AI-LOB addresses the need for Limit Order Book (LOB) forecasting systems in financial markets, where market makers, hedge funds, proprietary trading firms, and funds require accurate and reliable predictions to manage inventory, identify arbitrage opportunities, and optimize execution. Our solution reduces risks and improves returns for high-frequency market makers and traders and optimizes the execution of large orders for institutional investors, providing a more stable environment for all market participants.
AI-LOB builds on our machine learning models for LOB, published in leading AI journals, widely recognized, and among the most cited in the field. Examples are TABL model (cited >300 times) and recently developed LOBERT model (NeurIPS workshop best poster award). Our approach is to commercialize a low-latency AI framework that allows users to fine-tune, train, optimize, validate, test, and implement state-of-the-art ML foundation models. These models forecast LOB messages and features, generating real-time signals within short-term opportunity windows. The benefits of our solution lie in the fast, fine-tunable, and accurate models. The models can be integrated into clients’ existing in-house or commercial trading systems, directly translating into financial gains and improved profitability.
Given the highly international nature of the market, our business has strong global scalability, with significant growth opportunities in major financial hubs where demand for advanced AI-driven trading solutions is high.
Funding source
Contact persons
Juho Kanniainen
Professor, Computing Sciences
Juho Kanniainen