Hacker News new | past | comments | ask | show | jobs | submit login
Tales from Prediction Markets (misinfounderload.substack.com)
223 points by ikeboy 7 days ago | hide | past | favorite | 56 comments





Betting on whether someone is still alive by a certain date is probably against the rules of this prediction market, but this is a classic way (read: academic) to implement a hitman service with some level of safety between the participants (victim excluded).

All you need to do is bet a large amount of money that the target will still be alive by a certain date. Then someone takes you up on it, makes them have an accident and you just lose a bet – nothing more. You didn't discuss hiring a hitman, you've never even met a hitman: you're just placing bets online.


IDK if academic is the right term, but market manipulation is definitely the nerdiest way of hiring a hitman.

We could take this up a notch by making the bet (person A dies by date X, $100) tradeable. Maybe also create a derivatives market. Now we have a hitman market and a bodyguard market.

Derivatives have several beneficial effects. One is uncertainty, which should stimulate a diversity of contracts: dead by X, $100; dead by Y, $105. People love arbitrage. The second benefit of a two sided market is naked short sellers.

Best case scenario, naked short selling creates an infinite supply of tradeable hitman contracts. In theory, demand is always finite but as Keynes famously said... "no one f#&k-ng knows how demand works!"

So eventually... Robin Hood adds every person on earth to their murder exchange and the market singles out a random person for $gme status. Her name in Ana and next friday, $6 billion of her murder contracts expire. Half the mercenaries on earth are trying to kill her. The other half are her personal army.


Someone must have written a great book with that kind of a plot. Does anyone know such book?

'The Insurance Man' by Zadie Smith & Neil Gaiman

This book doesn't seem to exist on Google. Got a link?

There is one with that title by Timothy Squire, that has a close plot. Is that the one you are talking about?


They're making a joke.

Damn I really wanted to read that book. The plot is extremely interesting.

The trouble is, anyone can bet against the person remaining alive, so it becomes a public goods problem. Everyone's incentive is to let someone else do the assassination, and freeload. There's no extra profit in actually doing the hit, but a lot of extra risk.

>anyone can bet against the person remaining alive

Yeah, but it is 1. risky ( if the person stays alive you just lose money ). and 2. the incentive for a potential hitman is reduced by your bet ( so by placing the bet you are actually reducing your odds of success ).


So to increase your odds of surviving you can bet on your own death. That seems strangely philosophical to me. Another upside is, if you should die anyway, then at least you win the bet!

It's like life insurance that decreases your chance of death.

Knowing with certainty that the event will happen can lead to additional profit by being willing to accept less than fair odds or to increase leverage.

This is half clever, but taking up the other side of that bet in a public market would immediately make you a person of interest to law enforcement.

While not literally an assassination market, the souljahboy story demonstrates the site creates them: a player can use direct action to force an outcome to occur, rather than predict it as intended. For tweets, its silly. But what if someone sabotaged the vaccine distribution, for instance? There does seem to be room here for unintended and quite unfortuante outcomes.

How about someone sells a whole bunch of credit default swaps on some really risky asset backed securities and takes the proceeds and pays off the loans so default is impossible.

https://www.wsj.com/articles/SB124468148614104619


But isn't a bet on a prediction market open to anyone? Unless the assassin took the bet and killed the target moments before the deadline, couldn't a casual observer jump in on the action after they learned of the target's death?

Who'd be willing to take the other side of that bet at that point?

Sympathetic magic

A large enough bet is basically a voodoo doll into which you've just stuck a pin.


I could see a judge sending you to prison for this all the same.

They have markets on whether Biden, Cuomo, and Gaetz will still be in office as of various dates. You could pretty cheaply buy shares that pay off $1M+ if Biden were to die. But I'd expect the cost of assassination to be well over that amount by orders of magnitude plus you'd need to run forever.

