iGEN
Visit IGEN World Explore IGEN Expo
EXPLORE UPGRADE PLANS
BREAKING
Inside the rogue ChatGPT hack of Hugging Face: AI agents operate at superhuman speed but make clumsy mistakes Landstar Expects to Emerge a Winner After Supreme Court’s Montgomery Ruling Widens Broker Liability New Senate bill targets 'chameleon carriers' that reopen to escape penalties Werner Enterprises Posts Highest Revenue Per Truck Growth in One-Way Segment in a Decade CMA CGM and Stonepeak Launch United Ports LLC in $2.4 Billion Terminal Joint Venture UPS shift away from Amazon shows bigger payoff Lanesurf: 62% of Loads Get Vetted Carrier Offers Before Brokers Arrive India-China Border Trade Via Lipulekh Resumes Aug 1; China Permits 20 Traders Geopolitics Drives CMA CGM Q2 Profit Surge of 42% as Volumes and Rates Climb Benchmark Diesel Price Rises Third Week as Futures Plunge; Spread Hits Record Inside the rogue ChatGPT hack of Hugging Face: AI agents operate at superhuman speed but make clumsy mistakes Landstar Expects to Emerge a Winner After Supreme Court’s Montgomery Ruling Widens Broker Liability New Senate bill targets 'chameleon carriers' that reopen to escape penalties Werner Enterprises Posts Highest Revenue Per Truck Growth in One-Way Segment in a Decade CMA CGM and Stonepeak Launch United Ports LLC in $2.4 Billion Terminal Joint Venture UPS shift away from Amazon shows bigger payoff Lanesurf: 62% of Loads Get Vetted Carrier Offers Before Brokers Arrive India-China Border Trade Via Lipulekh Resumes Aug 1; China Permits 20 Traders Geopolitics Drives CMA CGM Q2 Profit Surge of 42% as Volumes and Rates Climb Benchmark Diesel Price Rises Third Week as Futures Plunge; Spread Hits Record
Home ›› Technology ›› Ai ›› Llms ›› MoCA-Agent: Market-of-Claims Code Agent Achieves Strong Results in Financial and Numerical Reasoning

MoCA-Agent: Market-of-Claims Code Agent Achieves Strong Results in Financial and Numerical Reasoning

The arXiv paper introduces MoCA-Agent, a market-of-claims code agent that decomposes questions into atomic claims and uses trader agents to buy or sell those claims. It achieved strong performance on ten benchmarks including FinQA (78.3%), FinanceMath (76.0%), and FinChart-Bench (85.6%).

iG
iGEN Editorial
June 20, 2026
MoCA-Agent: Market-of-Claims Code Agent Achieves Strong Results in Financial and Numerical Reasoning

Financial and tabular question answering systems often produce plausible but incorrect results due to misread cells or erroneous operations. A single mistake in unit, sign, or scale can silently undermine a high-stakes decision. To address this, the computer science article on arXiv titled "MoCA-Agent: A Market-of-Claims Code Agent for Financial and Numerical Reasoning" introduces a system that replaces free-form multi-agent debate with disciplined, claim-level verification.

How MoCA-Agent Works

According to the arXiv paper, MoCA-Agent decomposes each question into typed atomic claims. Specialist trader agents then buy or sell those claims based on their confidence. The system clears the orders into confidence-weighted accept or reject decisions. From the market-supported evidence, it synthesizes an executable Python program. A code-aware verifier checks the program for execution correctness, structural consistency, and common financial reasoning errors. The system allows at most one market-aware repair round if the initial program fails verification.

This approach shifts the aggregation of evidence from whole answers to atomic claims, which improves robustness in numerical reasoning tasks.

Performance Benchmarks

The paper reports results across ten public benchmarks covering financial numerical reasoning, general tabular reasoning, ESG question answering, and multimodal chart reasoning. Using a fixed Qwen3.6-27B backbone, MoCA-Agent achieved the following accuracy scores:

Benchmark Accuracy
FinQA 78.3%
FinanceMath 76.0%
MultiHiertt 71.2%
ESGenius 86.9%
FinChart-Bench (average) 85.6%

These results demonstrate that claim-level market mechanisms can outperform traditional multi-agent debate approaches in high-stakes numerical reasoning.

Implications for High-Stakes Reasoning

The paper's authors, including Abdallah, Abdelrahman, Elmadany, AbdelRahim A, Natour, Sameh Al, Hasan Cavusoglu, Adam Jatowt, and Muhammad Abdul-Mageed, designed MoCA-Agent specifically for domains where answers must be grounded in exact facts, formulas, units, signs, and scales. The system's ability to catch common financial reasoning errors through its code-aware verifier and market-based claim verification could make it suitable for applications such as automated financial analysis, audit support, and regulatory compliance.

The code and data for MoCA-Agent are publicly available, as noted in the paper. While the evaluation focuses on financial and numerical benchmarks, the methodology of claim-level verification could extend to other domains requiring precise, grounded reasoning.

By aggregating evidence at the level of atomic claims rather than whole answers, MoCA-Agent offers a robust alternative for error-prone numerical question answering, with potential to reduce costly mistakes in enterprise decision-making.


Sources:

Keep Reading

Recommended Stories

Vocabulary Dropout Technique Prevents Diversity Collapse in LLM Co-Evolution Training Technology

Vocabulary Dropout Technique Prevents Diversity Collapse in LLM Co-Evolution Training

A new method called vocabulary dropout prevents diversity collapse in co-evolutionary LLM training. Applied to Qwen3 models on mathematical reasoning, it improved solver performance by an average of 4.4 points, with largest gains on competition-level benchmarks.

June 16, 2026
New Research Shows Pretraining Data Composition Can Engineer Neural Scaling Laws for Particle Physics Technology

New Research Shows Pretraining Data Composition Can Engineer Neural Scaling Laws for Particle Physics

A new arXiv paper demonstrates that neural scaling laws in particle physics can be engineered by adjusting pretraining data composition. The study shows that including more diverse and task-aligned synthetic data can shift scaling behavior to require more data rather than larger models, offering insights for efficient AI training.

July 8, 2026
LLM Paraphrase Augmentation Boosts Sign Language Translation Performance Technology

LLM Paraphrase Augmentation Boosts Sign Language Translation Performance

A new study proposes using a large language model (GPT-4o) to generate controlled paraphrase variants of training targets for sign language translation (SLT). Evaluated on three datasets, the method yields a modest BLEU-4 improvement on PHOENIX14T and reveals gains in semantic fidelity not captured by lexical metrics.

June 21, 2026
Reinforcement-Aware Knowledge Distillation Boosts LLM Reasoning Efficiency Technology

Reinforcement-Aware Knowledge Distillation Boosts LLM Reasoning Efficiency

Researchers propose RL-aware distillation (RLAD) to address distribution mismatch and objective interference in knowledge distillation for LLM reasoning. The method uses Trust Region Ratio Distillation (TRRD) to selectively imitate teacher policies during reinforcement learning. RLAD outperforms offline distillation, standard GRPO, and KL-based on-policy distillation across logic and math benchmarks.

June 21, 2026