Anjan Roy
Agentic AI &
Portfolio Decision Systems
Portfolio Management as an Evolving Decision System
I apply Agentic AI and rule-based architectures to financial decision environments. My work focuses on the structural reality of markets: non-stationarity, assumption decay, and the necessity of rigorous feedback loops.
The lens is precise: portfolios, derivatives, and systemic risk. Unlike prediction-focused models that attempt to forecast price, decision systems focus on response, control, and adaptation across shifting regimes.
Scope of Practice
- Decision Systems Architecture
- Derivative Risk Control
- Regime-Aware Portfolio Logic
- Human-in-the-Loop Governance
* Note: This work is distinct from signal generation, high-frequency trading, or consumer products.
Areas of Research & Practice
Agentic AI & Decision Systems
- ■Agentic and rule-based system design
- ■Monitoring and feedback architecture
- ■Constraints and effective guardrails
- ■Observability and decision diagnostics
- ■Human-in-the-loop governance models
Portfolios, Derivatives & Risk
- ■Portfolios as dynamic decision systems
- ■Derivatives as control instruments
- ■Volatility, convexity, and payoffs
- ■Risk as system instability, not point estimates
- ■Structuring for survival and compounding
Market Reality & Regimes
- ■Managing assumption decay
- ■Detecting and adapting to regime shifts
- ■Non-stationarity in financial data
- ■Divergence of theory from practice
- ■Fragility of over-optimized models
Program Formats
Executive / Faculty Capsule
Duration: 2–3 Hours
A strategic session designed for faculty alignment and program anchoring. Focuses on why classical assumptions break down in modern markets and how agentic frameworks provide decision discipline.
Key Themes
- • Breakdown of classical assumptions
- • Agentic AI for decision discipline
- • Realistic derivative usage
- • AI value attribution & limits
Agentic AI for Portfolio & Derivatives Decision Systems
Core Immersion • 3 Days (Practitioner-Led)
A deep-dive immersion designed to complement existing finance, quant, or systems curricula. Offline-first, rigorous, and pilot-friendly.
Day 1: Decision Systems
Portfolios as systems, assumption decay, monitoring vs. prediction. Introduction to agents, feedback loops, and human-in-the-loop governance.
Day 2: Risk Limits
Volatility and convexity. Using derivatives to shape risk and stabilize portfolios across regimes. Practical payoff walkthroughs (conceptual).
Day 3: Agentic Integration
Mapping decision rules, monitoring across regimes, risk as system behavior. Failure modes, diagnostics, and responsible deployment.