Can Bitcoin Help Predict the Future?

shutterstock_167979914
24 May 2014

cookie fortune

Fancy taking a bet on bitcoin? Prediction markets are offering people the chance to wager on the virtual currency’s fortunes, among other things – and you can even use your bitcoins to do it.

Prediction markets rely on the wisdom of the crowd, the theory being that people can predict things better when in groups. Although this doesn’t work in some cases (just ask all those who bought real estate during the bubble), proponents say that it has merit, nonetheless.

How prediction markets work

In a typical prediction market, users make a bet, in some form or another. On Fairlay, for example, a user may simply state a prediction, and give odds, and wait for another user to match that bet.

Betting mechanisms can also get quite sophisticated. On sites like Predictious, an event may be created, and then contracts or shares issued in that event. Contracts are paid out only if the event occurs. If you buy a $1 contract in a prediction that bitcoin will reach $2,000 by 1st September, and it happens, then you get your $1 back. If it doesn’t, you lose your money.

On those market-driven sites, you can then trade the contracts you buy, and the market sets the value. The majority of others may not believe that prediction, making it harder to sell those contracts.

You may only be able to sell it for 45 cents, meaning that the market thinks there’s a 45% probability that it will happen. Just as in regular financial markets, the smart money finds predictions that are undervalued, and buys them up in order to cash in later.

In effect, prediction market bets are futures contracts.

These markets can work, say advocates. The University of Iowa runs a prediction market that forecasts presidential election results and claims to outperform traditional polls.

George Mason University ran a predictions market on geopolitical events for two years, and the predictions correlated with real-world outcomes far more than mere chance would allow.

The market uses a Brier score, a statistical method with an output ranging from 0 to 2.

Zero is a perfect, 100% correct prediction, explains Charles Twardy, Research Assistant Professor at GMU. Two is an entirely incorrect prediction. 0.5 equates to no better or worse than a chance outcome. The GMU market averaged roughly 0.2 on geopolitical questions, which is an excellent score.

Cryptocurrency-friendly predictions

Prediction markets are becoming more cryptocurrency-focused in a couple of ways: some of them are beginning to offer specific bitcoin-related predictions, while some are taking bitcoins as payment for bets.

Now, there’s a third, significant development: some people are proposing decentralized prediction markets, in which participants use block chain-based infrastructures to manage the betting.

[post-quote]

Last November, GMU launched a prediction market, called SciCast, which focuses purely on science and technology predictions, for example. As of last week it had only resolved four bitcoin questions, with a Brier of 0.49 – slightly better than chance.

On Fairlay, eight in 10 of the prediction events are bitcoin-related, with the remainder focusing on politics and sports.

The fact that prediction markets take their bets in bitcoins in some ways makes the audience self-selecting, pointed out Martin Köppelmann, Fairlay’s co-founder, saying:

“We narrowed the events in the beginning to bitcoin-related topics, since, as a bitcoin-only prediction market, this is a natural interest for all of our users.”

Bitcoin is a natural choice for prediction markets, because it’s an easy way of making small payments. It also helps not to have fiat currency floating around in a highly contested industry.

Predictions markets have suffered from regulatory scrutiny before: Irish firm Intrade famously shut down after being sued by the US Commodity Futures Trading Commission (CFTC) for selling futures contracts. There were also some financial irregularities, alleged reports.

The firm has since said that motivation is necessary for prediction markets to work, and that financial incentives are the best.

“We are thinking hard about replacement motivations, but there is just nothing quite like ‘cold hard cash’,” Intrade explained.

The need for incentives

Carrot and stickIncentives are an important part of prediction markets, said GMU’s Twardy.

“Each forecaster starts with a finite number of points, and therefore they have to choose a little more carefully where to spend them,” he continued.

“I might have opinions on a lot of things, maybe only a tenth of which I’d be willing to invest my limited points in. It’s a psychological trick to guarantee how I consider how [strongly] I believe in [the prediction]. Economists take this very seriously.”

This also helps to identify those people who really know their stuff, because they accrue more of these points over time, indicating their authority on a subject.

What satisfies Intrade’s call for financial incentives, and the more general need for points that can be easily exchanged? Bitcoin is turning out to be a grand incentive for prediction markets, according to some of the companies operating them.

“It was a good tool to build a prediction market,” says Flavien Charlon, creator of Predictious. “You can bypass banks and set up payments easily.”

The challenge of decentralization

Researchers are interested in other characteristics of cryptocurrencies too – particularly their autonomous, decentralized nature.

The creation of decentralized bitcoin markets will solve some of the biggest challenges facing prediction markets, argues Jeremy Clark, Assistant Professor at the Concordia Institute for Information Systems Engineering. Along with a team from Princeton and the University of Maryland, Clark recently published a long-anticipated paper on decentralizing prediction markets.

“The main goal with decentralization is to remove trust assumptions on single parties,” says Clark. In a centralized prediction market, you would need to let a single party hold money in your account, and to decide on the outcome of an event. You are also trusting it to execute trades in the best way possible.

“Finally the prediction market also gets to choose what markets to open, and what future events you can and cannot bet on.” If you want to bet on something obscure, you can open any event you like in a prediction market.

More than one academic is now exploring the possibilities for decentralized prediction markets using cryptocurrencies. Clark’s white paper suggests creating a new altcoin to facilitate this.

“Bitcoin could just extend their functionality to add this, but traditionally bitcoin has been reluctant to do that. There have been proposals for different extensions that could be used for bitcoin, but they’re very conservative,” he points out.

Another way to do it on top of the bitcoin block chain could be using coloured coins.

Truthcoin and the block chain

Another proposal, Truthcoin, comes from Paul Sztorc, a statistician and Associate in Research at Yale.

In this model, people publish ‘decisions’ on the blockchain, along with ‘states’ (outcomes). People then trade shares in those states on the blockchain. When the event occurs, people get to vote on the eventual outcome of that event, which makes the outcome consensual.

As we move towards decentralized prediction markets, we might escape some of the dangers of decentralized markets, such as the (probably temporary) disappearance of Intrade, and the concentration of decision-maker power in a single entity.

As to when a decentralized market for trading future predictions will be commercially deployed? That’s a bet we’re not taking money on today.

Fortune cookies and carrot and stick images via Shutterstock