Potential based no-regret dynamics are shown to be related to fictitious play. Roughly, these are ε-best reply dynamics where ε is the maximal regret, which vanishes with time. This allows for alternative and sometimes much shorter proofs of known results on convergence of no-regret dynamics...