Contact Us offers an unbiased, statistically-based probability analysis of the latest poll in each state to project the winner of the Electoral College for the Presidential election. Since 2010, we've applied the same analysis to project which party will control the Senate.
Each projection accurately predicted the election outcome. With each new state poll reported by my either RealClearPolitics or, LPR updates each candidate's probability of victory and the overall probability for the Democrats and Republicans to control the Senate.

LPR is tracking the latest projections of well-known analysts who also utilize state-level probability modeling and update their projections daily: FiveThirtyEight's Nate Silver, Princeton Election Consortium's Sam Wang, HuffPost Pollster,, The NY Times' TheUpshot, The Washington Post's Monkey Cage. We're including the projection for PredictWise, which draws its probabilities from bettors on Betfair.

In determining state probabilities, each analyst uses a different methodology in regard to poll selection, poll weighting, historical trends, etc. After the state probabilities for each candidate are determined, control of the Senate is then determined by either running thousands (or hundreds of thousands) of election "simulations" or by directly calculating the overall probability of winning. From the latest poll in each state, LPR calculates the probability of winning the state for each candidate. States are considered "safe" for a candidate if the probability exceeds 97.5%. In the remaining "toss-up" states, LPR evaluates each possible outcome to determine the overall probability of winning, thereby eliminating the need for election "simulations."

Since 2000, when we began our election analysis, using the final poll in each state, we have identified 195 state elections as being "safe" for a candidate. All 195 have been correctly called. The remaining "toss-up" states determine the election outcome. Applying current state probabilities from each handicapper to our "toss-up" states, we can determine an overall probablity of victory for each party by handicapper using LPR's model: