Post

once again reminding people that a market at 15% odds (as was the case coming out of that round 1) will go the other way about 1 out of 6.6 times. you don't have to like them but pretending like half a billion dollar markets are completely fake is actually really dumb

Philip Bump@pbump.com

“prediction market”

6 reposts92 likes

I’d say it’s more like acknowledging they are probably manipulated to sway public opinion. This race was always closer than the “prediction markets” claimed. Elections aren’t like a baseball game where the underdog might get lucky and 9th hitter smacks a grand slam to defy the odds.

Interesting how many geniuses on this site know exactly how prediction markets are consistently wrong and yet have no interest in making millions of dollars by exploiting said inefficiencies and structural flaws

i did this a few years ago with predictit and it was roughly spot on, with the tails (5% or lower) being chronically overpriced

I think the problem is more that they've made partnerships with news outlets like cnbc who report the results in lieu of actual reporting. They're pretty useless in that regard unless you want to know what the conventional wisdom is of the demographics that bet there.

What's more informative: A Brazilian reporter reporting how Lula made persuasion gains in the second rounds Or A 30 year old American male placing a bet on an election in Brazil?

but it's not analysis??? large liquid markets provide decent predictive value, discounted by time and uncertainty

What is "decent predictive value" and how would one go about either proving or falsifying such a claim with an election? The stock market has only limited predictive value and that isn't even getting into the thorny question of whether money splits have a direct relation to real-life odds.

i think the core thing for me is that it's not analysis when you can inside trade about it. "will x happen" sure does benefit insider knowledge about x, huh? i'm of the opinion it's either just gambling or insider trading and both need to burn.

Before a coin flip people are trying to analyze the coin, the wind, the flipping technique, whatever. But once the coin is in the air, that information becomes concrete and probability functions start to collapse into real data.