This issue of The Journal of Risk Model Validation brings together some rather disparate enquiries into topics in model validation and stress testing/backtesting. All our readers should find something of interest here. The flow of submissions continues to be healthy, and it is clear that our journal has much to offer the industry. The mix of excellent academic work and practical advice that The Journal of Risk Model Validation boasts is unusual among finance journals, which tend to focus on just one or the other.
Our first paper, “An optimized support vector machine intelligent technique using optimized feature selection methods: evidence from Chinese credit approval data” by Mohammad Zoynul Abedin, Chi Guotai, Fahmida-E-Moula, Tong Zhang and M. Kabir Hassan, deals with machine-learning models and their validation. This is an interesting and rather novel issue, with machine learning very much in vogue in the field of quantitative finance at the moment. It is seen as a procedure for man- aging complex and sometimes-hidden patterns of risk that are not detectable by conventional methods of risk modeling. I expect further papers in this area in future.
The issue’s second paper, “Model risk tiering: an exploration of industry practices and principles”, is by Nick Kiritz, Miles Ravitz and Mark Levonian. It addresses issues around current consultancy and regulatory recommendations for risk model validation. Tiering is an exercise whereby models are ranked by the inherent risk associated with their use. The authors’ focus on tiering is justified as an a priori exercise in resource allocation. Much has already been written by consultants on the subject, and this paper provides a very useful summary.
“Credit portfolio stress testing using transition matrixes”, our third paper, is by Radu Neagu, Gabriel Lipsa, Jing Wu, Jake Lee, Stephane Karm and John Jordan. The authors propose a new methodology for modeling credit transition probability matrixes using macroeconomic factors. This methodology is tested over a period including the last two major financial crises and is shown to perform well. We remind readers that stress testing and backtesting are topics that are very much of interest to the journal.
With that reminder in place, we come to the final paper in this issue: “Validation of the backtesting process under the targeted review of internal models: practical recommendations for probability of default models” by Lukasz Prorokowski. This paper provides practical recommendations for validating a backtesting process under the targeted review of internal models (TRIM) regulatory exercise. It offers advice on the introductory steps for validating a backtesting process and reviews the available statistical tests for calibration, discrimination and stability backtesting, According to the European Central Bank, "TRIM focuses on the reduction of unwarranted variability in risk-weighted assets (RWA) driven by inappropriate modeling which takes advantage of the freedom granted by the current regulation", This research fills a gap in the literature and is of both academic interest and practical value.
Trinity College, University of Cambridge
An optimized support vector machine intelligent technique using optimized feature selection methods: evidence from Chinese credit approval data
This paper focuses on feature selection methods for support vector machine (SVM) classifiers, checking their optimality by comparing them with some statistical and baseline methods.
This paper seeks to shed light on one critical area of such frameworks: model risk tiering, or the rating of risk inherent in the use of individual models, which can benefit a firm’s resource allocation and overall risk management capabilities.
In this paper, the authors propose a new methodology for modeling credit transition probability matrixes (TPMs) using macroeconomic factors.
Validation of the backtesting process under the targeted review of internal models: practical recommendations for probability of default models
This paper provides practical recommendations for the validation of the backtesting process under the targeted review of internal models (TRIM).