Your question is A/B Test for Bidding Algorithm. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
How would you design an A/B test to evaluate a new bidding algorithm at OpenX?
Address the hypothesis, primary and guardrail metrics, unit of randomization, traffic and duration requirements, sample-size calculation with an explicit MDE, and the analysis and ship criteria. Explain how you would handle peeking, sample ratio mismatch, novelty, and interference caused by repeated auctions or bidder adaptation.