Pith. sign in

REVIEW 4 major objections 4 minor 34 references

Quantum Powered Credit Risk Assessment: A Novel Approach using hybrid Quantum-Classical Deep Neural Network for Row-Type Dependent Predictive Analysis

T0 review · 4 major / 4 minor · reviewed 2026-08-09 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims a hybrid quantum-classical network, split by loan type, reaches 83% accuracy on personal loans and 81% on agricultural loans; simulator limits forced a cut to five principal components.

desk verdict Reported accuracies sit below the majority-class baseline and no classical comparator is given, so the quantum 'enhancement' claim is unsupported; the paper is transparent but thin. read the letter →

arxiv 2502.07806 v1 pith:3FLTWIQZ submitted 2025-02-06 q-fin.CP cs.AIcs.LG

Rath Minati , Date Hema This is my paper · ORCID
classification q-fin.CPcs.AIcs.LG
keywords quantumdeeplearningcreditriskassessmenthybridquantum-classicalneuralnetworkrow-typedependentpredictiveanalysisprincipalcomponentSMOTEdataaugmentationloanclassification
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper is trying to establish that credit risk classification can be improved by separating a loan portfolio by loan type and running each slice through a hybrid quantum-classical neural network. The proposed framework, HyQuC-DeepNN-RTDPA, encodes the first five principal components of each loan type into quantum states, processes them through entangling quantum layers, and feeds the measurements into classical dense layers. On a bank dataset of over 25,000 agriculture and personal loans, the authors report accuracies of 0.8348 for personal loans and 0.8112 for agricultural loans. The paper states plainly that quantum simulator limits forced the dimensionality reduction from 38-43 principal components down to five, and that the model's minority-class F1 scores remain low. The contribution is framed as a feasibility demonstration of quantum-classical integration for credit risk, not as a claim of industry-wide superiority.

What carries the argument

The central object is the hybrid circuit of Figure 3: classical features are angle-embedded into n qubits, processed by parameterized strongly entangling layers, and measured as Pauli-Z expectation values that feed dense neural-network layers. Around it sit RTDPA, the rule that a separate model is trained per loan type, and SMOTE, which creates synthetic minority-class samples before training. The quantum layer is the component the paper credits with extracting correlations that the classical layers then use for classification.

What would settle it

Run a purely classical deep network on the same five principal components and a classical network on the full 38-43 components; if either matches or beats the hybrid model's accuracy, the claim that quantum feature extraction improves credit risk assessment is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that a row-type-dependent hybrid quantum-classical deep neural network can produce workable credit risk predictions, with test accuracy 0.8348 for personal loans and 0.8112 for agricultural loans after PCA, SMOTE augmentation, and per-loan-type training. The quantum component angle-embeds the five PCA features into qubit states, applies strongly entangling layers, and measures Pauli-Z expectation values that become inputs to fully connected classical layers. Because each loan type has its own trained model, the network is meant to capture the distinct risk profiles of personal and agricultural loans. The authors are explicit that the evaluation is constrained: the full 38-43 component feature space could not be used because the quantum simulator could not handle it, and the minority classes (Sub Standard, and Loss in agriculture) show F1 scores as low as 0.1011.

Load-bearing premise

The load-bearing premise is that the first five principal components of each loan-type dataset preserve enough information to separate risk classes, even though the paper chose them only because its quantum simulator could not handle the full 38-43 components.

