Kelly Criterion for Optimal Credit Allocation

The purpose of this study is to address the critical issue of optimal credit allocation. Predicting a borrower’s probability of default is a key requirement of any credit allocation system but turning it into labeled classes leads to problems in performance measurement. In this paper the connection between the probability of default and optimal credit allocation is established through a conceptual construct called the Kelly criterion. Conflicting performance measures in dichotomous classification are replaced with coherent criteria for judging the performance of credit allocation decisions. Extensive testing on peer-to-peer lending data shows that the Kelly strategy enables consistent outperformance and efficiency in processing information relative to alternative credit allocation approaches.

[1]  Richi Nayak,et al.  Modeling Credit Risk: A Category Theory Perspective , 2021, Journal of Risk and Financial Management.

[2]  M. Zenga,et al.  Bank loan recovery rates: Measuring and nonparametric density estimation , 2010 .

[3]  William T. Ziemba,et al.  Long-term capital growth: the good and bad properties of the Kelly and fractional Kelly capital growth criteria , 2010 .

[4]  Sandra Paterlini,et al.  Using differential evolution to improve the accuracy of bank rating systems , 2007, Comput. Stat. Data Anal..

[5]  Daniel Rösch,et al.  Bayesian loss given default estimation for European sovereign bonds , 2020 .

[6]  Markus Viljanen,et al.  Predicting Expected Profit in Ongoing Peer-to-Peer Loans with Survival Analysis-Based Profit Scoring , 2019, KES-IDT.

[7]  C. E. SHANNON,et al.  A mathematical theory of communication , 1948, MOCO.

[8]  Edward I. Altman,et al.  FINANCIAL RATIOS, DISCRIMINANT ANALYSIS AND THE PREDICTION OF CORPORATE BANKRUPTCY , 1968 .

[9]  Peter Winker,et al.  Optimization heuristics for determining internal rating grading scales , 2010, Comput. Stat. Data Anal..

[10]  Turgay Celik,et al.  Statistical and machine learning models in credit scoring: A systematic literature survey , 2020, Appl. Soft Comput..

[11]  Michal Polena,et al.  Best classification algorithms in peer-to-peer lending , 2020 .

[12]  Jian Chen,et al.  Credit Card Fraud Detection Using Sparse Autoencoder and Generative Adversarial Network , 2018, 2018 IEEE 9th Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON).

[13]  Vlachostergios Eleftherios.  Basel Risk Weight Functions and Forward-Looking Expected Credit Losses , 2018 .

[14]  Josef Tvrdík,et al.  Hybrid differential evolution algorithm for optimal clustering , 2015, Appl. Soft Comput..

[15]  Tanja Verster,et al.  A proposed benchmark model using a modularised approach to calculate IFRS 9 expected credit loss , 2020 .

[16]  J. J. Kelly.  A new interpretation of information rate , 1956 .

[17]  Zhurong Zhou,et al.  Credit Scoring Model based on Kernel Density Estimation and Support Vector Machine for Group Feature Selection , 2018, 2018 International Conference on Advances in Computing, Communications and Informatics (ICACCI).

[18]  J. Suykens,et al.  Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research , 2015, Eur. J. Oper. Res..

[19]  C. Chu,et al.  Predicting the Loss Given Default Distribution with the Zero-Inflated Censored Beta-Mixture Regression that Allows Probability Masses and Bimodality , 2020 .

[20]  Florian Kaposty,et al.  Predicting loss given default in leasing: A closer look at models and variable selection , 2020 .

[21]  Roger M. Stein.  The relationship between default prediction and lending profits: Integrating ROC analysis and loan pricing , 2005 .