عنوان مقاله:

نویسندگان:

پگاه مرادی، سید محمدمهدی کاظمی

Generative models for financial time series often struggle to model practical financial objectives, which limits their applicability. In this study, we propose an improved Generative Adversarial Network framework that combines advanced sequence modeling structures, such as Temporal Convolutional Networks capable of capturing long-range dependencies and volatility patterns, with a novel generator function. This generator function is explicitly constructed to ensure that the resulting stochastic process can transition to its corresponding risk-neutral distribution. It also incorporates a new type of loss function, designed based on incorporating financial performance measures. These components aim to improve the profitability of trading strategies, enhance risk-adjusted returns, and manage the volatility of financial outcomes. Our numerical results demonstrate that the distributional characteristics at both short and long time lags closely match the expected patterns, indicating that the methodology can lead to the generation of more realistic and actionable market scenarios for robust financial decisionmaking.

Generative Adversarial Networks, Temporal Convolutional Networks, Stochastic Process, Time Series

برای لینک‌دهی به این مقاله از آدرس زیر استفاده نمایید:

برای ارجاع در منابع از عبارت زیر استفاده کنید:
مرادی، پگاه، محمدمهدی کاظمی، سید، 1404، بهینه‌سازی شبکه‌های مولد متخاصم برای سری‌های زمانی مالی: یک رویکرد تصادفی با معیارهای تابع هزینه کارآمد، پانزدهمین سمینار احتمال و فرایندهای تصادفی، https://conference.uok.ac.ir/fa/archive_index.php?c=HN-spsp15&aid=1518052