What was broken
Institution-scale compression cycles depended on consistent valuation across sources. A discrepancy found late could halt a live cycle and force an expensive resubmission.
Product decision
Create an independent valuation view and compare two sources before the live window. Move failure upstream, while it is still cheap.
In plain terms
- Compression cycles only run if valuations agree. A late mismatch stops the room
- $XT+ is the notional size of cycles under this check, not revenue or assets managed
- An independent view compared to internal marks; differences outside a tolerance band fail early
- Success was earlier detection and fewer expensive resubmissions. This is not a pricing product
What I owned
- Defined the product strategy: data quality as a control before live, not an ops cleanup
- Launched the dual-source valuation and validation engine for institutional cycles
- Tied success to detection lead time and resubmission rate across participating banks
How it works
- Produces an independent valuation view
- Compares results across two sources before live execution
- Surfaces discrepancies early enough for teams to resolve them
- Turns data quality from a late-stage failure into an explicit product control
Role & tags
Product Manager. Defined product strategy and launched the valuation and validation engine.
- Independent valuation
- Dual-source
- Shift-left controls
- Reliability KPI
Impact
Discrepancies surfaced 48 hours earlier. Failed-run resubmissions fell 91% across 24 banks running $XT+ cycles.

Also seeCompression Setup AgentIncident Triage AgentCross-Currency Compression
1-7 beats · j / k · ← → adjacent · y link · b brief