So what I’m learning from this is that unsurprisingly this unregulated market has a tonne of market manipulation. Not sure why anyone would get involved if they didn’t plan to cheat. How long before there’s a bomb threat at the CDC in order to win a prediction bet?

Well... You can cheat and lose. You can be cheated and win. Cheating hurts market integrity, but it doesn't kill it. Even consistent negative returns don't always kill a market. They can sometimes even live on as casinos, where average returns are guaranteed to be negative. Markets really hard to kill with market forces. Hence the Lenin-Luxembourg disagreement.

> Not sure why anyone would get involved if they didn’t plan to cheat.

Some markets are full of dumb money. Some people are mistaken about whether they are the dumb money or the smart money. Some people really are smart money and they can make large, fast nearly riskless returns on invested capital.

People make money in large liquid markets with many sophisticated participants all the time. This has need true as long as stock, bond and commodity markets have existed, in markets with and without bans on insider trading.

> How long before there’s a bomb threat at the CDC in order to win a prediction bet?

How long before there’s a bomb threat at Pfizer to make a short part off big? Criminals and terrorists are not generally known for their skills at long term or strategic planning.


> How long before there’s a bomb threat at Pfizer to make a short part off big?

Fair point, but the anonymity of the markets mentioned in the post matters here. You can short a stock or buy put options, but if the SEC gets suspicious they can trace them back to you (and there are occasional stories of this happening, like the guy who set off actual bombs at Target). With anonymous prediction markets, they would have a harder time figuring out who stood to benefit.


Hypermind [1] is an interesting prediction market which avoids many of these pitfalls, by 1) using play money instead of real money, and 2) requiring an application for new users.

I've been playing on it for a few years, and the predictions are usually pretty good.

[1]: https://predict.hypermind.com


There's also https://www.metaculus.com/ (which does not require an application)

> The number gets reported on the main page once a day, but someone found a page with the hourly data and figured out how to average together the data points to predict the daily number.

Isn't this THE point of prediction market? A user found a better source of information and used it to make money.


I would say technically the point of a prediction market is to predict things in advance of them happening, so I think you might get people who sensibly disagree with this point of view. But in the article it's a little confusing, because the person who did this could only trade IIUC with other people on the same day as the day the figure would be published so the people who traded with them were doing something very ill-advised to start with. I can't see how any of this would have affected people who tried to actually predict the co2 level properly, except that now they don't have a market any more.

Why isn't it the point? The point of prediction markets is to distribution information. If you know a better source the prediction markets incentivize you to aggregate that source and distribute that information widely. Once people find out, then that becomes the norm and the asset gets priced correctly. That's exactly the point of prediction markets.

[EDIT] Didn't realize the first section of OP was a quote and not an argument. I agree with OP


That's what the parent meant.

The more interesting question to me is when prediction markets/crowdsourcing works consistently well. Because there seem to be some specific scenarios where it does. e.g. [1]. But mostly it doesn't. And while there are some intutions around this, I've never run across a lot of substantial research.

[1] https://bitmason.blogspot.com/2012/04/crowdsourcing-predicti...


One thing these tales demonstrate is that they probably don't work so well when you can cheaply manipulate the outcome.

This is the first time I heard of this prediction market. Or, indeed, any heavily used real prediction market.

I wonder about two things:

1. Legality. Taking bets from Americans has landed other company owners in prison so I always wonder how they handle this on internet based gambling sites.

2. Oracle. Who decides who wins? It looks like it's decided by the company owning the platform. This seems a little dubious but maybe it works with the transparent nature of many of these sort of bets. Of course, they could always be in dispute.

The only information I could find on the oracle: https://dyor-crypto.fandom.com/wiki/Polymarket#Oracle_Method...


1. Most of the markets I've looked at have a letter of no-action from the CFTC. But I'm not aware of one for polymarket.

2. I imagine that any suspicion of foul play would taint the waters, the same way a 51 percent attack could technically make you a bitcoin billionaire, at the cost of completely destroying bitcoin value. IEM works the same way, but is affiliated with a non-profit / government entity.


