<p>As data trading markets expand, risks such as data breaches, misuse, and the circulation of falsified information increasingly threaten trust and security. This study constructs a tripartite evolutionary game model involving the government regulators, data providers, and blockchain service providers. Prospect theory is embedded into both the payoff structure and risk perception process to capture behavioral biases. The model compares static and compliance-contingent regulatory incentive mechanisms, with MATLAB-based simulations used to examine system stability and evolutionary trajectories under varying parameter combinations. Results show that: (1) Under static incentive schemes, excessively high penalties or rewards may dampen compliance incentives, suggesting that stricter regulation does not necessarily yield higher compliance; (2) compliance-contingent incentives, which link reward–penalty intensity to compliance levels, are more effective in driving the system toward high-compliance and high-security equilibria under lower regulatory pressure; and (3) cognitive biases such as overconfidence and loss aversion distort strategic decisions and weaken the effectiveness of traditional incentive tools. By integrating prospect theory into blockchain-based data trading governance for the first time, this paper identifies the effective boundary of punishment intensity and highlights the disruptive role of risk perception biases in incentive design. The findings offer theoretical and practical insights for optimizing differentiated regulatory strategies and platform security governance, contributing to the development of a safer and more sustainable data trading ecosystem.</p>