Methodology
Make every assumption visible.
CardMath separates two things: product facts that must be verified, and personal assumptions that belong to the person running the math.
RULE 01
A current number needs a current source
The issuer or official program source is the starting point. CardMath does not publish a current numerical claim from a source-of-truth row unless that row was verified within seven days. Stale or blocked data stays off the page.
RULE 02
A credit is not cash
Count a listed credit only when you would have spent that money anyway and can actually satisfy its terms. A headline credit amount is not automatically personal value.
RULE 03
A cap is not a target
A spending cap is a ceiling on eligible rewards, not a suggestion to spend up to it. Calculators should use expected normal spending, not manufactured behavior.
RULE 04
Different earning paths stay separate
Category earning, issuer-portal earning, and transfer-partner value are different mechanisms. Mixing them into one rate hides assumptions. CardMath keeps each layer explicit.
RULE 05
Point value belongs to the user
For points and miles, cents-per-point is an assumption. The calculator asks for yours rather than assigning a universal value and presenting it as fact.
RULE 06
The output is a result, not a verdict
CardMath can show that one scenario is mathematically higher than another under the same inputs. It does not turn that difference into an instruction to apply or a claim that one card fits a person.
The boundary
Product fee, rate, credit, offer, APR, cap, and eligibility fields come from verified source material. Spending, credit usage, and point valuation come from the user. The calculation sits between them.