Someone tried it - http://jcs.org/notaweblog/2009/03/06/trying_to_game_swoopo_c... - and couldn't get that to work.

The author does some interesting pattern analysis of how auctions actually work, and concludes (roughly) that you can't beat BidButler.

that's my site.

i'm pretty sure the last idea i proposed on part two would work fairly well. have the script watch bid activity and when it looks like there's only one other user using bidbutler, turn it on and let it spend money. as soon as a third user starts using bidbutler, turn it off to save money and let the other two fight it out. since both users probably won't use the same price ceiling, it should be pretty clear when one user drops out and the other keeps bidding with bidbutler. at that point, turn it back on and repeat until you're the last one using bidbutler.

I'm not a programmer either, but as someone with an econ undergrad degree going into grad school both the data/analysis and the method of obtaining it are really of interest to me. If anyone can point me in the right direction in terms of what I would need to create such a script I'd be willing to do the work and share the results here.

Well, you need a Firefox with installed greasemonkey, which is an extension that allows you to run your scripts on a page that you're looking at in Firefox. The scripts have to be written in Javascript. The script will have to locate the <div> element that contains the bidding history, then once a second copy the part of its content that was updated since last time to another <div> - where ALL bidding history on this page will accumulate. Then, when the auction ends, copy the contents of this div into excel or whatever you use for analysis and parse it (i. e. put time, price and username into different cells).

In the link above provided by JimmyL, the author says that he wrote such a script (and gives a description of his experiment). Try to email him, I think he'll share it.

Thanks. I saw the link after posting and saw that the author had occasionally had some trouble in the last second (due to swoopo's handling of all the simultaneous bids it seems). I wonder how much this will skew any results. Also, there are 179 live auctions right now, but I figure tracking about 25% should be more than enough for statistical purposes.

Good idea, but I would imagine human behavior is too irrational to model correctly.