If you not only want to buy an airline ticket but try to figure out when that ticket might be cheaper or more expensive, check out the new beta of Farecast, available at Farecast.com. The beta is so limited that this is mostly a technology showcase, but it’s an interesting technology showcase.

The limitations: outbound markets are limited to Seattle and Boston. Only roundtrip airfares are available. Sometimes lowest airfares are missed and there are some browser incompatibility issues. You can get a complete listing of the beta limitations here; note these limitations are all being worked on.

The front page has a query box for a trip leaving Boston/Seattle and going several places around the US, you also specify date and the number of people on the trip (a flexible search allows you to compare fares over a 30-day period.) I created a flight from Boston to Buffalo, leaving 7/14 and coming back 7/20.

Farecast thought about it for a minute and came back with a chart predicting that the lowest fare would rise an average of $36 over the next 7 days. Farecast was 78% confident of this prediction. There was also a chart available showing the average ticket prices for this fare over the last couple of months. (Run your mouse over the points of the chart to get prices and dates for those prices.)

Beneath that is a pretty standard fare listing showing low prices, airline, times of departure, etc. A filter system on the left allows you to narrow down the results several different ways — time of departure, airlines, number of stops, price range, etc. (There’s a grid listing tab at the top of the search results page if you’d rather look at the results that way.) Click on the name of the airline/fare provider to go straight to their site.

It’s interesting; I know airfares go up and down, but every single scenario I looked at the fares were going up. Sometimes they were only going up a little bit, but they were going up. You can get some overview of how Farecast is doing its predictions at http://www.farecast.com/about/ourTechnology.jsp.

Due to the extreme limitations of the beta, this isn’t really useful for ticket shopping, but the prediction technology is interesting. It would be useful to either have a third-party watching the success of their predictions over time, or perhaps a place on their site that keeps tabs on their own “track record.”