评分卡与金融风险评估研究综述
本文总结了评分卡的发展历程、金融风险评估模型的最新趋势、前沿技术应用、行业挑战及垂直领域研究成果
一、评分卡本质与发展
1. 起源与目标 (Fisher 1936)
核心：评分卡核心目标是最优区分度（非点估计）
2. 思维转变 (Kaplan, Norton, et al. 1992)
核心：从区分转向多维度平衡，提供企业全面视角（财务+运营指标），含"驾驶舱"概念
二、金融风险评估模型趋势
1. 传统模型偏好 (Lu and Zhang 2023)
核心：授信模型偏好可解释算法（逻辑回归/决策树）；反欺诈/反洗钱倾向复杂模型（图算法/GNN）
2. 行为序列创新 (Qiao et al. 2024)
核心：首创商户级折叠支付行为序列，增强违约风险检测
3. 数据颗粒度争议 (Whitrow et al. 2009; Jing, Gao, and Zeng 2023)
核心：交易级分类 vs 交易聚合策略
4. 自监督学习应用 (Liu et al. 2022)
核心：设计三任务学习欺诈知识（行为/意图/序列级）
三、前沿技术应用
1. 动态图融合 (Yuan et al. 2025)
核心：融合动态交易图（时序特征）与静态转账图（资金闭环），检测微信套现行为
2. 概念漂移应对 (Qiao et al. 2025)
核心：TRE方法通过测试时检索相似样本动态适应漂移
3. 关联规则挖掘 (sanchez2009associationrule?)
核心：用模糊关联规则识别信用卡交易欺诈模式
4. 多任务学习 (Liang et al. 2021)
核心：MvMoE（Multi-view-aware Mixture-of-Experts network）模型同步解决信用风险预测（Credit Risk Forecasting, CRF）和信用额度设定（Credit Limits Setting, CLS）问题
四、行业挑战与系统风险
1. 异常检测类型 (chalapathy2019?)
核心：点/上下文/集体异常区别 + 新奇性检测（非异常）
2. 系统性风控短板
- 套现导致数据污染、资金空转(Yuan et al. 2025)
- 经济资本新算法：账户级还款、余额建模优于PD预测（djeundje2025?）
3. 人行征信体系（未标注文献）
核心：5维度量化风险（画像/能力/意愿/习惯/动态）
五、垂直领域研究
1. 反欺诈综述 (Wang 2023)
核心：垂直关联（用户潜在交互）+ 水平关联（行为关系）
2. FTG 特征 (Huang, Wang, and Xiong 2022)
本质是一种固定效应。
3. 京东营销反作弊体系
以"感知-识别-拦截-复盘"闭环管理为核心，实时+离线模型双轨驱动，运用多模型融合（半监督/无监督）精准识别欺诈，平衡风险拦截（TPR）与用户体验（FPR）。
4. 番茄风控大数据 (2025)
信贷业务痛点：资金闲置、营销低效、逆向选择。意愿模型通过双场景（授信/预授信）分层优化额度定价与营销ROI，降低CPS成本。
参考文献
- Fisher, Ronald A. 1936. “The Use of Multiple Measurements in Taxonomic Problems.” Annals of Eugenics 7 (2): 179–88.
- Huang, Ho-Chuan, Xiuhua Wang, and Xin Xiong. 2022. “When Macro Time Series Meets Micro Panel Data: A Clear and Present Danger.” Energy Economics 114. https://doi.org/https://doi.org/10.1016/j.eneco.2022.106289.
- Jing, Phoebe, Yijing Gao, and Xianlong Zeng. 2023. “A Customer-Level Fraudulent Activity Detection Benchmark for Enhancing Machine Learning Model Research and Evaluation.” In IEEE International Conference on e-Business Engineering (ICEBE), 47–54. IEEE.
- Kaplan, Robert S, David P Norton, et al. 1992. “The Balanced Scorecard: Measures That Drive Performance.”
- Liang, Ting, Guanxiong Zeng, Qiwei Zhong, Jianfeng Chi, Jinghua Feng, Xiang Ao, and Jiayu Tang. 2021. “Credit Risk and Limits Forecasting in e-Commerce Consumer Lending Service via Multi-View-Aware Mixture-of-Experts Nets.” In Proceedings of the Fourteenth ACM International Conference on Web Search and Data Mining, 9. ACM.
- Liu, Can, Yuncong Gao, Li Sun, Jinghua Feng, Hao Yang, and Xiang Ao. 2022. “User Behavior Pre-Training for Online Fraud Detection.” In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’22), 9. ACM.
- Lu, Tian, and Yingjie Zhang. 2023. “1 + 1 > 2? Information, Humans, and Machines.” Arizona State University, Peking University. https://ssrn.com/abstract=4045718.
- Qiao, Yiran, Yateng Tang, Xiang Ao, Qi Yuan, Ziming Liu, Chen Shen, and Xuehao Zheng. 2024. “Financial Risk Assessment via Long-Term Payment Behavior Sequence Folding.” arXiv Preprint arXiv:2411.15056.
- Qiao, Yiran, Ningtao Wang, Yuncong Gao, Yang Yang, Xing Fu, Weiqiang Wang, and Xiang Ao. 2025. “Online Fraud Detection via Test-Time Retrieval-Based Representation Enrichment.” In The Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI-25), 12470–76. https://doi.org/10.1609/aaai.v39i1.24717.
- Wang, Cheng. 2023. Anti-Fraud Engineering for Digital Finance. Tongji University Press.
- Whitrow, C., D. J. Hand, P. Juszczak, D. Weston, and N. M. Adams. 2009. “Transaction Aggregation as a Strategy for Credit Card Fraud Detection.” Data Mining and Knowledge Discovery 18 (1): 30–55.
- Yuan, Qi, Yang Liu, Yateng Tang, Xinhuan Chen, Xuehao Zheng, Qing He, and Xiang Ao. 2025. “Dynamic Graph Learning with Static Relations for Credit Risk Assessment.” In Proceedings of the Association for the Advancement of Artificial Intelligence. Association for the Advancement of Artificial Intelligence.
- 番茄风控大数据. 2025. “意愿模型全解析：构建逻辑、应用策略与业务进化.” 微信公众号. June 20, 2025. https://mp.weixin.qq.com/s/TEphBNZPoyLuXelKFQ1oLg.