The Fintech Builder · review evidence

Keep the rival in the graph

Inspect every internal row against all external rows. See why a strong score can still fail mutual uniqueness or a runner-up margin.

Synthetic Cedar · 24 initial source events · 12 rows per side · all 144 comparisons retained. Scores are points, never probabilities.

All external comparisons for the selected internal rowEach bar is an exact score displayed on a zero-to-100-point scale. A zero-height outlined marker means a hard gate or unsupported text withheld scoring. Labels and the audit table identify the reason.1000Score points · a high score alone does not clear ambiguity
ProposedScored, held for reviewNo score

The full graph stays fixed for every decision. Choosing A2/B7 never promotes A1/B6 in a second pass.

Current row decision

Best exact score points; not match confidence.

Exact current-row fractions

Current row · every external candidate

External rowGap daysScore pointsRunner-up margins I / EDecision reason
Complete comparison audit
Internal / externalGap daysScore pointsRunner-up margins I / EDecision reason