LOB-Impact shows emergent market behavior in generative models
October 18, 2026
Research into generative limit order book models reveals that market impact emerges naturally in autoregressive models trained on next-token prediction. This allows these models to serve as more realistic world models for training trading agents.
HOW THIS AFFECTS YOU
●
researcherThis validates using next-token prediction models as high-fidelity simulators for financial market physics.
●
investorThis signals potential advances in highly realistic market simulation tools for quantitative trading.