November 07, 2004

Tradesports Worked Better Than Polls

Tradesports, an Irish inline trading site, lets you buy and sell shares to bet ont he outcome of various events. Sports and games are the most popular, but people will bet on anything - including politics. It's Irish, by the way, because the US test site for determining the predictive abilities of gambling was limited to small buys only. The Irish don't regulate it as much, so the gambling is much bigger.

The way it works, is the price of each share, from $.01 to $1 , is how confident the person is that the outcome will occur. It's more complex than that, but in essence it means that whenever the price is over $.50, the betters predict it will happen. If it's under $.50 then it will not come to pass.

In other words, it's gambling odds to predict a winner of elections. And it works.

These facts are poached from the Conspiracy to Keep You Poor and Stupid, as linked above. At the end of September, tradesports betting:
correctly predicted all 50 states except three (NH, WI, NM);
correctly predicted all 34 senate races except four (AK, FL, NC, SD -- in all cases the GOP won);
correctly predicted the GOP would keep Senate control; and
correctly predicted the GOP would keep House control.
.

That's not bad, considering it was over a month away.

As of the Friday before the election, tradesports betting:
correctly predicted Bush would win;
correctly predicted all 50 states except one (WI);
correctly predicted all 34 senate races except one (AK);
correctly predicted the GOP would keep Senate control;
correctly predicted the GOP would keep House control.


Not bad, four days away and only wrong on one state. All it did was overstate the Bush electoral victory and understate the GOP Senate victory.

How does it match up to exit polls? It's light years better.
According to Drudge, E-Day returns predicted Kerry would win Ohio by four points - he lost by two or three. Kerry was to win Florida by three points - he lost by five. Kerry was supposedl up a whopping eighteen points in New Mexico - where he lost by one point. The exit polls worked in some states passably well, such as Iowa where it was only a point off (it said tie, Bush won by one).

It was horrible on some states Kerry did win, but sucked on the margin. It said New Hampshire was sixteen points into Kerry country where he ended up with a one-point margin of victory. It said Kerry had eighteen point leads in Minnesota, which gave him a three point margin. And it said he was ahead 60-40 in Pennsylvania for a 20 point lead, but in reality it was a two-point win.

Exit polls sucked this year. Tradesports was much more accurate on the ultimate winner. Hopefully we'll listen to the pollsters when they say that exit polls are rough. Hopefully we'll pay more attention to tradesports-style betting next election, because it has better-proven results than exit polls.

0 Comments:

Post a Comment

<< Home