I track them as well to use as a control for a model I built to convert to 40 times (I would call it reasonably successful).
Kids are so different. A kid who focuses, learns a better start, gets his technique can improve .2-.5 seconds in a year. Most kids are going to see a wide variance within one year, let alone from year-to-year. Track conditions, wind conditions, when the event takes place all factors in.
Again, I'd say the regression on it shows it to be a nice variable to have your model, but fairly insignificant overall.