Optimization Score
An engine-derived 0–100, computed deterministically from your recommendations — never hand-set. Here is exactly what it is made of.
Execution — value captured
ProgressShare of the open actionable value already banked — Bayeto-validated hypotheses plus observed external optimization. The basis excludes held, superseded, stale, regressed and beyond-payback-policy moves on BOTH sides, so implementing a move the engine bands “not economic” cannot raise this.
Rises as hypotheses pass the measured loop. FALLS when new actionable opportunity is identified — deliberately: more found means more to do. That is discovery working, not the system worsening; the euro headroom beside this grade is the number that grew.
Evidence strength (open moves)
Portfolio qualityValue-weighted derived confidence across the open actionable moves. Weighted by value so one small low-confidence find cannot drag the grade of a large well-evidenced portfolio.
Higher when the euros still on the table are carried by well-evidenced moves. Reads 1 when nothing actionable is open — there is no open move left to carry weak evidence.
Context readiness (open moves)
Portfolio qualityValue-weighted context completeness across the open actionable moves — how much of the evidence that could change each conclusion the engine could establish.
Rises as you declare context and connect sources — the same dimensions the recommendation pages count.
Benchmark position
Portfolio qualityillustrativeWhere your benchmarked metrics sit between the fintech median and top quartile. Context only — the cohort is a hand-set reference distribution, so this carries no weight in the score.
Higher when your system already sits closer to top-quartile on benchmarked metrics. Does not raise the score.
The total is the weighted sum of the 3 components that carry weight — Execution — value captured and Evidence strength (open moves) and Context readiness (open moves). Execution moves as validated hypotheses bank value — and deliberately NOT when a move the engine bands beyond the payback policy is implemented, in either direction. Evidence strength and context readiness describe the open moves, weighted by value: finding one small low-evidence opportunity cannot drag the grade of a large well-evidenced portfolio, and declaring context raises readiness. Execution FALLS when new actionable value is found — more found means more to do; that is discovery working, and the euro headroom beside the gauge is the number that grew.
Benchmark position carries no weight and is shown as context. Its cohort is a hand-set reference distribution, not a measurement of your peers — and a score this page calls engine-derived cannot be moved by a number we chose.
When the engine surfaces a new opportunity, the €56,350/yr identified base grows — so the captured share can dip even though nothing regressed. A lower score here means more value found, not a system that got worse.
score model health-v2-2026-08 — the components, weights and exclusions above; a changed assumption bumps this version and can never move the grade silently.
ledger version 22c35e346df765c3 — identical on every page, export, and decision record of this workspace state · how to verify it