TIP-Search: Time-Predictable Inference Scheduling for Market Prediction under Uncertain Load
It addresses strict latency demands for high-frequency financial systems, offering a robust solution for market prediction under uncertainty, though it appears incremental as it builds on existing scheduling and model selection concepts.
This paper tackles the problem of real-time market prediction under uncertain workloads by proposing TIP-Search, a time-predictable inference scheduling framework that dynamically selects deep learning models to maximize accuracy while meeting deadlines, achieving up to 8.5% accuracy improvement and 100% deadline satisfaction on three real-world datasets.
This paper proposes TIP-Search, a time-predictable inference scheduling framework for real-time market prediction under uncertain workloads. Motivated by the strict latency demands in high-frequency financial systems, TIP-Search dynamically selects a deep learning model from a heterogeneous pool, aiming to maximize predictive accuracy while satisfying per-task deadline constraints. Our approach profiles latency and generalization performance offline, then performs online task-aware selection without relying on explicit input domain labels. We evaluate TIP-Search on three real-world limit order book datasets (FI-2010, Binance BTC/USDT, LOBSTER AAPL) and demonstrate that it outperforms static baselines with up to 8.5% improvement in accuracy and 100% deadline satisfaction. Our results highlight the effectiveness of TIP-Search in robust low-latency financial inference under uncertainty.