Far too often the favorites would be exponentially lower than rightful. The sites would see polls indicating a 3 or 4 point edge and somehow equate that to a football game, where 3 or 4 points equates to 60 or 65% likelihood. As Silver has demonstrated, a 3-4 point edge in politics in the late going is more like a 95+% edge. Many of us knew that all along. It was a gold mine for more than a decade. I really should have taken greater advantage than I did. I still shiver at some of the laughable numbers that were out there, like the 2006 Connecticut senate race where one major site used Joe Lieberman as only a -160 favorite against Ned Lamont in the general election. That price literally should have been -2000 or greater. Lamont won the Democratic primary narrowly over Lieberman but that had no relationship to the demographics in a general election, where independents and Republicans would favor Lieberman heavily. The GOP nominee was a token joke, guaranteed not to get many votes. My Excel model had Lieberman by 10, and that's roughly where it ended up. I guarantee the perspective of that race, and therefore the betting odds, would have been dramatically different if Nate Silver had been up and running at the time.
I started betting politics in the early '90s and won a 16-man betting pool in 1996, using primitive versions of some of the same concepts Silver is using. I had a PAN (Partisan Adjustment Number) for every state. That gave me a big leg up on my competitors. Quickly I became in tune with every state in terms of the partisan realities, like the percentage of self-identified liberals and conservatives in each state. That's a huge untapped variable, IMO, one that Silver himself does not allow enough weight. The self-identify number is more rigid and meaningful than Party ID. Once my parents' health declined about 5 years ago I stopped devoting much time to the political models, and I rarely post on political sites anymore, but it is fun to follow a little bit leading to election day.
Silver's followers on the original 538.com site got a bit annoyed with me in fall 2008 when I corrected a blatant error Silver made in one of his threads. He couldn't figure out why Obama's odds were so low on Intrade. It was clearly driving him nuts because he expected his percentage to align with the market odds, the so-called wisdom of the crowd. I was commenting regularly on Mark Blumenthal's great low traffic site Pollster.com at the time. There were only a couple dozen comments under those Pollster threads, despite the awesome content Blumenthal was providing. I mentioned that Intrade was notorious for scalping, in other words people betting both sides to guarantee a profit. Take 60 and give 40, that type of thing. Silver and Blumenthal are big buddies. Blumenthal was contacting me regularly via email. Silver obviously saw that "scalp" comment because immediately he wrote a long article on 538 saying the Intrade mystery had been solved, that it was arbitrage wagering. That's the British term for scalping, i.e. manipulation to guarantee a profit. Turning the bookie edge on your side. The problem was, Silver's lead example was very poor, in fact ignorant. He emphasized that it was like betting both sides of a football game with no risk, taking 6 points on one side and giving 3. Something like that. The numbers he used might not have been exact. Silver was spotlighting the wrong type of wager. There's no risk in scalping a money line, like taking +250 and giving -200. But there certainly is risk in giving -3 -110 and taking +6 -110. You're going to lose the vigorish unless it hits the range of 3 and 6. Trust me, I've been juiced out in short term samples while playing middle. You might have 40 or 50 in a row fail to fall. Anyway, I posted too arrogantly on Blumenthal's site and suddenly a bunch of Silver worshippers started attacking me, even though my point was 100% valid. Eventually some of them figured it out. These days Blumenthal's site is under the Huffington umbrella and Silver is at New York Times. It made sense for them financially but there's no intimacy or shared knowledge. Silver no longer responds to commenters, nor does Blumenthal. However, Sam Wang does. His site is excellent, and relatively low number of comments. Wang uses some of the same ideas as Silver, heavily invested in state polls.
I honestly never heard of the guy before he moved into political forcasting, but he definitely caught my attention with his 2008 predictions. So as a betting man, would you say he has influenced the betting lines on elections? I mean, since he's gone public, people could just piggy back on Nate Silvers work and dilute the payouts right?