A pattern needs its conditions.
Similar events can lead to different results under different conditions. We check where a pattern holds and where it breaks.
ABOUT REGULUS
We started with market data.
For Regulus Technology, AI transformation (AX) means giving people more room for creative work. We build tools for that purpose using what we learned developing and operating quant systems.
We checked the result, then revised the code.
With Merlu, we studied market data, including Bitcoin charts. A recurring movement might have been coincidence, so we formed hypotheses about the conditions under which it appeared.
We spent about 18 months building the quant system and about two years operating it. When a rule behaved differently than expected, we checked the conditions and exceptions and revised the code.
Market data and software work are different problems. The practice of dividing work into smaller steps, running them, and checking the result carried over. We are using it to build Asterism.
We want AI to carry out defined tasks and help check the results. People need time to define problems, connect ideas from different fields, and decide what to try next.
How we work.
Start with the smaller task.
We divide complex work into steps we can run. Then we check which procedures can be used elsewhere and connect them.
People choose the next direction.
Even when AI executes tasks and organizes results, people decide what the problem is and what to build next.
The builders also run the system.
We handle design through operation and fix problems found in use.
Sungdong Bae
배성동Founder · Product & systems design
Built and operated quantitative automation systems; leads product direction and core system development.
Dusan Baek
백두산Development · Operations
Has built and maintained enterprise software; works on the product and customer projects.