A premium is built from components: a base rate per unit of exposure (sum insured, turnover, number of persons), factors for the risk's characteristics (age, territory, claims history, deductible), minimum premiums, loadings and discounts, and taxes. In a well-designed engine every one of those is a row in a table, so the underwriter can change a rate without a developer and every premium can be traced back to the rows that produced it.
Determinism matters: the same inputs must always give the same premium, and a regulator, an auditor or a capacity provider must be able to see why. That rules out a machine-learning model as the pricing engine of record, though models are useful for proposing rates for a person to adopt.
On Soldek
On Soldek rate tables are data the MGA's team edits (on screen or in the product workbook), the engine is deterministic, every premium shows the lines it was built from, and the same engine answers the broker portal, the quote page and the Client API.
Insurance product builderQuestions
- Can underwriters change rates without IT?
- On a data-driven rating engine, yes: rates are rows in a table with a version and an effective date, so the underwriting team adjusts them and the change applies to new quotes immediately, with the old version kept for existing policies and the audit trail.