Home About Us Track Record Contact Us | |

LatestPollResults offers an unbiased, statistically-based probability analysis to project the winner of the Presidential election. Check out our perfect record since we began our analysis in 2000. With each state's probability and corresponding electoral votes in the Electoral College, we mathematically compute the probability for all possible electoral vote combinations for the candidates. Unlike most election analysis websites, we don't run thousands of in an attempt to estimate the overall probability of winning. We calculate the simulations overall probability for each party's candidate to win the Presidency. We also adapt our mathematical engine to project which party will control the Senate and the House. Below is a current snapshot of LPR's overall projection using state-level probabilities derived from market-based pricing:
exact | |

In presidential years, we typically begin our election coverage after the Republican and Democratic conventions have occurred and the candidates have been determined. The polling season begins in earnest and we have more data from which to make projections. This year, we have an additional source for determining state-level probabilities. PredictIt.org offers real-money market-based trading of political contracts. As of 6/21, there were 20 states with Presidential contracts related to which party will win that state. There were also 16 states with Senate contracts available. Traders purchase or sell contracts based on whether they expect the Democratic or Republican party to win that state. From PredictIt's market data, LPR derives the probability for each contract, after adjusting for PredictIt's commissions and after normalization. With the advent of PredictIt, we will include projections for both market-based probabilities along with probabilities derived from our poll-based analysis. The track record for both is impressive. We're not prepared to abandon our perfect record with poll-based projections, adhering to the adage that "if it ain't broken don't fix it!" However, there is quality research that suggests that market-based predictions have been more accurate than poll-based predictions. In analyzing 20 states, there are 2^20 possible outcomes - over one | |

Home About Us Track Record Contact Us | |