One policy for mortgage eligibility and price
Your guideline matrix and rate sheet answer the same question: can this loan be written, and at what price? Decisimo lets the people who own that policy author both in one place and answer them in one run.
score
price
decision
The run settles policy, not guesswork
One decision flow carries the lender's rules from supplied credit data to an eligibility result and a rate outcome.
Derive the qualifying score
Three custom functions handle three, two or one returned bureau scores. A rule set selects the applicable path, and a declared no-score path records the result.
Price the exposure
A par rate decision table uses qualifying score and loan to value bands. A rate adjustments scorecard then sums the lender's defined adjustments.
Apply eligibility rules
A rule set evaluates the lender's knock-outs, including exposure, payment history, seasoning, reserves, no-score exposure and points and fees.
Write the decision reason
The underwriting decision table returns approve, refer or decline. Each row writes a plain-language reason naming the condition that decided it.
Build the rate sheet beside the guideline
The lender's price begins in a decision table, takes its adjustments from a scorecard, then meets the floor in one custom function.
Each of the sixteen priced cells is a row edited in the portal, not a vendor change request or engineering release. An application outside every priced cell gets no nearest-cell rate; the decision refers it to a person.
The scorecard applies the lender's defined points, including negative adjustments such as a shorter term. A custom function takes the greater of par rate plus adjustments and the floor. Before the grid was built, it was tested at every band edge and at each edge plus or minus one step; validation reported zero overlapping row pairs and zero undecidable pairs.
Derive the score the policy prices on
The engine derives the qualifying score from the returned bureau scores, then keeps the selected path in the policy record.
- Three bureaus returned a score. A custom function returns the middle score, including ties. The qualifying score rule set selects this path when three bureaus answered.
- Two bureaus returned a score. A custom function returns the lower of the two scores. The rule set selects it when two bureaus answered.
- One bureau returned a score. The rule set selects the one score that came back. It does not substitute a value at design time.
- No bureau returned a score. The rule set records no score. Eligibility refuses the application above the exposure the policy allows without one, so it is not quietly priced.
A closed-end second uses the same policy design
A lien-position fork sends the application to its own exposure grid and adjustments, with shared credit and debt-to-income logic before it.
Combined loan to value divides the existing first-lien balance plus the requested amount by property value. The second-lien par grid has four bands: up to 70, 70.01 to 80, 80.01 to 85, and 85.01 to 90. Its adjustments cover occupancy, documentation and draw size.
Debt to income is shared and includes the first-lien payment. Both branches require the lien position; an application that does not state it is not priced. The same decision table, scorecard and two custom functions build both grids. Testing proved no second-lien cell prices below a first-lien cell at the same score and exposure.
Exceptions become decisions a person can work
The policy distinguishes an approval under the exception rules from a referral that needs review.
- Compensating factors hold. An application above the standard debt-to-income limit is approved when the qualifying score, exposure, reserves and purpose meet the lender's stated conditions.
- Compensating factors do not hold. The decision row records that the compensating factors did not hold. A person works the referral as a case, with the same rules available inside it. See case work for the file and task model.
- The reason stays with the outcome. The decision table returns approve, refer or decline and names the deciding condition in plain language. That reason is raw material for an adverse-action statement, not the notice itself. The wider credit lifecycle is covered in consumer lending.
The numbers illustrate the shape, not a rate
Every rate, band and threshold shown in the screenshots is illustrative. It is not a market rate, an indication, a quote or anyone's live pricing. The point is that your team can write the grid, its adjustments and its floor as one policy.
What US mortgage regulation asks you to evidence
Regulation B and the Equal Credit Opportunity Act require specific reasons for adverse action, or notice of the right to obtain them. Regulation Z requires a reasonable, good-faith determination of a consumer's ability to repay a dwelling-secured closed-end loan. A run that computes and records the relevant inputs and writes a reason produces evidence for your obligations; it does not discharge them or produce the notice.
Questions we get asked
Is the rate sheet separate from eligibility?
No. Eligibility and price come from the same decision flow. The run shares the report, qualifying score, debt to income and points-and-fees calculation before it follows the applicable lien-position path.
Can a credit analyst change a rate cell?
Yes. Changing a cell in the portal is an edit to the decision table, not a vendor change request or an engineering release. The analyst can run the regression suite, read impact analysis and send the change through approval.
What happens outside the priced bands?
The decision table writes no price. The application is referred to a person rather than receiving the nearest cell's rate.
How is a second lien priced?
A lien-position fork sends it to a second-lien par grid keyed on combined loan to value. Its own scorecard handles the defined adjustments, and shared debt-to-income logic includes the first-lien payment.
Does this decide a mortgage from application to closing?
No. It automates the policy's eligibility and price. Documents, appraisal and verification remain work for a person. A referral can be worked as a case using the lender's rules. Read more about case work.
Where do execution traces live?
Execution traces live on the lender's side, in infrastructure the lender controls. Decisimo retains the decision logic, its revision history and the platform audit log of portal actions. The execution endpoint receives data to make the decision, then execution data is not retained by Decisimo. See real-time decisioning for endpoint and integration information.
Can the policy cover a product with no vendor template?
Yes. A closed-end second is built from the same decision tables, scorecards and custom functions as the first-lien path. The lender writes its own grid and adjustments, then tests the resulting decision flow.
Put your guideline and rate sheet together
See how your team could author, test and release one mortgage decision flow for eligibility and price.