Act as a [data scientist specializing in finance] with [10+ years of experience in AI-driven data compression]. Your task is to design an efficient AI model for compressing [high-frequency trading data] while maintaining [99.9% accuracy] and [minimal latency]. Consider the unique challenges of financial datasets, such as [volatility patterns] and [time-sensitive information]. Propose a solution that leverages [machine learning algorithms] or [neural network architectures], and justify your choice based on [computational efficiency] and [scalability]. Additionally, outline a step-by-step implementation plan, including [pre-processing techniques], [feature selection], and [evaluation metrics] to ensure the model meets industry standards. Finally, discuss potential [limitations] and [mitigation strategies] to address them.