I have two friends running the marathon today and the BAA projections are driving me crazy because they are so lazy.  They employ a simple average the runner has done to date.  For starters they could at least weight it using the more recent 5k splits.  That is low hanging fruit.

With a little more effort they could pull in terrain data and/or typical pace patterns based on where people are in the course.  For instance if the 35-30k stretch is known to be difficult and runners are typically 5% slower through that portion factor it in.

Steve - I sent a note to info<at>baa.org as this is a great suggestion.  I think it would be very useful for spectators to have a better idea of where and when they should situate themselves to have the best chance of seeing their friends, family, or famous runners who are in the race.  What other reasons would you have for making more accurate projections?

Anyways, I’ll respond back here if I hear anything.  If I don’t, I will try to track down their timing vendor as I am guessing they are probably the ones responsible for these projections.