• Marianne Williamson was the first candidate to drop out after January 7th. 14% of contestants accurately predicted that. Most others thought she would drop out 3rd or 4th.
• ⚬
• 45% of participants predict Biden will be the eventual nominee. Warren (25%) and Bernie (23%) are the next most common guesses.
• ⚬
• Michael Bennet is by far the most popular guess for first-to-drop (30% of submissions). John Delaney is next, with 15% predicting he'll be out first.
• ⚬
• 69% of participants predict Bennet will be among the first 3 candidates to drop out. Delaney (48%), Deval Patrick (45%), and Cory Booker (38%) are all also common 'first 3' picks.
• ⚬
• Contest participants disagree the most over Cory Booker's chances - it's common to see him all over the place in guesses. Participants also often disagreed about Tusli Gabbard.
• ⚬
• Tom Steyer left 7 candidates still running when he dropped out on Feb 29th. 49% of contestants predicted Tulsi Gabbard would drop out before any of the other candidates still running. Meanwhile, 34% of contestants predicted Amy Klobuchar would be the first out of the remaining bunch. Compare this to 11% for Bloomberg, 2% for Pete, 1% for Bernie, and no one for Warren or Biden. In the end, the sole 2% who predicted Pete Buttigieg would go before the other final 7 were correct - the former Mayor dropped out March 1st.
• ⚬
• As of March 2nd, the four candidates expected to win delegates on Super Tuesday are Bernie Sanders, Joe Biden, Elizabeth Warren, and Michael Bloomberg. Just over 40% of contestants accurately predicted those four would be among the final five candidates running.
• ⚬
• Also as of March 2nd, Tulsi Gabbard is the fifth candidate still running, defying the expectations of 90% of contestants. Just 2 players accurately predicted the final five candidates.

Between December 3rd and January 7th, contest participants were asked to predict the order in which the remaining 14 Democratic primary candidates would drop out. Because players were guessing drop out order and not actual primary performance, they had to consider factors such as candidates’ finances, desire to preserve their reputations, or even how stubborn they are.

Scoring is calculated using Kendall tau distance, which compares how similar any two ranked lists are. For example, I predicted the candidates would drop out in this order: Sanders 1st, Biden 2nd, Warren 3rd, but let’s say they actually end up dropping out in this order: Biden 1st, Sanders 2nd, Warren 3rd. In that case, my Kendall tau score would be 1, because 1 swap would be needed to make my rank order match reality’s rank order. The number of swaps needed to make a player’s list match reality is their Kendall tau score.

The lowest possible Kendall tau score is 0 (perfectly predicting reality); the highest possible score in this case is 91 (perfectly failing). I converted all scores to a 0-1 scale (where 0 is still the most similar and 1 is the least similar) using the formula: Kendall tau score / (14 candidates * (14 candidates - 1) / 2.

Your accuracy percentage represents how close you are to a perfect score. It is calculated by: 100 - 100 * your (0-1 scaled) Kendall tau score.