When asked to localize the onset or offset of a dynamic moving object, systematic biases occur. At stimulus onset, the Fröhlich effect (forward error) is typically observed for fast speeds, whereas the Onset-repulsion effect (backward error) occurs for slow speeds. Conversely, at stimulus offset, Representational Momentum (forward error) dominates for slow to medium speeds, while the Offset-repulsion effect (backward error) emerges at very high speeds. The Speed Prior Account (Merz et al., 2022) offers a unified explanation for these phenomena: It proposes that the perceptual system integrates sensory input with an internal prior expectation of speed. Consequently, perceived speed is "pulled" towards the prior. Stimuli faster than the prior are perceived as slower (resulting in backward shifts/repulsion), while stimuli slower than the prior are perceived as faster (resulting in forward shifts). The three experiments tested whether this internal speed prior is adaptable. We manipulated the experimental context using different speed statistics (blocks with slow vs. fast average speeds). Results show a systematic shift of the localization patterns: The "crossover point" — where biases flip from forward to backward — shifted towards higher speeds in fast contexts and towards lower speeds in slow contexts. These findings suggest that the speed prior is not static but dynamically aligns with the statistical properties of the environment, supporting the predictions of the Speed Prior Account.