alex_korr wrote:
Who told you that it is based on 6 data points? Read up on his methodology, it is all out on his blog. He does a lot of sophisticated data wrangling, based on 6 years (and a large number of marathons) worth of history.Thanks. I just read up on his methodology. It sounds pretty thorough, but he also hasn't been able to exactly reproduce the cutoff time. His estimate for Boston 2016 was off by more than the range that he's predicting for 2017. Don't get me wrong, he's definitely close, but he has some pretty strong assumptions in his model. Most of them include consistency across years for interest in Boston. I imagine there's quite a bit of variance there. He's also not taking into account factors such as a growing interest in running across the US and factors such as a stronger economy which make it economically viable for more athletes to travel to Boston (a real factor - I skipped my first year because I didn't have the money to travel despite hitting the BQ time.) I know these factors are small and hard to model, but they do impact the results - possibly by more than five seconds. For transparency, this matters to me because my girlfriend is 54 seconds under her BQ time which is right on the limit of this guy's model. Fingers crossed that she makes it!!