Practice Notebook – Overview

Why this notebook exists

The premise is simple: your institution already has data, models and policies. What is often missing is a learning layer that remembers what you try, how those decisions perform through time and stress, and where you still have room to stretch without betraying your risk appetite. This notebook is where that layer takes shape in human language.

Most portfolios “learn” in a fragmented way. One team runs a pilot, another adjusts a cut-off, a third negotiates a new collection strategy. Six months later, the institution remembers that something was tried, but not exactly what, for whom, or with what consequence. The result is a kind of institutional amnesia: experience is rich, but memory is poor.

The Practice Notebook assumes the opposite posture: every meaningful change in who you lend to, how you price, how you collect, and how you steer under stress is treated as an experiment with a hypothesis, a capacity limit and a link to stress and capital.

How to read this notebook

You do not have to read the chapters in order, but there is a quiet logic in the sequence shown in the sidebar.

• Overview – this page – explains the frame: why we talk about a learning book at all.
• False Negative Revival shows how to recover near-miss borrowers safely, by turning “borderline cases” into designed corridors.
• Rates & Pricing Dynamics treats rates as bands and price as the full package, making clear which segments can carry what, and why.
• Collections Lift reorganises delinquency from “more pressure” into a friction map: product misfit, cash-flow stress, execution gaps.
• Stress Navigation connects all of the above to forward-looking stress views, so growth and caution speak the same language.
• Integrity Shield covers the lending perimeter: how to reduce avoidable leakage without creating unnecessary friction for clean customers.
• Insurance Pricing – Term Life extends the logic to premium bands, persistency and claims, where horizons are longer but the trade-offs are similar.
• Adaptive Layer describes the learning engine itself: hypotheses, experiments, logs and guardrails.
• ROI Lens translates the practice into numbers the board actually cares about: risk-adjusted income, capital use and growth.
• Partner Handoff shows how all this is packaged as APIs, dashboards and routines that can live alongside your existing systems.
• Terminology is the appendix where we make our vocabulary explicit and precise.

If you are a CEO, you may start with revival, stress and ROI. If you are in risk, you may start with guardrails and the adaptive layer. If you run a P&L, you might go first to rates, collections and insurance. Every path is allowed; the learning layer is the same underneath.

The lending book as a learning system

In a conventional set-up, pilots and parameter tweaks live and die in slides. This notebook takes the opposite view: it treats each deliberate change as part of a learning system. AltVector’s practice anchors that system around four families of decisions:

• Who you lend to – the false negatives you revive and the segments you prioritise.
• On what terms – the rate and price bands you offer, including insurance.
• How you collect – the way you respond to different frictions in delinquency.
• How you steer under stress – the way you react to comfort, watch and action zones.

You are not committing to endless tinkering. You are committing to remembering why you did what you did, and to comparing that decision against a clear counterfactual: what would the book have looked like without this lever?

Corridors, bands and guardrails

Across the chapters you will see the same shapes repeat:

• Corridors – narrow zones where you allow a different decision logic (for example, reviving a slice of declines under stricter terms).
• Bands – rate or premium ranges that are sensible for a given segment, given your appetite and stress views.
• Guardrails – hard limits on volume, movement and exposure without new approvals.

False Negative Revival will show corridors along the decline–approve boundary. Rates & Pricing and Insurance will show bands along the yield and premium axes. Collections and Stress Navigation will show how these corridors and bands behave when the cycle turns.

A learning book becomes governable when the board can ask, at any time, three simple questions:

1. Which corridors and bands are we currently running?
2. What guardrails protect us if they misbehave?
3. What have we learned from the last cycle of decisions?

The notebook is designed to make those questions answerable without digging through disparate decks and emails.

Who owns this notebook?

Formally, ownership will sit with a joint cell between Risk, Business and Finance. Practically, the most senior sponsor matters: the person who is willing to say, “We will lend more, but only in ways we can explain calmly three years from now.”

AltVector’s role is to provide the scaffolding:

• The analytics that identify corridors, bands, frictions and stress trajectories.
• The learning engine that logs experiments, reads outcomes and proposes adjustments.
• The structure you see in this Practice Notebook.

But the real work – deciding which corridors to open, which bands to move, which guardrails to enforce – remains firmly in your hands. This is your book, your capital and your reputation. The notebook exists so that your decisions can compound into institutional knowledge, not just individual experience.

If you only have twenty minutes

If you are short on time, here is a pragmatic path through the Practice Notebook:

1. Read this Overview to anchor the idea of a learning book.
2. Go to False Negative Revival to see how a concrete lever is designed.
3. Jump to Stress Navigation to understand how all levers tie into forward-looking comfort, watch and action zones.
4. Glance at the ROI Lens chapter to see how the economics stack up.
5. Keep Terminology handy as an appendix; it will save time in meetings.

The rest of the chapters deepen the same logic. Taken together, they are a proposal for how a modern lender can lend more, risk less – not by magic, but by tightening the link between decisions, stress and memory.