Several Upgrade: Travel Better readers took Farecast.com for a spin last week. The Boston Globe did, too, and they kept track of how accurate the predictions ended up being. Their answer: so-so, but still with lots of potential.

Twice the website accurately predicted price increases, but twice it missed the boat, once predicting an increase when the fare dropped and once predicting a decrease when the fare went up. On the fifth route, from Boston to Philadelphia, Farecast was fairly close. The website predicted the lowest fare would drop $6 over the next seven days, but it stayed the same.

2 wins, 2 losses, 1 tie.

That’s better than my results. I checked fares for two routes last week: Boston-Chicago (Farecast predicted decline, but the price went up) and Seattle-Chicago (Farecast predicted increase, but the price stayed flat). 0 for 2 for me. Not a huge sample size, admittedly.

Farecast’s excuse? Their software estimates the AVERAGE price over the course of the WHOLE WEEK ahead, not the price exactly one week ahead. Hmm, I see… Well, then they should re-label their “tip” from “wait” to “wait, and check airfares obsessively for the next 7 days.”

P.S. More referrals to the private beta of farecast are available.

Earlier in the week, I commented on Farecast, the new service that seeks to predict the direction of airfares. Access to the site is currently by invitation only.

I have 25 login invitations to give away. If you want a login to beta-test the site, please use the “contact me” link at the top right of the blog to send me a note. First come, first served.

Remember, the site is in beta, so it only searches fares departing from Boston and Seattle, but you’re welcome to kick the tires.

The Boston Globe spends quite a few words on a new fare search site that hasn’t even gone live. It hasn’t even gone truly beta. The new site, Farecast, is in “private beta” — meaning you need an invitation to be able try out the site.

Farecast goes beyond a cute name. It promises to go a step further than FareCompare’s graphical plotting of the past year’s fares, as reported here last week. Farecast’s goal is to predict the direction fares will go, with a confidence percentage (how confident it is of its prediction) so you can decide whether to buy now or wait.

A neat idea, but it’s clearly not quite ready for primetime yet.

The downsides? 1) So far, the site only includes searches from Boston or Seattle. 2) Southwest fares are excluded, which distorts predictions, since a good Southwest fare sale can send the other airlines into a tizzy. 3) If you see a fare you like, you can’t always book it through the site. You may still need to go to Kayak, Travelocity, Orbitz, or the airline to buy the fare. 4) With current oil prices, and rising fares, it doesn’t take much sophisticated programming to create a site that always suggests buying now.

Farecast isn’t alone in the space either. Coming soon: FlySpy.com, offering a similar product. I can imagine FareCompare or Kayak adding a prediction function, too. Who will win out in the end? Let me check my crystal ball…