Editorial extensions

If this is right

  • If the framework holds, banks can tune a separate credit-risk model for each loan product and slot a quantum feature-extraction layer into existing deep-learning pipelines.
  • The reported test accuracies (0.8348 personal, 0.8112 agricultural) give a concrete feasibility point for hybrid quantum-classical models on today's simulators.
  • Because minority-class F1 scores are low (0.1600 for personal Sub Standard, 0.1011 for agricultural Sub Standard), the current model is not yet usable for the very decisions where risk is concentrated.
  • The reliance on five principal components means the full benefit of the row-type split remains untested until simulators or encoding methods can handle 38-43 components.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • We infer the paper leaves implicit that, without a classical-only baseline trained on the same five principal components, the reported accuracies cannot be attributed to the quantum layer; the reader should test that baseline before crediting quantum features.
  • We infer the PCA truncation is the main confound: if the risk-separating signal lives in components beyond the first five, neither the quantum nor the classical part of the model could see it, making any quantum advantage untestable on this evidence.
  • We infer the RTDPA idea is separable from quantum computing: a purely classical deep network trained per loan type on the full features might match or beat the hybrid system, which would mean the row-type split is the active ingredient.
  • We infer a natural next experiment is to keep the same five principal components and classical layers but swap angle embedding for amplitude encoding, since the paper names this as future work and it would isolate the encoding's contribution.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. This paper proposes a hybrid quantum-classical deep neural network for credit risk classification, with separate models for personal and agricultural loans under a framework called Row-Type Dependent Predictive Analysis (RTDPA). The workflow preprocesses each loan type separately, applies PCA for dimensionality reduction, encodes the reduced features into a parameterized quantum circuit via angle embedding, and feeds the measurement results into classical dense layers. On a proprietary dataset from a named-but-anonymized bank, the authors report test accuracies of 0.8348 for personal loans and 0.8112 for agricultural loans, along with per-class precision, recall, F1-score, ROC AUC, and Cohen's kappa. The paper also discusses computational constraints, lists limitations, and includes a toy example of hybrid training in the appendix.

Significance. The proposed QRTDPA architecture is a plausible template for combining per-loan-type preprocessing with a quantum feature map followed by classical layers, and the manuscript is unusually candid about its limitations. If the hybrid model were shown to outperform classical baselines on the original features, the contribution would be of interest to the quantum-finance community. However, the empirical evidence does not establish such an improvement: the reported accuracies are below the trivial majority-class baseline, no classical comparator is provided, and the input space is truncated to five principal components. The candid limitations section is a strength, but it concedes exactly the issues that block the central claim. No code or data are provided, and the novelty relative to the authors' prior work [3, 21] appears incremental.

major comments (4)
  1. [§8, Tables 3–4] The headline accuracies are below the trivial majority-class baseline, and no comparator is provided. In Table 3, the personal-loan test set has 870 Standard instances out of 932; an all-Standard classifier would achieve 93.3% accuracy, whereas the model reports 83.5%. In Table 4, the agriculture test set has 3,525 Standard instances out of 4,116; the majority baseline is 85.6%, above the reported 81.1%. Because the paper does not report a classical deep neural network on the same five components, an ablation without the quantum layer, or any other baseline, the abstract's claim that the hybrid framework 'enhance[s] the accuracy' of credit risk evaluation is not supported by the evidence.
  2. [§7.3 and §10] The evaluation is performed on the first five principal components only, chosen post hoc because the quantum simulator could not handle the 38–43 components suggested by the scree plots. Section 7.3 states this explicitly, and Section 10 concedes that this reduction 'may have resulted in the loss of critical information' for distinguishing loan statuses. Since no classical model is evaluated on the same five-component representation and no experiment uses the full feature set, the reported accuracies cannot be attributed to the quantum layer; they are, at best, properties of a drastically reduced input space.
  3. [§6, §7.4, §10] There is an internal contradiction about whether SMOTE was used. The algorithm in Section 6 (Step 2) and the framework description say class imbalance is addressed through SMOTE and data augmentation, but Section 10 states that SMOTE and its variants 'were not incorporated into our analysis' and are only suggestions for future work. Section 7.4 describes why augmentation is useful but does not report any augmentation actually applied to the data. This makes the experimental procedure ambiguous and prevents reproducibility of the class-imbalance handling.
  4. [§8.1, Table 3; §9] The reported metrics are internally inconsistent and lack statistical support. In Table 3, Training Accuracy is 0.6734 while Validation Accuracy and Test Accuracy are both 0.8348; the text describes the lower training accuracy as 'potential overfitting,' which is the opposite of the usual interpretation, and no separate test-set description or cross-validation details are given. Section 9 explains that hardware constraints prevented confidence intervals, but without repeated runs or error bars the single-point accuracies in Tables 3–4 cannot be distinguished from noise, especially for minority classes with supports of 30–53.
minor comments (4)
  1. [§3] The paragraph beginning 'The versatility and adaptability of deep learning algorithms...' appears verbatim twice in Section 3.
  2. [Throughout] There are several typographical issues, including 'Synthetic Minority Over-sampling Technique Synthetic Minority Over-sampling Technique' and the broken abbreviation 'HyQuC-DeepNN-R TDP A'; a careful proofreading pass is needed.
  3. [Appendix 12.1.1] The matrix labeled RZ(0.325) is a real rotation matrix, not the standard RZ gate, and the parameter-shift example yields a zero gradient for both data points, so the toy example does not effectively demonstrate the training step.
  4. [§2, References [3], [20], [21]] The literature gap is asserted largely through the authors' own prior work; independent corroborating references for the claimed novelty would strengthen the positioning.

