As political pundits continue to fill the airwaves with predictions on how this year’s midterm elections will shake out, NYU politics professor Sanford Gordon has claimed to have devised a way to accurately predict the results of this year’s House elections.

According to Gordon, the model was developed by aggregating survey data from the Cook Report and the Rothenberg Report, two well-known organizations that predict the outcomes of individual political races, and compiling those findings into two larger forecasts.

“The idea behind my method is simply to take those rankings and see how they did in the past at predicting the outcome of the election,” he said. “We treat each election like a little coin flip and the coin is weighted by whatever the historical performance of those ratings [was] in the past … You can flip 435 coins, each corresponding to a house race depending on what rank they got, and that would give you a hypothetical election outcome.”

Using a computer program, Gordon was able to act out 10,000 of these simulations. According to his findings, using the data from the Cook Report, Republicans have a 37 percent chance of a House takeover. Using data from the Rothenberg Report, Gordon predicted that Republicans have a 14 percent chance of retaking the House.

Politics professor Patrick Egan said Gordon’s method was creative.

“By basing its predictions on the accuracy of past results, professor Gordon’s forecast is just as viable as the others that are out there,” Egan said. “The fact that it reaches a vastly different conclusion than most other forecasts should give the pundits some pause about the certainty of their predictions.”

But statistics professor Aaron Tenenbein cautioned against using model-based election forecasting.

“Model-based forecasts, which may have worked well in the past, may not work well in the future if the model assumptions are violated,” Tenenbein said. “I would be very interested to see how well his forecasts work after November 2.”