Innovative Credit Risk Assessment System Using Artificial Intelligence-Based Reinforcement Learning

Read full text Summarise with VIRAT
Authors: Sumanth Patibandla
Cited by: 0
References: 21

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The optimization and application of reinforcement learning (RL) in credit risk assessment strategies for cryptocurrency trading focus on leveraging AI-driven models to enhance decision-making in dynamic and volatile markets. Reinforcement learning algorithms are trained to maximize returns by continuously learning from market conditions, historical data, and reward feedback. This involves finetuning parameters, exploring optimal policy strategies, and adapting to rapid price fluctuations. By applying RL, traders can develop automated systems that efficiently execute trades, manage risks, and identify profitable opportunities, providing a competitive edge in the cryptocurrency market. The paper introduces the Quantitative Spider Swarm Reinforcement Learning (QSSRL) algorithm and explores its optimization and application in cryptocurrency trading. Leveraging reinforcement learning techniques, QSSRL adapts trading strategies through interactions with market data to maximize profits while managing risks. The optimization process fine-tunes key parameters such as learning rate, discount factor, and exploration rate, enhancing the algorithm's performance. Application exploration across multiple epochs demonstrates the algorithm's ability to consistently improve profitability and adaptability over time. Results show increasing cumulative returns, Sharpe ratios, and average trade profits, indicating the algorithm's effectiveness in navigating dynamic cryptocurrency markets. The optimization process finetunes key parameters such as learning rate, discount factor, and exploration rate, resulting in significant improvements in performance. Application exploration across multiple epochs demonstrates the algorithm's ability to consistently improve profitability and adaptability over time, with cumulative returns increasing from 20% to 35% and average trade profits rising from $500 to $700.

Fields of Study & Subjects

Stock Market Forecasting Methods Blockchain Technology Applications and Security FinTech Crowdfunding Digital Finance

Publication details

Journal / source
Journal of Computer Allied Intelligence (JCAI).
Volume
4
Issue
1
Pages
23–23
Publication date
28 February 2026
Language
EN