|
|
 |
|
Viewing report
|
|
|
|
RISK CYBERNETICS: Training Manual Version 1.0
Finamatrix, Jan 2010, Pages: 300
Volatility Feedback Control for Automated Trading with Genetic-Algorithm Neural-Network (GANN) Optimization: An Application on FX Markets (EURUSD and USDJPY)
Objectives: High Volatility Trading, Low Latency Automated Trading, Algorithmic Trading on Forex, Index Futures, Commodities, and other volatile financial instruments. Decoding the Black Box; Creating Expert Advisors.
Towards achieving the Chartered Risk Cybernetician (CRC) global designation.
Abstract
The major elements of this exploratory research, which contribute to the epistemology of financial knowledge, are risk management, specification and control techniques using advanced human thinking and computer technologies with circular causal volatility feedback in a Genetic Algorithm Neural Network (GANN) framework developed in Microsoft Visual C++, which is termed Risk Cybernetics. This is the foundation of automation in trading with automated trading robots and exotic systems. This study augments over four decades of research and is pioneering work in the field of risk-profiling of foreign exchange and in the application of the golden-ratio, genetic algorithms and neural networks in an empirical framework for forecasting and risk management. Inspired in part by optimizing random games of chance in casinos, the study proposes a modern and flexible research methodology that is easily implementable in signal forecasting for volatility control, and provides a solution to the problem of unpredictability in currency rates. The study offers a possible relationship between prices, returns, volatilities and a full suite of technical explanatory variables.
This research uncovers the explanatory power of technical variables with stepwise GLS estimators and tests signal prediction with upside and downside risk statistics, momentum and other variables. This includes a randomized data series implemented with pseudo random number generators that test for random-efficiency of the optimization model. When volatility breaches GF-risk threshold, we observe the existence of volatility fractures, which is also tested with the defined volatility-cluster-fracture (VCF) analysis, supporting previous studies of periods of fractals and stability patterns. Data tested herein is the weekly prices of the US dollar per Euro (EURUSD) and the Japanese Yen per US dollar (USDJPY) for a 200-period trading strategy from 2005 to 2009. By developing a genetic-optimization engine in a neural network and applying a modified value-at-risk (MVaR) framework, we create a decision support system for risk and trading management. The results show that the GANN model with feedback control is more efficient in producing high efficiency trading signals which forms the basis of risk cybernetics and autobots. Also, MVaR estimates under GANN provide values of 4 to 5 times smaller than MVaR without GANN signals. Implementing this research is in accordance with the latest ISO 31000:2009 guidelines and a promising tool for monetary policy.
Product samples
A sample for this product is available. Please Login/Register to download this sample.
|
|
|
|