Circularity Check

2 steps flagged · score 4.0 of 10

The empirical results are not circular, but the paper's novelty framing reduces to a self-cited RTDPA concept that renames standard subgroup modeling.

  1. renaming known result [Section 1 (RTDPA definition) and Section 2 (Literature Review gap claim)]
    "By analyzing loan subcategories individually, predictive models can be tailored to the specific attributes of each loan type. This results in more precise risk assessments and a deeper understanding of potential vulnerabilities within a credit portfolio [3]. ... most existing research on hybrid quantum-classical models for credit risk assessment lacks a focus on Row-Type Dependent Predictive Analysis (RTDPA)"

    RTDPA is defined in the paper as the practice of analyzing loan subcategories individually and tailoring models to each type, which is the standard idea of segmented or conditional modeling. The paper then claims a research gap because existing work 'lacks a focus on RTDPA.' That gap is true only because RTDPA is a newly coined acronym for a known modeling strategy; the novelty claim is manufactured by renaming rather than by a new derivation. The quantum experiment itself is independent, but the claimed conceptual gap is circular.

  2. self citation load bearing [Section 1, Introduction (RTDPA citation [3]); Section 2, Literature Review (gap claim)]
    "RTDPA represents a shift in credit risk assessment by recognizing that different loan types exhibit distinct characteristics and risk profiles. ... This results in more precise risk assessments and a deeper understanding of potential vulnerabilities within a credit portfolio [3]. ... Our proposed framework aims to fill this gap ..."

    The only cited basis for RTDPA and for the asserted gap in the literature is the authors' own prior arXiv paper [3]. The paper uses this self-citation to establish both the central organizing concept and the claim that existing hybrid quantum-classical credit-risk research lacks it. Thus the load-bearing novelty premise is supported by a self-referential citation chain rather than by external evidence. The reported accuracy numbers do not derive from [3], so this circularity affects the framing of the contribution, not the empirical measurement itself.

full rationale

The reported accuracies (0.8348 personal, 0.8112 agricultural) come from training the hybrid model on PCA-reduced data and are not derived from, or equal to, any fitted input by construction; no equation in the paper defines the test metrics as the model's own assumptions. The absence of a classical baseline and the fact that both accuracies are below the majority-class baselines are serious evidential weaknesses, but they belong to correctness risk, not circularity. The circularity that exists is at the framing level: the paper's 'novel' RTDPA concept is defined in the authors' own prior work [3], and the literature gap it claims to fill ('most existing research ... lacks a focus on RTDPA') is true only because RTDPA is a self-coined name for the standard practice of fitting separate models to loan-type subgroups. The quantum encoding methods are also attributed to the authors' own [21], though that is an external comparative study and does not supply the reported numbers. Hence the empirical core is independent, but the novelty/gap argument reduces to a self-citation chain and a renaming of segmented modeling.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central empirical claims rest on three load-bearing choices: the RTDPA split (from the authors' prior work), the 5-PC reduction imposed by simulator limits, and the unvalidated assumption that a quantum embedding layer improves feature extraction. None is tested against alternatives, and the paper's own limitations section concedes most of these points.

