Hacker News Reader

Comment by jgrahamc | Hacker News Reader

Makes you wonder with all of Palantir's massive data-mining if Thiel knew something the public didn't.

You didn't need Palantir to see the Trump result coming.

Oh come on. Anyone using a data driven approach to predict the election was putting money on Hillary. If Palantir had data to predict a trump win, then that would be interesting to hear about.

Anyone? I didn't. I put money on Trump long ago.

parent said data driven approach.

Care to outline your thought process? I was trying to find unbiased election coverage, and I read Nate Silver's book a few years back and liked it, so I read 538.

As I mentioned above, I was disappointed that he couldn't resist inserting petty personal insults about Trump into his articles. He was clearly invested in the outcome of the election, which isn't what you want from an analyst.

While he wasn't as bad as the NYTimes, and he hedged his language, he was pretty firm in claiming a likely Clinton victory for the 3 months before the election.

So I was honestly a bit surprised when Trump started winning, and then won decisively. I'm always interested in the areas where reality breaks models, where practice differs from theory, and this is an overt example.

Firstly, looking at politics in the UK and France (two countries that I follow closely) it was clear that there were a large group of people disadvantaged by the 2008 financial crisis who were outside the standard discussion seen in the media of politics.

Secondly, Trump won the nomination with a huge number of votes.

Thirdly, Brexit happened despite opinion polling being the other way.

Then I started paying close attention to crowd sizes at Trump rallies and when Hillary appeared in public. The crowds never got smaller for Trump despite the press and stories around him. I looked at Nate Silver's analysis and guessed the firewall could be breached. Yet the betting markets had incredibly long odds on Trump; so I bet. Modestly since I'm no (b|m)illionaire.

This was hardly a clear deal. Trump won the election with fewer votes than the last two candidates lost with.

Of course it wasn't clear, but the point is that the odds were better than conventional wisdom. That makes it a good bet.

OK thanks, that makes sense. I wasn't aware there were markets, but in retrospect, the piling on by the media should have been a signal of a "mispricing". I tend to buy stocks when I see this behavior. (I believe the media heavily influences the price of stocks, and it's probably even more true for elections.)

I think the process isn't just putting money on the winner. I think the bet was made because a majority of people overvalued the winner. So hedging the bet would win big especially if the margins are close.

I didn't read much of what Nate Silver wrote, but I did listen to the podcast. Clearly, he would have preferred a Hillary win, but I'm not sure it biased the forecasts (he's said it's just an algorithm he wrote and they just plug in the polls). He and his team said many times, just because Hillary was ahead it wasn't a guarantee. Since the election they've said it's very difficult to convey probability to people and they're trying to get better at that. He also pointed out that he was less confident of Hillary than most other sources (even if they had her ahead). They were very clear that they had no expectations of correctly guessing all 51 states because it was so close. I'm not sure if this was because if him calling Trump wrong in the primaries or if he was objectively looking at the numbers.

So hedging the bet would win big especially if the margins are close.

Yes. At the time I placed my bet on a Trump win I simultaneously bet on Hillary with an amount that would cover my Trump bet. Thus Hillary win would have meant that I had neither lost nor gained money.

Interesting to note here that Alex Karp offered to help Hillary.

Reply on news.ycombinator.com