Compare baseline and candidate numerical predictions against reference values. Inspect mean squared error, declared costs and fixed-correction bounds, then freeze and verify saved records. Includes deterministic Python tools, strict input checks and a synthetic solver example. Requires Python 3.9 or later and file execution. Costs and references are supplied by the user; no automatic cost measurement, statistical confidence or scientific certification is provided.