USING SYNTHETIC DATA TO MODEL A PORTFOLIO IN CONDITIONS OF HIGH VOLATILITY. HOW SYNTHETIC DATA ALLOWS YOU TO TEST STRATEGIES FOR RARE MARKET EVENTS. EXAMPLES OF GENERATIVE MODELS APPLICATION

Аннотация

INTRODUCTION involves the use of synthetic data generated to Modern financial markets are characterized by analyze various market scenarios, including high volatility, limiting the applicability of financial crises, sharp asset price fluctuations, and traditional methods for forecasting and testing changes in exchange rates. investment strategies under extreme fluctuations. In such circumstances, it is essential to develop According to estimates by the McKinsey Global methods that model asset behavior in rare and Institute, generative AI could deliver between $2.6 unconventional market situations. These methods trillion and $4.4 trillion annually across various facilitate accurate assessments of portfolio industries worldwide in the 63 scenarios analyzed. stability and profitability. One such approach Among industries, the banking sector is expected https://www.theamericanjournals.com/index.php/tajmei 55 THE USA JOURNALS THE AMERICAN JOURNAL OF MANAGEMENT AND ECONOMICS INNOVATIONS (ISSN- 2693-0811) VOLUME 06 ISSUE12 to realize some of the greatest benefits, with an surges are addressed through conditional annual potential ranging from $200 billion to $340 generative adversarial networks, as described in billion (equivalent to 9–15 percent of operating the work of Chen Y. et al. [9]. profit), primarily driven by productivity gains. The Research by Tepelyan R., Gopal A. [5], Gibson L., economic impact is likely to be positive across all Hoerger M., and Kroese D. [6] confirms the banking segments and functions, with the highest effectiveness of these networks in forecasting absolute benefits projected in the corporate and returns. These methods aid in the development of retail sectors ($56 billion and $54 billion, strategies that account for changing market respectively) [11]. conditions. These models replicate dependencies The article by Naritomi Y. and Adachi T. [2] characteristic of real markets, including discusses the use of generative networks to extraordinary situations. The application of such improve trading forecasts, while Dogariu M. et al. methods enables the testing of strategies under [3, 10] explore the potential of generative conditions that are rarely represented in historical networks for testing strategies. data, improving risk forecasting and enhancing The method proposed by Li J. et al. [7] focuses on portfolio resilience. the application of Wasserstein in generating The use of synthetic data requires the synthetic flows of market orders, facilitating the development of new methods for risk assessment modeling of market liquidity. Coletta A. et al. [8] and portfolio optimization under conditions of use conditional generative adversarial networks to economic instability. Amid environmental changes simulate market scenarios, expanding the and market uncertainty, employing synthetic data possibilities for modeling various strategies. for testing and adapting strategies represents a The practical examples described in this study critical task, as it expands modeling capabilities draw on sources [13-14], available on the websites and improves the quality of investment decision- *masterofcode.com* and making. *www.orientsoftware.com*, which detail the The purpose of this article is to explore the experiences of companies using generative potential of using synthetic data for portfolio networks for portfolio modeling. Source [11], modeling under conditions of high volatility, as located on the website *www.mckinsey.com*, well as to examine practical examples. provides statistical data related to the use of METHODS generative intelligence. Studies dedicated to forecasting rare events The article employs a method of analyzing existing highlight the necessity of employing generative scientific literature and reviewing real-world adversarial networks (GANs) to create synthetic examples in the field. data. Articles by Labiad B., Berrado A., Benabbou L. RESULTS AND DISCUSSIONS [1], and Dannels, S. [4] demonstrate the application Financial markets in recent decades have been of these models for generating market data under characterized by high volatility, complicating asset conditions of increased volatility. This approach management. In such conditions, methods that enables the modeling of phenomena that are adequately reflect ongoing changes are essential. challenging to capture using traditional methods. Forecasts based on historical data analysis lose Issues related to the scarcity of data on market their effectiveness in predicting future https://www.theamericanjournals.com/index.php/tajmei 56 THE USA JOURNALS THE AMERICAN JOURNAL OF MANAGEMENT AND ECONOMICS INNOVATIONS (ISSN- 2693-0811) VOLUME 06 ISSUE12 developments. Synthetic data generated using consist of two neural models: one generates data, mathematical models offer opportunities for while the other evaluates their conformity to developing risk management techniques. predefined parameters. GANs serve as a tool for generating data, includin