CAME: Company-Aware Evidence-Memory Experts for Interpretable Quarter-Ahead Revenue Forecasting

2026-09-27

AI Summary

This study addresses the scaling bias of large language models in quarterly revenue forecasting and their inability to capture timely signals from historical anchors. To overcome these limitations, we propose a residual forecasting framework that integrates statistical anchors, semantic evidence, and error memory. By incorporating leakage-prevention mechanisms and traceable evidence cards, the approach achieves both high predictive accuracy and interpretability. Furthermore, evaluation robustness is ensured through rolling backtesting combined with an evidence-memory expert system. Empirical experiments on 336 company-quarter samples demonstrate that the proposed method attains the lowest aggregate error, significantly outperforming existing baselines.

Abstract

Quarter-ahead revenue forecasting requires company-scale numerical accuracy, strict temporal validity, and company-specific interpretation of narrative disclosures. LLMs can distill textual evidence but can produce scale-misaligned forecasts, whereas history-based anchors are stable but miss forecast-time signals such as product transitions, supply constraints, and management guidance. We introduce CAME (Company-Aware Evidence-Memory Experts), a residual-forecasting framework that refines a no-leakage statistical anchor when current semantic evidence and prior error patterns justify an adjustment. On a development-inclusive rolling backtest of 336 company-quarters from 12 large public technology and platform firms, CAME achieves the lowest aggregate point-estimate error among the reported methods, with statistically supported macro-sMAPE gains over the matched Statistical Anchor, and outperforms History + Guidance on all six aggregate metrics. CAME also links adjustments to source-linked evidence cards and guarded memory traces, supporting forecast inspection, provenance, and failure localization.

Problem

• Revenue Forecasting
• Large Language Models
• Interpretability
• Temporal Validity
• Scale Alignment

Innovation

• Residual Forecasting
• Evidence-Memory Experts
• Interpretable Forecasting
• Revenue Forecasting
• LLM