<p>High-frequency trading (HFT) environments expose delegated proof of stake (DPoS) blockchains to transaction ordering manipulation, where privileged delegates may front-run transactions for profit. This undermines fairness and trust in time-sensitive applications. We propose a modular security optimization framework that integrates transaction scheduling, deterministic order commitment, anomaly detection, and behavior-based enforcement into the DPoS delegate workflow. Our system filters and prioritizes incoming transactions, commits execution order cryptographically, detects suspicious reordering behavior, and dynamically adjusts delegate influence based on detection scores. Evaluated in a simulated DPoS environment under adversarial HFT workloads, our mechanism reduces front-running success by over 90%, detects malicious delegates with 0.83 AUC accuracy, and maintains over 95% of baseline throughput. Unlike consensus-level adjustments that increase complexity or latency, our framework introduces a modular, in-line security layer within DPoS delegates. This layer integrates transaction filtering, cryptographic ordering, anomaly detection, and penalty enforcement directly in the transaction pipeline, offering a novel, tamper-resistant approach to ensure fairness without modifying core consensus behavior. These results demonstrate that our approach achieves secure and fair transaction ordering without compromising performance, providing a practical enhancement to the integrity of DPoS-based financial systems.</p>