Always-valid sequential A/B testing engine: mSPRT confidence sequences make peeking safe by construction, CUPED cuts variance up to 49%, SRM gates bad data. Built-in adversarial peeking harness proves the claim: naive daily peeking hit 27.5% false positives in 2,000 simulations; this engine held 1.7%. All numbers reproducible from committed seeds.
python statistics monte-carlo experimentation ab-testing hypothesis-testing causal-inference variance-reduction cuped sequential-testing confidence-sequences always-valid-inference msprt srm-detection
-
Updated
Aug 4, 2026 - Python