This is an expanded version of Chart 3 that appeared in my FiveThirtyEight article:
These were the five routes in which following Kayak's recommendations would get me to a lower price than the 14-day-out price for each flight. The purple dot was when Kayak first suggested buying. The gray dotted line was the price on my first day of search.
A few key observations here:
- The high variability of prices along many routes
- The variability of the price variations: Los Angeles to Honolulu was rather flat throughout; Los Angeles to Chicago experienced two stages of price hikes; New York to Boston showed a parabola.
- The algorithm does not always find the lowest possible price but did pretty well. As I explained before, using ex-post data to complain about decisions that must be made without such data is bad form.