Tired of waiting for election day?

Inevitably, every four years all sorts of different "experts" will spring up, each claiming to know the single best way to predict who will win the US Presidential Election. This usually involves pointing to some obscure poll, survey, or historical precedent that has apparently always been correct ever since election X way back when.

Here are some of the most commonly-cited predictors that have an unbroken track record of legitimacy. There are many others as well, but most of those have been wrong at least once. These ones have been 100% accurate up till now:

Predictors which were voided in 2004

When Senator John Kerry lost the 2004 Presidential election several longstanding election-predicting traditions proved they were no longer reliable:

How wrong were the poll sites?

There were a ton of sites in 2004 who all claimed to have established an authoritative method of predicting the outcome of the electoral college. They used elaborate mathematical formulas and charts that averaged polls and calculated voter trends. They even claimed to be able to predict how the undecideds would break, and how the swing states would swing.

So how reliable did the inter-web prove itself? Here's what the leading electoral college sites predicted on November first. Generally speaking, these sites predicted whatever their ideological bent wanted to happen. That is, the conservatives predicted Bush would win and the liberals predicted Kerry would.

The actual final tally was Bush 286, Kerry 251. The conservative predictor site "electionprojection" was exactly right, the liberal sites tended to be wrong by margins of at least 30 votes or more.