I’ve been skeptical of how well farecast‘s recommendation and airline price forecast works. So I did alittle crude experiment. I searched a flight from sea to oak, Sep 21-26(thu-tue), the most remost future farecast woudl predict at the time. Iran the first prediction on july/3 (right before july 4th).Unfortunately, my PDF printer didn’t print the trend graph for the July3 prediction. The chart shows a steep increase of price in “recenthistory” due to the holiday season. The prediction according to recenthistory was a continued increase, which resulted in a recommendation of “BUY.”
July 3, 2006 prediction
Two weeks later, July 17, I ran the same search. The result was as expected, a steady and lower price. The recommendation? “BUY”. Had I really needed to order the ticket then, not to mention it was 3 months in advance, and trusted Farecast 2 weeks ago, I would have wasted $80. I admit this is not a good test yet given that the predicted time is so far ahead. Next time I will need to try something with the period of two weeks, when the prediction really matters.