Spot the Gap (Exercise)

The brief. A new submission lands on your desk: Léman Assurances SA (Lausanne), Personal Accident XL 2027, one layer 3m xs 2m at 6.5% rate on line, share offered 100%, subject premium 18m — and no stated loss ratio anywhere in the pack. Derive what can be derived; return null for what cannot. The battery fails you for inventing a zero.

This is the null-honesty principle from Lesson 1 at working size, and it is the exact behaviour of the product's SubmissionMetrics scope: the combined ratio is derived only when all four components are on file — stated loss ratio, brokerage, ceding commission, expense ratio. A missing figure must surface as a data-quality referral to the Head of Line of Business (Lesson 2 turns that into a rule), never as CR = 0 — which would be the most aggressive possible stance taken on the worst possible data.

The rules

Premium100 must still work: 195,000 (0.065 × 3m). Missing economics never block premium math.
CombinedRatio returns double? — the sum of all four components when all are present, else null.
With brokerage 0.10 and expense 0.06 on file but loss ratio absent, the answer is null — and the battery's referral line must read "Data-quality referral → Head of LoB".
A control case with loss ratio 0.70 on file must yield 0.86.

Your workbench

using MeshWeaver.Layout;
using MeshWeaver.Layout.Composition;
using MeshWeaver.Utils;

record Pack(string Name, double Limit, double RateOnLine, double? LossRatio, double? Brokerage, double? CedingCommission, double? ExpenseRatio);

// YOUR METRICS — mirror the product's SubmissionMetrics scope.
static double Premium100(Pack p)
    => 0.0;                 // TODO: rate on line × limit — never blocked by missing economics.

static double? CombinedRatio(Pack p)
    => 0.0;                 // TODO: lr + brokerage + ceding commission + expense — ONLY when all four are on file, else null.

var leman = new Pack("Léman PA XL 2027 — as received", 3_000_000, 0.065, null, 0.10, 0.0, 0.06);
var control = new Pack("Control — with LR 0.70 on file", 3_000_000, 0.065, 0.70, 0.10, 0.0, 0.06);

var premium = Premium100(leman);
var crLeman = CombinedRatio(leman);
var crControl = CombinedRatio(control);

var premiumOk = Math.Abs(premium - 195_000) < 0.5;
var lemanOk = crLeman is null;
var controlOk = crControl is double c && Math.Abs(c - 0.86) < 1e-9;

Controls.Stack
    .WithView(Controls.H4($"Null-honesty battery — {(premiumOk ? 1 : 0) + (lemanOk ? 1 : 0) + (controlOk ? 1 : 0)}/3 checks green"))
    .WithView(Controls.Markdown(
        $"- Premium @100% = **{premium:N0}** (must be 195,000) {(premiumOk ? "✓" : "✗")}\n" +
        $"- Léman CR = **{(crLeman is double v ? v.ToString("P0") : "null")}** (must be null) {(lemanOk ? "✓" : "✗")} — " +
        (lemanOk ? "**Data-quality referral → Head of LoB**" : "an invented figure would decide a deal on data nobody supplied") + "\n" +
        $"- Control CR = **{(crControl is double w ? w.ToString("P0") : "null")}** (must be 86%) {(controlOk ? "✓" : "✗")}"))

The stub fails the Léman check on purpose: it returns 0.0, which is the aggressive-zero bug. Your fix is not a bigger formula — it is refusing to answer.

All three green? Reveal the solution — the pattern-match that keeps null null.

← Lesson 1 exercises · Lesson 1 — The submission