free parameters (2)
  • Number of principal components retained = 5
    Chosen because the quantum simulator could not handle 38-43 components (Section 7.3), not from data. This affects all downstream results and is admitted as a limitation.
  • Final quantum circuit and training hyperparameters = not reported
    The algorithm lists a grid search space in Section 6 Step 8, but the final selected values for qubits, layers, learning rate, batch size, and epochs are never given, making the model under-specified.
assumptions (4)
  • domain assumption Different loan types have distinct risk profiles and should be modeled separately (RTDPA premise)
    Introduced in Section 1 and the literature review; no statistical test is provided to show per-loan-type models outperform pooled models.
  • ad hoc to paper The first 5 principal components preserve enough information for credit risk classification
    Section 7.3 and Section 10 admit this reduction may lose critical information; it is driven by simulator limits, not by data.
  • domain assumption Angle embedding and StronglyEntanglingLayers provide a useful quantum feature representation for tabular data
    Assumed in the Section 6 algorithm and in Section 8.3; no comparison against other embeddings or classical feature maps is provided.
  • domain assumption Standard PCA assumptions (linear correlations, orthogonal components) hold for this dataset
    Stated in Section 10 as a limitation; if false, the dimensionality reduction is suboptimal.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Quantum Powered Credit Risk Assessment: A Novel Approach using hybrid Quantum-Classical Deep Neural Network for Row-Type Dependent Predictive Analysis." pith.science (2026). https://pith.science/paper/3FLTWIQZ

@misc{pith2026250207806,
  author       = {Pith},
  title        = {Pith review of: Quantum Powered Credit Risk Assessment: A Novel Approach using hybrid Quantum-Classical Deep Neural Network for Row-Type Dependent Predictive Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3FLTWIQZ}},
  note         = {Machine review of arXiv:2502.07806}
}
read the original abstract

The integration of Quantum Deep Learning (QDL) techniques into the landscape of financial risk analysis presents a promising avenue for innovation. This study introduces a framework for credit risk assessment in the banking sector, combining quantum deep learning techniques with adaptive modeling for Row-Type Dependent Predictive Analysis (RTDPA). By leveraging RTDPA, the proposed approach tailors predictive models to different loan categories, aiming to enhance the accuracy and efficiency of credit risk evaluation. While this work explores the potential of integrating quantum methods with classical deep learning for risk assessment, it focuses on the feasibility and performance of this hybrid framework rather than claiming transformative industry-wide impacts. The findings offer insights into how quantum techniques can complement traditional financial analysis, paving the way for further advancements in predictive modeling for credit risk.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 27 canonical work pages

  1. [3]

    Adaptive Modelling Approach for Row-Type Dependent Predictive Analysis (RTDPA): A Framework for Designing Machine Learning Models for Credit Risk Analysis in Banking Sector

    Minati Rath and Hema Date. Adaptive modelling approach for row-type dependent predictive analysis (rtdpa): A framework for designing machine learning models for credit risk analysis in banking sector. https://arxiv.org/abs/2311.10799, 2023

    work page Pith review arXiv 2023

  2. [21]

    Quantum data encoding: a comparative analysis of classical-to- quantum mapping techniques and their impact on machine learning accuracy

    Minati Rath and Hema Date. Quantum data encoding: a comparative analysis of classical-to- quantum mapping techniques and their impact on machine learning accuracy. EPJ Quantum Technology, 11(1):72, 2024

    work page 2024

  3. [17]

    Egger, Ricardo Garcia Gutierrez, Jordi Cahue Mestre, and Stefan Woerner

    Daniel J. Egger, Ricardo Garcia Gutierrez, Jordi Cahue Mestre, and Stefan Woerner. Credit risk analysis using quantum computers. IEEE Transactions on Computers , 70(12):2136 – 2145, 2021. Cited by: 30; All Open Access, Green Open Access

    work page 2021

  4. [1]

    Machine learning-driven credit risk: a systemic review

    Si Shi, Rita Tse, Wuman Luo, Stefano D’Addona, and Giovanni Pau. Machine learning-driven credit risk: a systemic review. Neural Computing and Applications , 34(17):14327–14339, September 2022

    work page 2022

  5. [2]

    Quantum Computation and Quantum Information, 2000

    MA Nielsen and IL Chuang. Quantum Computation and Quantum Information, 2000

    work page 2000

  6. [4]

