|
Supercomputing-based
Data Mining for Equities
The battle in today's
financial markets is increasingly being waged at computer
speed and not human speed. More and more traders now rely
on computerized decision-making approaches to help them
develop winning investment strategies. The use of a supercomputer
is of continuing importance especially in financial modeling
because of the massive amount of data that needs to be processed
for intra-day equity strategies. Also anyone failing to
exploit these new tools may find survival difficult.
SD2000 is a supercomputing
based Decision
Support System using Mathematical, Statistical and Artificial
Intelligence based techniques for optimizing returns while
trading in financial markets. The solutions tend to improve
the odds of winning in a complex, uncertain world.
The primary objective
is to do Quantitative Modeling of global financial markets
for investment analysis and risk management. The system
focuses on data of very short term price movements, including
every single trade, in a broad range of global markets.
The density of this information requires the power of a
supercomputer to perform effective analysis. The "tick-by-tick"
database that is being built is one of the largest in the
world dedicated to financial markets research.
 The
overall objective is to forecast not price, but "ideal trader
behavior". The system will emulate trading actions of a
superlative trader in the real world market environment.
The system enables traders to monitor foreign exchange and
stock markets by incorporating and integrating human-expert
analysis with human-independent quantitative analysis of
the markets.
The traders can deploy
this system for developing their own proprietary strategies,
back-testing them on the historical data and analyze their
performance. The ultimate goal is to filter the successful
strategies for deployment in the real market.
Salient Features
- Centralized Repository of terabytes
of Historical Data of Stocks and Foreign Exchange traded
in global Stock Exchanges
- Technical and Quantitative Analysis
based on time-tested Mathematical and Statistical Models
- Proprietary Trading Strategies
using Artificial Intelligence Techniques such as Neural
Networks, Genetic Algorithm, Fuzzy Logic ...
- Strategy Development Library enabling
traders to code proprietary strategies
- Back-testing of the Trading Strategies
on Historical Data
- Trades and Strategy Performance
Reports capturing Net Profit/Loss, Net Position, Maximum
Drawdown, Strategy Parameters ...
Success Stories:
Currently, SD2000 is operational
on C-DAC's PARAM 10000 supercomputer for the financial engineering activities of
Man Drapeau Research Pte. Ltd., Singapore. This is one of
the first and largest supercomputing facilities in the world
deployed for financial modeling and engineering.
|
