Within the scope of predicting political elections, polls such as those from Gallup can only go so far.  Take the Michigan democratic primary for example, in which Hillary Clinton was favored to win by 21 points before the election. On the evening of March 8, most political pundits and poll-readers were comfortably settling in to watch the wave of votes cascade towards Clinton’s campaign. But instead, Bernie Sanders won; Clinton’s momentum began to dry up simply because of Michigan’s unpredicted outcome. This is just the sort of thing that can change political tides. And it did, in both barrels.

But again, there is only so much polls can do, and sometimes they can be at odds with political news and reports.

There is one organization that is the flip side of the coin when it comes to polls—PredictIt, which can turn the incessant media analysis, polls, incredulous political speeches and the American election process into what it really is: a game. PredictIt is the largest political trading system in America. And for those who use it, PredictIt tosses the entire political syndicate into a compartmentalized machine that only reads back two words: yes or no.

PredictIt also predicted the Michigan primary accurately.

The site, which was created only 16 months ago, turns every election, decision, or happenstance into a simple yes or no situation, where users then raise bets for or against a situation and buy multiple “shares” against the dollar at a price the user and the market set, much like the New York Stock Exchange, but at a miniature scale. It’s a prediction market, which means it was designed to trade around specific outcomes of events which make up the market; in this case a purely political prediction market.

This is PredictIt’s first presidential election season, and in the state of New Hampshire, there were over 770,000 trades of democratic shares prior to the vote. On March 8, Michigan had over two million trades.

According to Brandi Travis, Chief Marketing Officer at PredictIt and parent company Aristotle Inc., these shares, called contracts, have a ceiling to them, which ultimately leaves the company with little profit.

“We’re actually currently not making money,” Travis said with a laugh. “There’s a 850 dollar limit per contract (per each singular “yes or no” bet). And we charge a fee of only 10 percent profit and 5 percent withdrawal but that’s to help pay for expenses, our main objective is to find out more (accurate) information. [PredictIt] is an educational, prediction market project.”

In the last few days before the March 8 primary, Sanders “shares” in Michigan went up, much past Clinton’s; the PredictIt market was making a very clear prediction that the polls and pundits did not see.

And the Michigan polling bundle made ripples in the election. Not only did Michigan light a fire under the Sanders campaign, it undermined the subsequent elections whose polls and pundits predicted Hillary Clinton to win. Up until Michigan, Clinton was winning 64 percent of the voted states, but she went on to only win 53 percent of the states that voted after. And most recently, Sanders has won 5 of the last 6 states.

PredictIt is legal, vetted by the CFTC (Commodities Futures Trading Commision), but there are still risks of insider trading, which, according to Travis, that the company does little to prevent. But to PredictIt, which was created by the non-profit University of Wellington in New Zealand and is based in D.C., the business goal is not to make profits or secure fairness but simply to make a clear and accurate prediction model based around markets, and insider trading only makes that model more accurate.

To UNH finance professor Richard Kilbride, the market always knows best.

“The market itself is a polling and voting machine,” said Kilbride.

He also explained that the market predicts election outcomes through value (of shares) fluctuations, but should not been seen as superior to polls, just different to them. And when asked about the PredictIt concept and prediction markets as a whole, he gave a thumbs up.

“I think it’s a fabulous [political] barometer, and usually politics is more noise than signal. But I can also tell you that the market can be wrong. But it’s generally right. And I want to make a distinction, none of this is insider trading or trading on nonpublic information,” said Kilbride. “Most students of the market know [the market] is much smarter than them; the media, really, is afflicted with the ‘to catch a trader’ mentality.”

For those who use the political trading site, it’s simply an interesting way to profit off the U.S.’s crazy elections.

“Oh it’s a ton of fun, to use pointless political knowledge to make actual monetary gains, it’s awesome,” said sophomore UNH finance student and PredictIt user Connor Courtwright. “I actually made a good amount of money off of Michigan, and New Hampshire, though I personally do not support [Sanders].”

There will undoubtedly be more polls and pundits that will mess up, like the Michigan primary, and provide a nice boost to a candidate’s campaign. But like always, there are two sides to a coin, and PredictIt can help you pocket that coin afterwards.

For most, like Kilbride says, politics is more noise than signal. But for those who see the mosaic of policies, speeches, buzz words and contorted verbatim as clear signals, politics may actually get to be worth its weight in gold.