FOUNDATION
Exploration
The founding team started from zero, exploring fully automated trading. The first-generation quant model began trading proprietary capital.
YEAR ONE
Year One
A dedicated quantitative trading model was established, investing through mathematics and artificial intelligence. The team set its ambition: to become a world-class quantitative hedge fund.
FIRST AI MODEL
The first AI model
In April 2016 the first stock position generated by a deep-learning model went live, computed on GPUs. Before this, algorithms relied on linear models and classical machine learning, running on CPUs.
DIRECTION
AI becomes the direction
Complex models hit the limits of single-machine training. Growing demand collided with finite compute; the team began building its own GPU cluster.
FULL AI
Strategies fully AI-based
The AI algorithm and hardware teams expanded continuously. By the end of 2019 almost all quantitative strategies were computed by AI models.
GLOBAL
The year of going global
Trading moved beyond China A-shares. The Tianma T+0 model was founded as an independent effort, with Tan Daoyi as CEO of the research team, and the Hong Kong Hang Seng market as the primary back-testing ground.