Vitalik Buterin: Crypto Prediction Markets Outperformed Forecasts
3 min readEthereum co-founder Vitalik Buterin said today that prediction markets are proving more reliable than traditional polls when it comes to forecasting the current US election.
Traditional polls—that use statistical models of predicting election outcomes—strongly favored a Biden victory in this year’s presidential race. But crypto prediction markets mostly leaned towards a tight 50:50 race. And with the election going down to the wire, Buterin argued that prediction markets had the edge.
“Regardless of who wins from here, I definitely think that the prediction markets have proven themselves more accurate than the polls/models this time around,” said Buterin.
Prediction markets use Ethereum-based tokens to let people bet on the outcome of major events, like elections. The tokens are on sale for between $0 and $1, and when the election is decided, the tokens on the winning side are valued at $1, while those on the losing side are valued at $0. So, if you make the right guess, then you end up in profit—make the wrong guess and you take home nothing.
How each token is priced ahead of the election is used to estimate who will win. If everyone thinks one candidate will win, the resulting demand makes his tokens more expensive—reducing the profitability of buying those tokens. This is kind of similar to how betting odds are worse for favored candidates.
Predicting a tight race
Ahead of the election, both tokens representing Biden and Trump were similarly priced across multiple crypto prediction platforms. This meant, unlike with traditional polls that predicted a likely Biden win, they were more evenly matched.
Buterin said there is a “big difference” between prediction markets and traditional statistical models that try to predict the outcome of elections. This difference, for Buterin, may be down to the notion that bets on prediction markets take into account the possibility of election interference, and other adverse influences.
“Bets on prediction markets correctly incorporate the possibility of heightened election meddling, voter suppression, etc affecting the outcome, but statistical models just assume the voting process is fair,” Buterin said in a tweet.
Yet, there remain other possible explanations. For example, the fact that prediction markets are inaccessible to political experts, or perhaps simply that experts are “incorrigbly dumb and just haven’t learned their lessons around detecting surprise pro-Trump voters as happened in 2016,” Buterin added.
A balanced approach
A final advantage for prediction markets is the fact their methodology do not require a single person’s model to be used.
“The advantage of a prediction market is that rather than being based on one person’s model, it lets the world come together and essentially negotiate a clearing price — the point at which people who think it’s too high and people who think it’s too low are balanced,” Sam Bankman-Fried, CEO of FTX, told Decrypt.
It wasn’t only the crypto community that noticed the difference between the expected outcome and the current tight race.
“Again, I’d urge people to view tonight as another example of some very major systematic polling errors across all kinds of domains rather than ‘under-performance’,” said MSNBC host Chris Hayes, adding that, “literally everyone was working off the same data. The data was bad.”
Although, considering the same thing happened last election, that excuse is starting to look a bit flimsy.