    Marqu´ es, V

    A.I. Marqu´ es, V. Garc ´ ıa, and J.S. S´ anchez. A literature review on the application of evolu- tionary computing to credit scoring. Journal of the Operational Research Society , 64(9):1384 – 1399, 2013. Cited by: 69; All Open Access, Green Open Access

    work page 2013

  7. [5]

    A comparative study of forecasting corporate credit ratings using neural networks, support vector machines, and decision trees

    Parisa Golbayani, Ionut ¸ Florescu, and Rupak Chatterjee. A comparative study of forecasting corporate credit ratings using neural networks, support vector machines, and decision trees. The North American Journal of Economics and Finance , 54:101251, 2020

    work page 2020

  8. [6]

    Ferreira- Martins, and Samurai Brito

    Sohum Thakkar, Skander Kazdaghli, Natansh Mathur, Iordanis Kerenidis, Andr´ e J. Ferreira- Martins, and Samurai Brito. Improved financial forecasting via quantum machine learning. Quantum Machine Intelligence , 6(1):27, 2024

    work page 2024

Show all 34 references
  1. [7]

    Quantum machine learning for finance

    Maria Schuld, Mark Fingerhuth, and Francesco Petruccione. Quantum machine learning for finance. arXiv preprint arXiv:1907.03044 , 2019

    1907 arXiv

  2. [8]

    P.W. Shor. Algorithms for quantum computation: discrete logarithms and factoring. In Proceedings of the 35th Annual ACM Symposium on Theory of Computing (STOC) , 2002

    2002

  3. [9]

    Lov K. Grover. A fast quantum mechanical algorithm for database search. In Proceedings of the Annual ACM Symposium on Theory of Computing , volume Part F1294, pages 212–219. Association for Computing Machinery, jul 1996

    1996

  4. [10]

    Jha and L

    A. Jha and L. Cincio. Quantum algorithms for classical machine learning tasks. Physical Review Letters, 124(12):120504, 2020

    2020

  5. [11]

    Farhi and H

    E. Farhi and H. Neven. Classification with quantum neural networks on near term processors. arXiv preprint arXiv:1802.06002 , 2018. 28

    2018 arXiv

  6. [12]

    Quantum algorithms for supervised and unsupervised machine learning

    Seth Lloyd, Masoud Mohseni, and Patrick Rebentrost. Quantum algorithms for supervised and unsupervised machine learning. arXiv preprint arXiv:1307.0411 , 2013

    2013 arXiv

  7. [13]

    Quantum Machine Learning in Feature Hilbert Spaces

    Maria Schuld and Nathan Killoran. Quantum Machine Learning in Feature Hilbert Spaces. Physical Review Letters, 122(4):040504, feb 2019

    2019

  8. [14]

    Quantum neural networks: Concepts, applications, and challenges

    Yunseok Kwak, Won Joon Yun, Soyi Jung, and Joongheon Kim. Quantum neural networks: Concepts, applications, and challenges. In 2021 Twelfth International Conference on Ubiqui- tous and Future Networks (ICUFN) , pages 413–416, 2021

    2021

  9. [15]

    Quantum neural networks

    Jacob Biamonte, Ville Bergholm, Peter Wittek, Nicola Pancotti, Patrick Rebentrost, and Seth Lloyd. Quantum neural networks. arXiv preprint arXiv:1711.11240 , 2017

    2017 arXiv

  10. [16]

    Supervised learning with quantum-enhanced feature spaces

    Vojtˇ ech Havl ´ ıˇ cek, Andr´ es M C´ ardenas-Valencia, and Al´ an Aspuru-Guzik. Supervised learning with quantum-enhanced feature spaces. Nature, 567(7747):209–212, 2019

    2019

  11. [18]

    A novel dynamic neural system for nonconvex portfolio optimization with cardinality restrictions

    Xinwei Cao and Shuai Li. A novel dynamic neural system for nonconvex portfolio optimization with cardinality restrictions. IEEE Transactions on Systems, Man, and Cybernetics: Systems , 53(11):6943–6952, November 2023

    2023

  12. [19]

    A novel recurrent neural network based online portfolio analysis for high frequency trading

