CRAFT Funds New Research Projects Advancing Financial AI and Tokenized Markets

The Center for Research toward Advancing Financial Technologies (CRAFT) at Stevens Institute of Technology continues to expand its role as a leading hub for industry-connected fintech research through the funding of two new faculty-led projects approved at CRAFT’s nineth and 10th Industry Advisory Board (IAB) meetings.

CRAFT is a university-industry research center housed within the Stevens School of Business that brings together faculty, students and corporate partners to address critical challenges in financial technology, artificial intelligence, blockchain, analytics, and digital markets. Through its Industry Advisory Board model, CRAFT member organizations collaborate directly with researchers to identify emerging industry problems and fund high-impact applied research projects with real-world relevance.

The newly funded projects reflect two of the most important areas shaping the future of finance: trustworthy AI systems for financial regulation and next-generation infrastructure for tokenized securities markets.

Internalizing Financial Regulations for Financial LLM Agents

Funded at CRAFT’s 10th IAB Meeting (Spring 2026) Funding Amount: $100,000 1 Year Full Research Project Principal Investigator: Denghui Zhang; Co-Principal Investigator: Steve Yang

Financial AI systems are increasingly being used to support accounting, auditing and compliance workflows. However, current large language model (LLM) systems struggle to reliably interpret and apply complex financial regulations such as FASB’s ASC 606 and ASC 842. Existing approaches often require feeding entire regulatory texts into the model for every query — making systems expensive, inefficient and unreliable when multiple regulations interact simultaneously.

“Financial LLM agents need to reason with complex financial regulations, but repeatedly placing long regulatory texts into prompts is costly, inefficient and fragile,” says PI Denghui Zhang. “Our project, Internalizing Financial Regulations for Financial LLM Agents, explores a different path, a more token-efficient approach by helping models internalize key regulatory knowledge rather than relying only on long-context retrieval at inference time. The goal is to build financial AI agents that are more efficient, reliable and regulation-aware for high-stakes accounting and compliance tasks”.

The research team’s preliminary testing highlights the severity of the challenge: while leading models such as GPT-4o can perform adequately on isolated rule checks, performance drops dramatically when evaluating complex multi-rule compliance scenarios.

This project seeks to fundamentally improve how financial AI systems reason about regulation by teaching AI agents to internalize financial rules directly within the model itself. The team will develop:

• A public benchmark for evaluating AI reasoning on financial regulations

• A training framework for embedding financial regulations into language models

• An open-source financial compliance agent capable of performing realistic accounting and auditing tasks

The project aims to create faster, more reliable and lower-cost compliance AI systems that could directly benefit audit firms, financial institutions and RegTech companies across the CRAFT network.

Algorithmic Market Making for Tokenized Securities

Funded at CRAFT’s 9th IAB Meeting (Fall 2025) Funding Amount: $30,000 Six-Month Pilot Project Principal Investigator: Zachary Feinstein; Co-Principal Investigator: Ionut Florescu

As tokenized securities markets continue to grow toward a projected multi-trillion-dollar opportunity, major challenges remain in the stability and reliability of decentralized trading platforms. Existing systems often struggle to maintain accurate pricing and sustainable liquidity, limiting institutional adoption and increasing risk for investors.

“The current market for tokenized securities is caught in a catch-22: a lack of liquidity leads to unstable prices, yet that very instability deters the institutional capital needed to provide liquidity”, says PI Zachary Feinstein. “We are looking at this problem because existing automated market makers are information agnostic and unsuitable for assets that have a strong external price reference. By building an 'Oracle-Aware' framework, we can ensure on-chain prices remain accurate and stable. As tokenization moves toward a multi-trillion-dollar reality, our work will establish the new standards for market design and provide the open-source tools necessary to accelerate global adoption”.

This project addresses these foundational issues by developing a new generation of decentralized market-making mechanisms connected to reliable real-world pricing data. The research will establish mathematical frameworks for more stable, efficient, and trustworthy tokenized asset exchanges.

By improving the underlying market infrastructure for digital securities, the project has the potential to accelerate adoption of tokenized financial products while reducing systemic risks associated with emerging digital asset markets.

Expanding Research Opportunities for Students

In addition to advancing leading-edge fintech research, both projects will provide opportunities for Stevens students to participate directly in high-impact research initiatives. The project teams expect to recruit and engage graduate and undergraduate students to assist with research, development, data analysis, modeling and implementation activities throughout the duration of the projects.

These opportunities reflect CRAFT’s broader mission of connecting students with industry-relevant research experiences that prepare them for careers at the intersection of finance, technology, AI and analytics.

As CRAFT continues to expand its portfolio of industry-supported research initiatives, the center encourages faculty across Stevens to engage with the IAB process and explore opportunities for future project funding and collaboration.