R-Quant

1
Dr. Anton Fokin The Svedberg Laboratory, Sweden
2
Outline
R-Quant is a software toolbox, which provides a
financial researcher or quantitative investor
with an advanced object oriented data analysis
framework. R-Quant is a stand-alone extension of
the ROOT and it is especially designed to work
with financial objects.

• What is R-Quant
• Physics meets Finance
• R-Quant tools
• Conclusions

3
Where physics meets finance
Quark-Gluon Plasma?
4
Where physics meets finance
Or Market Crash?
Phase transition?
5
What is inside?
6
R-Quant tools
Modern financial data management reflects all the
features of full-scale data acquisition and
storage systems we used to deal with in our
experiments. In finance we talk about tens and
hundreds of Gigabytes of off-line data. Real time
data come with tick intervals ranging from
seconds to hours.

• Data management

7
R-Quant tools

• Base instruments
• TAsset
• TRiskFreeAsset
• TBankAccount
• TObligation
• TRiskyAsset
• TStock
• TBond
• TFund

• Financial derivatives
• TDerivativative
• TFutures
• TOption
• TSwap

Inheritance trees representing base and
derivative financial instruments make a perfect
example of the object oriented software
development.
8
R-Quant tools

• Asset pricing

• A number of standard pricing methods including
• Black-Scholes
• Finite difference
• Binomial
• Trinomial
• Monte-Carlo

9
R-Quant tools
Statistical tools for time series analysis. Tens
of embedded indicators and signals. Opportunity
to add new indicators via ROOT macro processor
and script compiler using C interpreter as a
script processor. Powered with ROOT visualization
facility.

• Time series and technical analysis

10
R-Quant tools

• Portfolio management and optimization

• Modern Portfolio Theory (MPT)
• Capital Asset Pricing Model (CAPM)
• Quadratic optimization problem for thousands of
  variables with a number of constraints
• ROOT MINUIT
• Simulated annealing with Metropolis algorithm
• Genetic optimization

11
R-Quant tools
ANN are used in finance for pattern recognition
and forecasting. ANN have the capability to learn
underlying market dynamics from noisy and complex
time series data. GA can help to build optimal
neural network topologies, select good
indicators, create new indicators from existing
ones, etc.

• Artificial Neural Networks and Genetic Algorithms

12
R-Quant tools
In our experiments as well as in finance we use
rather fuzzy definitions such as high or low
which have different numerical value in different
situations. R-Quant implements forward
(conclusion) and backward (explanation) chain
techniques. Fuzzy objects may also serve as
inputs for neural network applications.

• Fuzzy logic and expert systems

Facts dollar value goes up while operating on
the US stock market Rule database if stock
market US and dollar up then interest rate
down if interest rate down then stock market
up Conclusion US stock market goes up
13
Conclusions

• Technology transfer is possible in both
  directions (ROOT2000)
• Thanks to the ROOT team
• R-Quant is an open (source) project
• Welcome to use
• Welcome to join
• Contacts
• http//garbo.lucas.lu.se/kosu_fokin/rquant.htm
• Emailfokin_at_tsl.uu.se