Research & Publications
Discover our latest research contributions in Vision Language Action for trading systems, scientific publications, and ongoing studies that push the boundaries of intelligent trading and financial AI.
FinFlowRL: An Imitation-Reinforcement Learning Framework for Adaptive Market-Making
ConferenceDeveloped two-stage framework combining MeanFlow generative policy networks with PPO for high-frequency market-making, achieving superior performance in adaptive trading strategies.
FinMem: A Performance-Enhanced Large Language Model Trading Agent with Layered Memory
JournalDesigned layered memory system for LLM trading agents improving decision accuracy and computational efficiency, demonstrating state-of-the-art performance in financial decision-making.
FlowHFT: Imitation Learning via Flow Matching Policy for Optimal High-Frequency Trading
PreprintPioneered flow matching approach for high-frequency trading achieving state-of-the-art performance in optimal execution and market-making strategies.
FlowOE: Imitation Learning with Flow Policy from Ensemble RL Experts for Optimal Execution
PreprintBuilt generative imitation learning framework achieving superior cost efficiency compared to classical execution baselines through ensemble reinforcement learning experts.
ByteGen: A Tokenizer-Free Generative Model for Orderbook Events
PreprintCreated novel 32-byte packed order book event model with hybrid H-Net and Mamba-Transformer architecture for realistic market simulation and orderbook prediction.
TradingGPT: Multi-agent System with Layered Memory and Distinct Characteristics
ConferenceDeveloped multi-agent reasoning system for trading decision support with retrieval-augmented capabilities, combining financial text understanding with structured market data encoding.
