Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Prediction Markets: Tales from the Election

vitalik.ca

Post reply on HN
Print view View original

61–70 of 191 posts

Re: Prediction Markets: Tales from the Election

#61

Post by notahacker »

jetrink wrote: post #39
ErikAugust wrote: post #3

"The best way to predict the future is to create it." I always thought the the problem with prediction markets is you have the incentive to make the prediction come true. And that gets ugly fast. But hell, I guess that's what all this is at the end of the day.

One recent example is the man who ran onto the field at the Super Bowl after having proxies place $370,000 in bets on that occurring. However, he bragged about what he had done on a radio show and those bets were invalidated.

Another example that springs to mind is a British bookmaker that ran a promotional betting offer on a fat lower division substitute goalkeeper eating a pie during a televised cup match.

The fat goalkeeper ate a pasty whilst sitting on the bench, and was promptly banned for a couple of months even without any evidence of financial gain for "intentionally influencing a betting market"

Re: Prediction Markets: Tales from the Election

#62

Post by TerminalWarrior »

ahelwer wrote: post #6
ErikAugust wrote: post #3

"The best way to predict the future is to create it." I always thought the the problem with prediction markets is you have the incentive to make the prediction come true. And that gets ugly fast. But hell, I guess that's what all this is at the end of the day.

Yep, it clearly leads into assassination markets - for example, opening a large bet that person X will die on a specific date.

Or, cheap life insurance for the suicidal.

Re: Prediction Markets: Tales from the Election

#63

Post by phpnode »

The problem with making this bet is that the value of the underlying crypto can move by far more than the possible 15% return while the result is pending.

2 months is more than enough for the price to double or halve and people want to be able to exit to fiat at a moment's notice when they think things are heading south.

Re: Prediction Markets: Tales from the Election

#64

Post by rspeele »

tantalor wrote: post #31
bigpumpkin wrote: post #7

One thing he didn't talk about was capital gains tax. If I wanted to bet on the election, I would first have to realize my short term capital gains.

Prediction market positions are not capital gains in the same way that lottery tickets are not capital gains. It is gambling.

Yeah, my PredictIt winnings (also from betting against Trump after Trump lost) are on a 1099-MISC as "Other Income", which I believe is what gambling would be. Still, the point stands that gambling sucks since not only are you likely to lose more often than you win after fees, but taxes eat your winnings too! But then again, it's hardly gambling when you are betting against Q.

Hilariously, on PredictIt you could see the market inefficiency in action as the site has a cap of $850 risked per bet, but after the election you could make essentially the same bet a dozen different ways. Which party will win? Which candidate will win? Will the winning candidate also win the popular vote? What will be the electoral college margin of victory? Will a woman be elected VP in 2020?

Of all those different ways to make the Trump/Biden bet, the most straightforward ones like "which party wins the 2020 election" had the highest Trump prices, while the less prominent markets like "Will the SC Dem Primary winner win the general election?" had lower Trump prices.

Now, why would I and other gamblers not sell my bets from the expensive 88c "SC Dem Primary Winner" market, and move the money into the cheaper 84c "Which party wins" market? Why would the prices not equalize since they are, at this point, bets on the same outcome? Because I'm already at the $850 cap in the "Which party wins" market. I can't put more in. Or, if I wasn't betting on that before the election, I might not even be able to because PredictIt also caps each market at 5,000 total traders.

In other words, the more obvious markets filled up to their limits earlier in the process and had their price influenced more by people who thought Trump would win, and had less price movement after that due to the distribution of traders already locked into the market. The obscure markets were low-volume before the election, but filled up after-the-fact with people like me eager to find "free money" bets, so the price on the Biden side ended up a few cents higher.

Re: Prediction Markets: Tales from the Election

#65

Post by opo »

btilly wrote: post #38
m463 wrote: post #29

Earlier quoted context omitted.

That got ugly too until they prevented people from taking out life insurance on people without their knowledge/consent

Concrete example, Walmart took out life insurance on its employees because they were betting that their horrible working conditions made people more likely to die than the insurance companies thought. And then were incentivized to make those conditions worse. See https://www.bankrate.com/insurance/life-insurance/dead-peasa... for an overview of current regulations that resulted from that.

>...Walmart took out life insurance on its employees because they were betting that their horrible working conditions made people more likely to die than the insurance companies thought.

No, Walmart was doing it since they thought the policies would be tax deductible. They ended up losing money and eventually ended up bringing up a case against their insurance companies:

>...Discount retailing giant WalMart cannot sue its insurers just because it gambled and lost $1.3 billion on getting a tax break from thousands of insurance policies it took out on employees, according to a brief filed by the insurers in the Delaware Supreme Court.

