Every number is traceable. Instantly.
CapVex flags every hardcoded value, broken link, and circular reference. Not with a highlight. With a rule ID, a cell reference, and a plain-English fix.
FINANCIAL MODEL BUILD & REVIEW · IB & PE
Models that stand up in Investment Committee.
LBO model, DCF model, and M&A models — built and reviewed with institutional-grade auditability, accuracy, and traceability.
No case studies published by design.
findings_triage
3 of 40CRITICAL
CF!G44IRR calculation window hardcoded
OpenHIGH
IS!E40Tax allowances not applied to EBT
OpenHIGH
Bridge!F22D&A add-back double-counts Q3 adj
Opencapvex_meridian_v4 · reviewed 12 Jan 2025 · 40 findings
200+rule checks per model
< 0.3%false positive rate across reviewed deals
Analyst-controlledzero output without analyst sign-off
Full audit trailon every workflow
THE PROCESS
Built with intent. Defended with confidence.
In a deal process, the narrative matters as much as the numbers. CapVex ensures every assumption is explicitly set and approved, so the model tells your story with full traceability from input to output.
Step 1
Upload
Ingest Deal Materials
Automated
Step 2
Sign-off 1
Extraction Confirm
Human sign-off required
Step 3
Sign-off 1.5
Circularity Check
Human sign-off required
Step 4
Sign-off 2
Assumptions Sign-off
Human sign-off required
Step 5
Report
+ Export
Automated
Speed With Control
CapVex replaces black-box modeling with controlled, step-by-step construction - where every assumption is reviewed, every dependency is explicit, and every output is fully defensible.
WHY IT MATTERS
Every number in your model should be defensible.
Every assumption is benchmarked before it enters the model.
Every assumption is mapped against sector-specific percentile ranges. Outliers require documented rationale before sign-off. No exceptions.
Every decision is documented — and defensible
CapVex creates an immutable audit trail of every stage sign-off and every rationale entry. Timestamped, attributed, exportable. No data is used or shared without approved consent.
MODEL STRESS TESTING
Built on rules that have been stress-tested, not prompted.
Our rule library is not generated from a language model. Each of the 200+ rules was written, tested, and validated against a corpus of real deal models. False positives reduce analyst trust. We treat a spurious flag as seriously as a missed error.
| Rule ID | Category | False Pos. | Status |
|---|---|---|---|
| IRR-WINDOW-01 | Formula Integrity | 0.2% | Validated |
| LINK-HARDCODE-07 | Linkage | 0.4% | Validated |
| BENCH-EXIT-01 | Benchmark | 0.1% | Validated |
| DA-BRIDGE-03 | Formula Integrity | 0.3% | Validated |
| TAX-ALLOW-02 | Tax Logic | 0.5% | Validated |
Full rule documentation available under NDA.
< 0.3%
average false positive rate across all rules, validated on 200+ reviewed models.
“We stress-test each rule the way you stress-test a model: scenario by scenario, edge case by edge case, until the failure rate is documented and acceptable.”
Built for institutional data.
Your models never leave your control without your consent. CapVex processes files in isolated, ephemeral environments. We do not train on your data. We do not retain model files after workflow completion.
SOC 2 Type II
In audit
Ephemeral processing
Files purged post-workflow
No training on your data
Contractual guarantee
US-hosted by default
EU data residency on request
Bring your model. Leave with conviction.
Pressure-test it in a real-world context. We'll go deep on what actually matters — model defensibility, workflow fit, and firm-wide rollout.
40 minutes. Your model. Reviewed live.