    Xinwei Cao, Adam Francis, Xujin Pu, Zenan Zhang, Vasilios Katsikis, Predrag Stanimirovic, Ivona Brajevic, and Shuai Li. A novel recurrent neural network based online portfolio analysis for high frequency trading. Expert Systems with Applications , 233:120934, 2023

    2023

  13. [20]

    Quantum-assisted simulation: A framework for designing ma- chine learning models in the quantum computing domain

    Minati Rath and Hema Date. Quantum-assisted simulation: A framework for designing ma- chine learning models in the quantum computing domain. https://arxiv.org/abs/2311.10363, 2023

    2023 arXiv

  14. [22]

    Quantum machine learning

    Jacob Biamonte, Peter Wittek, Nicola Pancotti, Patrick Rebentrost, Nathan Wiebe, and Seth Lloyd. Quantum machine learning. Nature, 549(7671):195–202, September 2017

    2017

  15. [23]

    A quantum-inspired classical algorithm for kernel learning

    Sergey Bravyi, Theodoros Kapourniotis, and Barbara Kraus. A quantum-inspired classical algorithm for kernel learning. arXiv preprint arXiv:1611.09347 , 2016

    2016 arXiv

  16. [24]

    K. Cai, Z. Zhang, and X. Ye. Quantum algorithms for linear algebra: Theoretical insights and practical considerations. arXiv preprint arXiv:1901.10796 , 2019

    1901 arXiv

  17. [25]

    Park, Carsten Blank, and Francesco Petruccione

    Matt Lourens, Ilya Sinayskiy, Daniel K. Park, Carsten Blank, and Francesco Petruccione. Hierarchical quantum circuit representations for neural architecture search. npj Quantum Information, 9(1):79, 2023

    2023

  18. [26]

    Quantum computing for finance: Overview and prospects

    Rom´ an Or´ us, Samuel Mugel, and Enrique Lizaso. Quantum computing for finance: Overview and prospects. Reviews in Physics , 4:100028, 2019

    2019

  19. [27]

    Quantum computing for financial risk measurement

    Sascha Wilkens and Joe Moorhouse. Quantum computing for financial risk measurement. Quantum Information Processing, 22(1):51, January 2023. 29

    2023

  20. [28]

    Quantum computing reduces systemic risk in financial networks

    Amine Mohamed Aboussalah, Cheng Chi, and Chi-Guhn Lee. Quantum computing reduces systemic risk in financial networks. Scientific Reports, 13(1):3990, March 2023

    2023

  21. [29]

    Coyle and D

    B. Coyle and D. Leung. Quantum machine learning for financial applications. IEEE Trans- actions on Quantum Engineering , 3:1–10, 2022

    2022

  22. [30]

    Financial credit risk assessment: a recent review

    Ning Chen, Bernardete Ribeiro, and An Chen. Financial credit risk assessment: a recent review. Artificial Intelligence Review , 45(1):1–23, 2016

    2016

  23. [31]

    Credit risk classifica- tion: an integrated predictive accuracy algorithm using artificial and deep neural networks

    Mohammad Mahbobi, Salman Kimiagari, and Marriappan Vasudevan. Credit risk classifica- tion: an integrated predictive accuracy algorithm using artificial and deep neural networks. Annals of Operations Research, 330(1):609–637, November 2023

    2023

  24. [32]

    Li et al

    Y. Li et al. A new approach to data augmentation for credit risk modeling using machine learning. International Journal of Information Technology & Decision Making , 20(4):1277– 1296, 2021

    2021

  25. [33]

    Quantum circuit learning

    Kosuke Mitarai, Makoto Negoro, Masahiro Kitagawa, and Keisuke Fujii. Quantum circuit learning. Physical Review A , 98(3):032309, 2018

    2018

  26. [34]

    Evaluating analytic gradients on quantum hardware

    Maria Schuld, Ville Bergholm, Christian Gogolin, Josh Izaac, and Nathan Killoran. Evaluating analytic gradients on quantum hardware. Physical Review A , 99(3):032331, 2019. T able 5Column Description COLUMN NO COLUMN NAME DESCRIPTION 1 Q Quarter 2 BRCD Branch Code 3 CUSTID Cus...

    2019

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.