PersonaLedger: Generating Realistic Financial Transactions with Persona Conditioned LLMs and Rule Grounded Feedback

TL;DR

PersonaLedger addresses the lack of public, high-fidelity financial transaction data by combining a persona-conditioned LLM with a rule-grounded, stateful engine to generate diverse yet internally consistent transaction streams. The system maintains explicit ledger state and enforces accounting invariants, using a closed-loop interaction where the LLM proposes transactions and the engine applies $cash_{t+1}=cash_t+income_t-payment_t$ and $bal_{t+1}=bal_t+spending_t-payment_t+interest_t+fees_t$ with $0\le bal_{t+1}\le limit$, among others, to prevent drift. The authors release a public dataset of 30 million transactions from 23,000 personas, plus benchmarks for illiquidity classification and identity theft segmentation, along with full generation logs for reproducibility. This work provides a scalable, privacy-preserving resource for rigorous evaluation of forecasting and anomaly detection in financial AI and lays groundwork for extending rules and markets without retraining.

Abstract

Strict privacy regulations limit access to real transaction data, slowing open research in financial AI. Synthetic data can bridge this gap, but existing generators do not jointly achieve behavioral diversity and logical groundedness. Rule-driven simulators rely on hand-crafted workflows and shallow stochasticity, which miss the richness of human behavior. Learning-based generators such as GANs capture correlations yet often violate hard financial constraints and still require training on private data. We introduce PersonaLedger, a generation engine that uses a large language model conditioned on rich user personas to produce diverse transaction streams, coupled with an expert configurable programmatic engine that maintains correctness. The LLM and engine interact in a closed loop: after each event, the engine updates the user state, enforces financial rules, and returns a context aware "nextprompt" that guides the LLM toward feasible next actions. With this engine, we create a public dataset of 30 million transactions from 23,000 users and a benchmark suite with two tasks, illiquidity classification and identity theft segmentation. PersonaLedger offers a realistic, privacy preserving resource that supports rigorous evaluation of forecasting and anomaly detection models. PersonaLedger offers the community a rich, realistic, and privacy preserving resource -- complete with code, rules, and generation logs -- to accelerate innovation in financial AI and enable rigorous, reproducible evaluation.

Figures (8)

• Figure 1: Example demonstration. Our PersonaLedger dataset consists of user personas (Left) and their corresponding transactions and payments sequences (Right).
• Figure 2: Example demonstration. Left: LLM reasons about the trajectory plan. Right: the generation is based on the trajectory plan. Personas and generated transactions are strongly bounded.
• Figure 3: Average monthly spending and error bars grouped by different persona attributes.
• Figure 4: Average monthly/daily spending and error bars grouped by different temporal features.
• Figure 5: A case study of unrealistic transactions from a naive prompting baseline with Llama-3.3-70B.