> a 51 percent attack could technically make you a bitcoin billionaire

That's not the case.

A 51 percent attack doesn't allow you to steal bitcoin, if that's what you have in mind.


A successful 51% attack lets you double-spend your coins, which is effectively the same thing. (Until the market catches on and BTC plummets, anyway.)

You are not going to be able to double spend a billion dollars. And it would cost hundreds of millions, possibly over a billion, to accumulate enough mining power to perform a 51% attack.

I don't know about the legality, but they've raised millions in VC funding. I suspect they have some kind of plan. They don't hold your money, all they do is act as an oracle as you note, which might help legally.

So far there's not been any major issues with markets resolving incorrectly, across over $100M of total volume and dozens of markets.

Predictit is also heavily used but limits individual bets to $850 per market. They have an official no action letter from the CFTC, so that's all legal.


> they've raised millions in VC funding. I suspect they have some kind of plan

In most cases the plan is to raise millions.


Some predictions seem pretty vague, even if the probability of hitting edge cases is low. E.g. "Will ETH be above $1,500 on January 27th?" begs the question on which exchange? Or smaller ones like at which hour, does it have to be higher all day or just for a single trade, if for a single trade, if Alice sells Bob 0.01 ETH that day for $2k does that mean it was higher etc.

In general oracle role seems pretty hard when big money is involved.


For most of those there are specifics on how it will be resolved in the description. Usually a specific url they base it off.

They can hardly put all of it in the title.


That page says "nor do we host any markets ourselves. Polymarket displays existing markets live on the Ethereum blockchain," so it's going to be a variety of oracles, none of them being Polymarket. They're just an aggregator.

The leading prediction market on Ethereum is Augur, which doesn't have trusted oracles at all. They have a way of crowdsourcing the bet resolutions, which has worked well so far.


Polymarket is in fact the oracle for all of the markets on their site. They don't host them in the sense that it's all on the blockchain.

Augur has no actively traded markets last I checked. It was good by the election but has been basically dead since. Polymarket has way more volume than Augur as of now.

Omen is more decentralized but doesn't have much volume. There's also a version of omen on level 2 on the xDai chain.


Isn’t the company holding your money anyway? Seems pretty reasonable to trust them to resolve the markets as best as they can, at least if they seem to be interested in staying in business for a while.

Interestingly, the setup used means they don't actually hold your money. It's all on the blockchain. They can resolve incorrectly but they can't outright steal everyone's money.

Can you withdraw your money? Presumably not, since that wouldn’t make for a very interesting market. So you’re still trusting the company with control over your money. The fact that there’s no direct incentive for the company to outright steal your money (unless they’re on the other side of every bet, lol) ought to make them even easier to trust!

You can withdraw your money of course. You have the key to your money and under the hood it's smart contracts between users.

> Or, indeed, any heavily used real prediction market.

Predictious, Bitbets and others have had some success before. It didn't work well because none of the cryptos are stable, and also - given the semi illegal nature and the need to rely on anonymous operators - exit scams happen every now and then.

Oracle is simply a URL where the info will be sourced from to resolve. For example, bloomberg.com/in/fo, or trusted.g.uy/bet12


Polymarket uses USDC, a stablecoin, and is run by a NYC based company with millions in VC funding.

Looks more like sports parlay gambling

There's been some interesting markets. The vaccine market produced some interesting models, and there's a few markets now on how many covid cases will be reported in various states and nationwide.

The market on the Ever Given getting unstuck was super fun.

very interesting stories. tells you a lot about how inefficient markets are. reminds me of "reminisces of a stock operator"

timeless book

I would recommend the book: "Broken Markets"

This entire concept strikes me as pure exploitation of gambling, with extra steps to hide the situation from the typical non-critically thinking mark.

Sure, this concept is not obviously evil, but is damn close, and I'll bet your ass in practice operating such an endeavor is dancing with the devil.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: