SEBI's PolicyTwin Prototype Wins First Place at C:>DIR Global 'Agentic Regulator' Hackathon
Securities and Exchange Board of India (SEBI), represented by Team Pheonix, has secured first prize at the C:>DIR Global 'Agentic Regulator' Hackathon for their prototype called "Policy Twin". The hackathon was organized by Financial Innovation for Impact (Fii), an international not-for-profit organization that spun out from the Cambridge Centre for Alternative Finance at the Cambridge Judge Business School.
The hackathon was conducted from July 8 to September 18, 2026, bringing together regulators, policymakers, AI researchers, academics, financial institutions, and FinTechs to build practical agentic AI solutions. The event drew 336 submissions involving 942 individuals from more than 65 countries, with 36 teams reaching the Final Round. Based on presentation and live voting, the top six teams were selected for final jury evaluation.
Team Pheonix comprised four SEBI officials: Shreya Chaturvedi (AGM), Suchita Jain (AGM), Apoorva Singh Rathore (Manager), and Shahityaa D (Manager). The winning team received a prize money of USD 35,000 and has been invited to present their solution at the Singapore FinTech event scheduled for November 2026.
PolicyTwin is a prototype of a virtual "practice market" that enables financial market regulators to test proposed policies before implementation. Using artificial intelligence, it creates a realistic, computer-generated version of the securities market with AI agents representing various market participants including retail investors, large institutions, brokers, high-speed traders, and potential bad actors. These agents place orders as they might in real markets, allowing regulators to enter draft rules and run simulations to see how they could affect trading, settlement, investor losses, and market fairness.
The prototype specifically checks for warning signs such as market manipulation, shortages of margin money, heavy losses for small investors, and misuse of automated trading. If safety limits are crossed, it tests safeguards including freezing collateral and slowing down order flow. Throughout the process, humans remain in control with regulators able to review and change every assumption before simulation runs. Findings and recommendations are presented as evidence but never acted on automatically, with officials able to provide feedback, re-run simulations, and compare results.
PolicyTwin is currently at the prototype stage and is planned to be adopted by SEBI after refinement and thorough testing. The winning team was felicitated by the Chairman of SEBI for their outstanding achievement.