WHAT WE DO?

Lanna Digital Tech is a quantitative technology company built on large-scale data and driven by engineering and algorithms.

We adhere to a scientific, rational research paradigm, emphasizing verifiability and controlled risk. Leveraging mathematical statistics, machine learning, and deep neural networks, we construct and train trading models on multi-scale, high-frequency market sequences. We benchmark against major indices, pursuing stable, reusable excess returns through continuous iteration and research-engineering optimization.

WHO WE ARE?

Lanna Digital Tech’s core team is interdisciplinary, covering the full stack from AI research to trading engineering. Team backgrounds include core R&D at large tech and fintech platforms, quantitative hedge funds and exchange infrastructure, top universities in fintech/computer science/mathematics, and long-term practice in data engineering, low-latency systems, and model deployment.

HOW WE DO?

Our underlying philosophy is “react, not predict,” aligning trading risk control with multi-horizon sequence modeling. On the risk side, we replay historical stress and set parameter bounds matched to maximum drawdown tolerances, combine millisecond-level volatility detection with disciplined stop-losses, and prioritize protection against unpredictable shocks. On the signal side, we use multi-timescale price-channel models (hybrid RNN/LSTM and Transformer) to estimate and update the probabilistic bounds for the next few hours, entering contrarian positions during sentiment shifts. Execution is event-driven with instantaneous matching as the decision aligner, augmented by multi-factor robustness filters, forming an integrated “signal–decision–execution” loop. With this framework, we have achieved about 86% model win rate in out-of-sample and live-online execution, and, with end-to-end risk control and millisecond stop-losses, maintained portfolio-level average win rates around 96% in historical statistics (past results, not a guarantee).