Discover Top Posts Tagged with #check my work

Epistemic status: Check my reasoning please.

So in a prediction market on how long a politician will live, every bet on 'at least X time' is also a bounty on assassinating that politician before X is up. For reasons that aren't very clear to me this is considered Unethical and there are no legal markets in politician lifetimes. That's too bad but I can see I'm not going to make any headway on convincing people otherwise so I will reluctantly make do without the ability to anonymously hire assassins.

However! How do people feel about prediction markets on the longevity of laws? For example, "The Jones Act will be repealed in 2023". Every 'no' bet in that market is an incentive for politicians and lobbyists to make an opposing bet and cash in a humongous bribe profit when they repeal the thing! The market therefore allows the mass of people who are moderately screwed by the Jones Act to pool their money together, in effect concentrating the opposition to the distributed harms. If the Act is repealed, the 'no' bettors are better off due to the freedom of the market in shipping - and obviously they each bet less than such freedom is worth to them. And if not, at least they get a tiny amount of money back from the bet.

Apologies if this was completely obvious to everyone else and I'm catching up years later. I haven't seen the theory explicitly laid out anywhere, and today I was thinking of how one could mobilize the diffuse interest in repealing rent-seeking laws. I realized that one cannot expect people to put up $X/2 for repeal in advance, where $X is the value to them of the repeal, because of the freeloader issue; but if one could do a conditional pledge, Kickstarter style, that might work. And then I realized that hey, that's a bet! And if you make it a prediction market, you even get paid for making the bet! And might even be able to extract some of the rent being sought, since the law's beneficiaries have an incentive to bet on repeal.

#prediction markets #economics #check my work #can we get this market up on Kalshi?

Putting this out there for others to check my intuition: Prediction markets as currently configured cannot give odds on events more than 10 years out, and the further the odds are from 50%, the shorter this time horizon.

Reasoning: Suppose the price is 50% and I'm correct; I double my money. But if the payout is in ten years then I only need a return of 7-8 percent to double my money anyway. (Ok that's a pretty high return but you could make it 12 years without much affecting the problem.)

What if the proposition is trading at 0.05, though? Twenty times the money! Well, that's great for me but who is taking the other side of the thing, with a measly 5/95=0.052 return if they are correct? Not per year, total! Less market fees! Even in today's interest environment you can find a money market fund that does better. There are probably people who would do it for the Imaginary Internet Points you get along with the money, but I suggest you cannot run a very liquid real-money market just on those. And anyway they're likely more attracted to the contrarian one-in-twenty side.

Buy now, plan to sell in a year when more evidence has come in and it's clearer that you're right? Maybe, but if large amounts of evidence can be expected in a year, why is the payout a decade away? I don't think this can work in general.

Obviously this only applies to markets in the style of PredictIt, where you have to buy Yes or No bonds and thus tie up your money. But simple bets, where no money changes hands, have the obvious problem of counterparty risk - is the other guy going to be good for it in ten years? Can you even track him down? Is he good for it now? PredictIt does solve those problems. And if it was twenty years, well really now. Are you seriously expecting to still be active in this market twenty years from now? On that sort of timescale most people with real money are going to have to think about contingencies like how their heirs could get access!

And yet it does seem that we make long-term predictions with real money sometimes; investing in an early-stage startup has a time horizon of at least a decade for any Serious payout. But in that case the counterparty is "the whole economy" which presumably will be good for it, indeed a successful startup will help make it so. Is there any way to duplicate that with arbitrary predictions?

Well duh, buy ten-year bonds for the respective stakes and pay out the whole accumulated sum, not the original dollar. That gives you the bet winnings plus the "risk-free" return. Sorry, I'm an idiot. Although afaik no prediction market is doing this right now, so I stand by "as currently configured".

#prediction markets #long term predictions #economics kinda #check my work

s is a new thing I'm trying. I say my piece first — all my bias, all my gut, no notes, no Googling, no homework. Then I run the whole thing past Claude to check me. What I got right, what I got wrong, where I was loud and confidently incorrect. The point isn't to be right on the first pass. The point is to show the seams. So the Supreme Court just gutted the Voting Rights Act. The way I understand it, they said you can't draw districts based on race. And on its face — fine. If you're using race to limit a minority's voting power, that's exactly the thing the Voting Rights Act was built to stop. But that's not what this was. This was the opposite. They took a tool built to undo Jim Crow and used it to dismantle the remedy. And then — this is the part that broke my brain — they basically said it's still fine to carve up districts by party.