Mazino Space — finance domain knowledge, data science, and data engineering notes. https://mazino2d.github.io/ Label design, scorecards, vintage deep-dives, and closing the loop from score to portfolio outcome - Every decision discussed in this series — who to approve, at what limit, at what rate, when to intervene, when to tighten policy — is ultimately informed by a model. This final part connects the analytical tools behind those decisions to the portfolio outcomes they are designed to influence. The... Tue, 14 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-14-6-risk-models/ https://mazino2d.github.io/2026-04-14-6-risk-models/ Policy levers, vintage monitoring, concentration risk, and stress testing - Measuring a portfolio is not the same as managing it. Part 4 gave you the instruments; this part describes how to use them. Portfolio management is fundamentally about making calibrated decisions under uncertainty — when to tighten credit policy, when to investigate an anomaly, how to distribute risk across channels... Tue, 14 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-14-5-portfolio-management/ https://mazino2d.github.io/2026-04-14-5-portfolio-management/ Volume, credit quality, and profitability — the three layers every portfolio manager tracks - If Part 1 defined the product and Parts 2–3 described how individual loans behave, this part assembles the view from above — the metrics that let a portfolio manager understand the health, scale, and economics of an entire book. These metrics are organized in three layers that answer three distinct... Tue, 14 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-14-4-portfolio-metrics/ https://mazino2d.github.io/2026-04-14-4-portfolio-metrics/ From origination to charge-off — and why the first payment is the strongest signal you have - Every installment loan traces the same arc: it is originated, disbursed, repaid (or not), and eventually closed — either through full repayment or through the accounting event called charge-off. Understanding this lifecycle is not just theoretical. Each stage generates different risk signals and calls for different management responses. The Lifecycle... Tue, 14 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-14-3-loan-lifecycle/ https://mazino2d.github.io/2026-04-14-3-loan-lifecycle/ The four archetypes that determine how your portfolio performs under stress - A lending portfolio is not a single homogeneous pool of risk. It is a mixture of borrower types, each with different repayment behaviors, different responses to hardship, and different long-term value to the business. Treating them as identical is the first mistake a portfolio manager can make. Why Segmentation Matters... Tue, 14 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-14-2-borrower-segmentation/ https://mazino2d.github.io/2026-04-14-2-borrower-segmentation/ Installment loan anatomy, revenue model, cost structure, and why this is a thin-margin business - Before you can manage a lending portfolio, you need to understand the product itself — how money moves, where revenue comes from, what it costs to run, and why the margin is thinner than it looks. This part builds the economic foundation that every subsequent part will reference. Anatomy of... Tue, 14 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-14-1-product-economics/ https://mazino2d.github.io/2026-04-14-1-product-economics/ Scoring, vintage analysis, roll rates, and CLTV — the quantitative foundation of portfolio management - The strategies in Part 5 work because they are backed by models. Scoring tells you who to approve and at what terms. Vintage analysis tells you whether your approvals were good decisions. Roll rate analysis tells you where the portfolio is heading. CLTV tells you what a customer relationship is... Thu, 02 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-02-6-risk-models-analytics/ https://mazino2d.github.io/2026-04-02-6-risk-models-analytics/ The levers that drive profitability, control risk, and extend account relationships - Metrics tell you what is happening. Strategies are what you do about it. Credit card portfolio management has a well-defined toolkit of interventions across the lifecycle — from credit line adjustments to collections segmentation to vintage-level stress testing. Credit Line Management (CLM) The credit limit is the most direct lever... Thu, 02 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-02-5-management-strategies/ https://mazino2d.github.io/2026-04-02-5-management-strategies/ The KPI framework for managing a credit card book from acquisition to charge-off - A credit card portfolio generates enormous amounts of data. The challenge is not measuring everything — it is knowing which metrics matter, when they matter, and what they tell you. This part builds the complete measurement framework across the lifecycle. Acquisition Metrics Approval rate — the percentage of applications that... Thu, 02 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-02-4-portfolio-metrics/ https://mazino2d.github.io/2026-04-02-4-portfolio-metrics/ From underwriting to recovery — the full arc of a credit card account - A credit card account is not a static asset. It moves through distinct stages over its life, and each stage demands different management actions. Understanding the lifecycle — and having the right strategies at each point — is the operational core of portfolio management. The Full Credit Card Lifecycle Acquisition... Thu, 02 