>WalMart is contending in an appeal that it was entitled to rely on its expert insurance brokers to warn the company of the inherent dangers of buying COLI policies. WalMart has asked the high court to revive its bad-faith and breach-of-duty claims against its insurers, which the Delaware Chancery Court had dismissed.

>Such policies lost their attraction for corporations long ago but disputes over liability for the tax consequences continue for companies in a position similar to WalMart’s.

https://www.snopes.com/fact-check/dead-peasant-insurance/

Re: Prediction Markets: Tales from the Election

#67

Post by whatshisface »

avarun wrote: post #23
whatshisface wrote: post #9

Earlier quoted context omitted.

That's not quantifiable evidence of a political bias. You win in a prediction market by guessing who will win, not by stating who you want to win.

Right, and as mentioned in the article: The markets were guessing Trump would win after he had already lost. If that's not evidence of a political bias to you I'm not sure why.

When hurricane forecasters predict that a hurricane will strike a city, would that be evidence of their pro-hurricane bias? If they were consistently overeager to predict hurricane strikes, would that be evidence that they supported hurricanes, or merely were overcalibrated to predicting them?

Re: Prediction Markets: Tales from the Election

#68

Post by didibus »

He did need 1 million dollar simply to make a one time 50k return, and it also seemed like it took considerable effort to do it. I don't know how many people with that money also have the know how, are aware of these betting markets, and care much that they'd make an extra one time 50k.

I think it'll be interesting to see how crypto based betting markets evolve and if regulations will strike them or not. I don't really know what to think about people simply full on gambling on whatever world event, is that a good thing?

Re: Prediction Markets: Tales from the Election

#69

Post by fallingfrog »

DC1350 wrote: post #58
fallingfrog wrote: post #45

Earlier quoted context omitted.

The original meaning of “right wing” meant supporters of monarchy or aristocracy. (The supporters of the aristocracy sat on the right in the Estates General assemblies right before the French Revolution, the commoners on the left). What you are describing is a free market libertarianism that denies or pretends that class doesn’t exist at all, which generally serves the interests of economic aristocracy.

Yes - I just finished reading a book about the French Revolution and it was surprising how far the modern definition of right vs left has strayed. Imagine living in a time when the leftists were pro-war nationalists who loved prison and the death penalty! But in the current political climate, libertarian market ideas are generally included in the basket of ideas labelled 'right'. That's all I meant.

Yes, I think also that the word freedom is a bit ambiguous which lends itself to being distorted. The left wing view is that freedom means not being dominated by authority figures, and the way you fight back against authority is to band together to pursue your common interests, which means that to obtain freedom you have to take collective action. So freedom is seen in terms of power relations. The right wing idea of freedom is being alone or being independent; but we on the left would say that being alone just makes you vulnerable, unless you are already in a position of considerable power. But you know, independence from other people is a kind of freedom too, so I can see the idea there.

Re: Prediction Markets: Tales from the Election

#70

Post by oramit »

I actively participated in the prediction market this election cycle on Predict-It. I didn't know anything about the Ethereum applications so this was an insightful read.

My main takeaway from this whole election cycle was that the betting markets were not that great at predicting the outcome. As the article points out, and I agree with, a big reason is just that these markets are underdeveloped. You have to really be interested in politics or gambling to get involved and the transactions costs are high. Because of this you end up with a mix of die-hard partisans and gamblers looking for an edge. No normal people are participating on these sites. In fact the markets on predictit were sometimes comically wrong because of partisan money. As the author mentions there were contracts trading at 15 cents for Trump to win states after he had already lost them. I bought the NO contracts for 85 cents and made good money on that. For quite a while on predictit there was a persistent pro Trump bias to almost every state level contract. It persisted even after the count was over!

The second reason is that the betting markets themselves were just a reflection of polling data. I'm sure some sophisticated people were using other sources to make bets but most of the predictit discussions were just about polling and how to weight certain polls. There didn't seem to be some big insight that analysts like Nate Silver had missed that the prediction market sniffed out.

In the interest of full disclosure I fall into both the partisan and gambling camps. I wanted Biden to win but knowing that I avoided contracts that were political since I knew I would bias myself. I went instead for proxy contracts where pricing was out of whack. I settled on the total vote count: https://www.predictit.org/markets/detail/6882 Each bracket has a 3 million vote range and about a month out from the election I saw that the higher range contracts were underpriced. For some reason the predictit market had concentrated the betting around the 2016 vote total which made no sense to me. I knew this was going to be a big turnout (even more partisan than 2016 and everyone was at home with covid). At the very least it was going to be more than 2016. The prices were so low on the contracts I just bought all of them above the 2016 numbers with a bias to the upside and it worked out quite well.

It was a fun game but I didn't think it gave me much more insight into the final result.

Post reply on HN