Our Senate model puts Democratic control at 54% as Florida and Minnesota move

Our model gives Democrats a 54 in 100 chance of winning control of the Senate, against 45% for Republicans, as of 6:32 a.m. ET on September 28, 36 days before the election. The model's average outcome is 50.7 Democratic seats. Democrats need 51 seats for control. Republicans need 50, because the vice president is a Republican and a 50-50 tie goes to them. The average sits just short of the Democratic threshold, which is why the chamber remains close to even in our estimate.

The House picture is less close. Our model gives Democrats a 98% chance of winning a majority and Republicans 1%, with an average outcome of 245.7 Democratic seats against the 218 needed. That figure is somewhat firmer than it should be. Our simulation correlates polling error nationally, by census division, by state and within each district, but it has no term linking demographically similar districts in different parts of the country, because we lack the data to build one. Without that term, the House probability is slightly more confident than a fuller model would be.

Our generic ballot average stands at D+8.3, drawn from 27 polls over a 45-day window.

Two Senate races moved this cycle that we have not yet covered. In Florida, our model's Democratic win probability fell to 18% from 35%. In Minnesota, it rose to 90% from 77%. Iowa has also moved further than we previously reported: our model now gives the Democrat, Josh Turek, a 62% chance, up from 42%.

The new polling of note comes from Michigan. A Big Data Poll survey sponsored by the Public Polling Project, fielded September 22-24, found Abdul El-Sayed, the Democrat, at 46.7% and Mike Rogers, the Republican, at 42.1% among 678 likely voters. The same poll's registered-voter sample of 725 showed El-Sayed at 45.4% and Rogers at 41.0%.