The reviewer's claim is not supported by the paper, and that is precisely the problem. When a rebuttal acknowledgement misstates your results, the instinct may be to soften your response or assume the reviewer simply misunderstood. But the evidence is on your side, and the most respectful thing you can do for both the reviewer and the field is to present that evidence clearly, calmly, and without apology. This is not about winning an argument. It is about protecting the integrity of your work.
You have already done the hard part. The paper includes a comprehensive list of hyperparameter comparisons, and the reviewer's assertion that your method underperforms baselines under certain settings does not appear in that analysis. That discrepancy is not a minor detail. It is the foundation of a productive rebuttal. Point to the exact tables or figures where the comparisons live. Walk through the specific settings the reviewer mentioned, and show what the data actually says. If the claim is simply absent, say so directly. Reviewers are people too, and most will respond to a straightforward correction delivered without condescension.
What you should not do is ignore the comment or dismiss it as baseless. Even an incorrect claim can signal that your paper's presentation is not as clear as it could be. Use this as an opportunity to strengthen the clarity of your writing. If a reader can walk away thinking your method fails under certain hyperparameters, then somewhere in the text or figures, the story is not being told as explicitly as it could be. Revisit the relevant sections. Add a sentence that directly addresses the hyperparameter space you explored and the range of performance you observed. Make it impossible for the same misreading to happen twice.
In practical terms, your response should do three things: restate the claim, present the evidence that contradicts it, and reaffirm the scope of your contribution. Do not overexplain. Do not speculate about the reviewer's intent. Stick to the data. If the reviewer persists after seeing the evidence, that is a different conversation, and one that the area chair can help mediate. But your job in this moment is to be precise, patient, and unflinching. The work stands on its own. Let the numbers speak, and let the reviewer see that you are not asking for trust. You are asking for a second look at what is already there.