself is influenced by every single factor of the world, economic simulation also requires a large amount of data. Moreover, effect of the factors are so entangled and complicated that many techniques from artificial intelligence are needed to imitate them. So these applications not only rely on supercomputer and efficient parallel algorithm, but also rely on huge data base, data mining, and artificial intelligence. This property is a challenging but also exciting point of these applications.

 

5. Evaluation/Conclusion

    Researches in this area are not open to public. Information about applications is also hidden. So I could not get much help concerning parallel computing and algorithm. But, this gives a possibility of commercial use of parallel computing. Market simulation system has unlimited applications. Stock market simulation attracts dealers in Wall Street, crop market simulation will give very useful information for producers to control production amount so that they can prevent a price crash. Economic simulation can be applied to any kind of market, even though it would be quite challenging. This is an excitement and a chance for us.