Originally Posted by
mahasamatman
Extrapolation of historical data, including expected traffic delays, anticipated routing, winds aloft, planned fuel loads, and probably a lot more.
What I will perhaps never understand is that it's all fake precision. You can estimate one flight to be two minutes longer an average because of more taxiing delays at SFO, but departure (push) time variance is going to dwarf this computation so I really wonder.. what value can it possibly provide?