1. Develop future return prediction models based on short-term bar and order book data.

2. Employ nonlinear methods (tree-based models, neural networks, RNNs, Transformers, etc.) to uncover robust signals.

3. Process large-scale tick and K-line data from multiple exchanges, establishing a reliable research workflow.

4. Build a long-term iterative feature engineering framework to explore effective market microstructure factors.

5. Use research outputs as an auxiliary branch to enhance main process signals and deploy into production.

Requirements

1. Experience researching predictive models for 1–5 minute or shorter intervals.

2. Familiarity with nonlinear modeling frameworks and proven feature engineering techniques.

3. Capability to handle hundreds of millions of records or terabytes of data with solid engineering expertise.

4. In-depth understanding of order book dynamics, market microstructure, and short-term market behavior.

5. Experience in mid- to high-frequency trading or market making is preferred.

Preferred Qualifications

Ability to independently build research pipelines.

Demonstrable models or signal generation results.

Established understanding of strategy implementation and model stability.