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-02-3-lifecycle-management/ https://mazino2d.github.io/2026-04-02-3-lifecycle-management/ Transactors, revolvers, hibernators, and defaulters — the four archetypes that shape portfolio economics - Not all cardholders behave the same way. A portfolio that looks healthy in aggregate can contain a very different mix underneath. Understanding that mix — and how it shifts — is the foundation of effective portfolio management. The Four Behavioral Archetypes Transactor A transactor pays the full statement balance every... Thu, 02 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-02-2-customer-segmentation/ https://mazino2d.github.io/2026-04-02-2-customer-segmentation/ How credit cards work, who profits, and why the risk is two-sided - Before you can manage a credit card portfolio, you need to understand what you are actually managing: a financial product that sits at the intersection of consumer credit, payment infrastructure, and multi-party economics. How a Credit Card Works A credit card is a revolving credit product — the bank grants... Thu, 02 Apr 2026 00:00:00 +0700 https://mazino2d.github.io/2026-04-02-1-product-overview/ https://mazino2d.github.io/2026-04-02-1-product-overview/ From trustworthy data to clear rollout decisions - This final lesson turns analysis into action through a three-step framework. Step 1: Trust the Experiment? Before discussing uplift, verify trust conditions: No major SRM No severe tracking asymmetry Baseline behavior is explainable No single-day dependency for conclusion If trust fails, do not ship based on this run. Step 2:... Wed, 11 Mar 2026 00:00:00 +0700 https://mazino2d.github.io/2026-03-11-6-decision-framework/ https://mazino2d.github.io/2026-03-11-6-decision-framework/ Twyman's Law, underpowered tests, peeking, and overdue experiments - This lesson covers the mistakes that often look harmless in the moment but can quietly destroy decision quality. Twyman’s Law: Too Good to Be True When a result looks extraordinary, suspect instrumentation before celebrating. Signals: Unrealistically large uplift Extremely tiny p-values with weak business rationale Sudden jumps inconsistent with historical... Wed, 11 Mar 2026 00:00:00 +0700 https://mazino2d.github.io/2026-03-11-5-pitfalls-part-2/ https://mazino2d.github.io/2026-03-11-5-pitfalls-part-2/ Lucky day traps, SRM, base-rate mismatch, and data loss - Even perfect statistics cannot fix broken experiment data. In this lesson, we focus on high-impact quality failures. Lucky Day Trap A result can appear significant mainly because of one anomalous day. Warning signs: Most lift comes from a single date window Removing one day changes the conclusion entirely Traffic behavior... Wed, 11 Mar 2026 00:00:00 +0700 https://mazino2d.github.io/2026-03-11-4-pitfalls-part-1/ https://mazino2d.github.io/2026-03-11-4-pitfalls-part-1/ How to design experiments that can actually detect real effects - Many teams run experiments that cannot detect meaningful effects. This is a planning failure, not an analysis failure. What Is Statistical Power? Power is the probability of detecting a real effect of a specific size: \[\text{Power} = 1 - \beta\] Typical target: 80% or 90%. Low power means you can... Wed, 11 Mar 2026 00:00:00 +0700 https://mazino2d.github.io/2026-03-11-3-power-and-sample-size/ https://mazino2d.github.io/2026-03-11-3-power-and-sample-size/ P-values, confidence intervals, and Type I/II errors - In Part 1, we defined the mechanics of an A/B test. Now we answer the hard question: when is a difference convincing? P-Value in Plain Language A p-value answers: If the null hypothesis were true, how likely is data this extreme (or more extreme)? Small p-values indicate that the observed... Wed, 11 Mar 2026 00:00:00 +0700 https://mazino2d.github.io/2026-03-11-2-significance/ https://mazino2d.github.io/2026-03-11-2-significance/ Core concepts: hypotheses, conversion rates, and random variation - A/B testing is a disciplined way to make product decisions using evidence instead of intuition. In this first lesson, you will build the mental model needed for every experiment that follows. What Is an A/B Test? An A/B test compares two variants: Variant A: current experience (control) Variant B: new... Wed, 11 Mar 2026 00:00:00 +0700 https://mazino2d.github.io/2026-03-11-1-foundations/ https://mazino2d.github.io/2026-03-11-1-foundations/ Why doing well with money has little to do with intelligence and everything to do with behavior - The Psychology of Money: Timeless Lessons on Wealth, Greed, and Happiness by Morgan Housel explores the complex relationship between money and human behavior. Unlike traditional finance books that focus on technical analysis, this book reveals how psychology, emotions, and personal experiences shape our financial decisions. In this blog, I will... Sat, 13 Dec 2025 00:00:00 +0700 https://mazino2d.github.io/2025-12-13-psychology-of-money/ https://mazino2d.github.io/2025-12-13-psychology-of-money/ How tiny changes create remarkable results and why 1% better every day matters - Atomic Habits: An Easy & Proven Way to Build Good Habits & Break Bad Ones by James Clear is a practical guide to understanding how habits work and how to change them. The book provides a proven framework for improving your life through small, incremental changes. In this blog, I... Fri, 12 Dec 2025 00:00:00 +0700 https://mazino2d.github.io/2025-12-12-atomic-habits/ https://mazino2d.github.io/2025-12-12-atomic-habits/