Larger interval better for instantaneous velocity possible?

any function where [2,3] is better estimate for the instantaneous velocity at 2 min than the average velocity interval [2,2.5]?