You Pick BestYouPickBest

King of the Hill Voting: How Winner-Stays Ranking Works

By YPB Team · Published September 23, 2026

King of the Hill voting picks one option out of many through a chain of two-way matchups: the winner of each matchup stays, and the loser is replaced by the next challenger. When the challengers run out, the option still standing is the pick. With N options the format takes exactly N − 1 choices, and every option except the last winner leaves after a single loss.

The rule is older than the web. Children play it on a mound, arcades ran their queues on it, and parliaments use the same sequence to vote on amendments. Since 1948 voting theorists have studied one property of it: the order in which challengers arrive can change who wins.

From the playground to the Street Fighter II cabinet

The name comes from the children's game in which one player holds the top of a hill and everyone else tries to push the "king" off and take the spot.

Arcades turned that rule into a queue. On Street Fighter II, which Capcom released in 1991, a second player could drop a coin in at any moment and interrupt with "Here Comes a New Challenger!". The loser of the fight walked away and the winner kept playing. In a 2006 PopMatters essay, Ryan Smith compared the scene to pickup basketball at the YMCA and described players holding their place in line with quarters lined up on the cabinet.1

Twenty years later Mortal Kombat rebuilt the arcade queue online. Its 2011 release shipped with a "King of the Hill" mode designed, Game Informer reported, to recreate the waiting line and the winner-stays policy of a real cabinet.2 That game, usually called Mortal Kombat 9, is one of twelve in the best Mortal Kombat game ranking, and Street Fighter II sits next to the original Mortal Kombat among the 18 greatest arcade games. Pickup basketball keeps the same convention on the court: the winning team stays on until somebody beats it.

The same sequence in parliaments

Legislatures vote on amendments with the winner-stays rule. The Scottish economist Duncan Black described it in 1948 as "ordinary committee procedure": a motion is put, an amendment is set against it, and whichever survives meets the next proposal.3

Robin Farquharson later called it the amendment procedure. A 2015 study of parliamentary voting lists it in use in the parliaments of Finland, Sweden, Switzerland and the United Kingdom, and in the US Congress.4

Why the order of challengers can decide the winner

Take three voters. The first prefers A to B to C, the second B to C to A, the third C to A to B. A majority prefers A over B, B over C and C over A, so the group's preference runs in a circle, a situation known as the Condorcet paradox. Run King of the Hill on it and the result depends only on who enters last.

Order of entry First matchup Winner stays, meets Final pick
A, B, then C A beats B C beats A C
B, C, then A B beats C A beats B A
C, A, then B C beats A B beats C B

Each option wins under one of the orders, and it is always the last to arrive. The COMAP textbook For All Practical Purposes uses a four-option version of the same example, where reshuffling the agenda turns the winner from D into A.5

One case is safe. If an option beats every other option head to head (a Condorcet winner), honest voting elects it whatever the order, because once it enters it cannot lose.4

Without such an option, whoever sets the order holds real power. In a 1977 Virginia Law Review article, the economists Michael Levine and Charles Plott reported designing the voting agenda for their own flying club's choice of a new fleet. The agenda produced the fleet they had aimed for, although another option would have beaten each alternative one on one.6 A year earlier, Richard McKelvey had proved that when options differ on two or more dimensions and no Condorcet winner exists, an agenda setter can almost always build a chain of pairwise votes from any starting point to any outcome.7

King of the Hill with a single voter

The circle needs a group. One person with consistent tastes (liking A over B and B over C means liking A over C) cannot produce it. For that voter each matchup keeps the better option, so the favourite wins from any position in the queue.

The format is also as short as a search for the best can be. Carnegie Mellon's algorithms course proves that finding the maximum of N items takes at least N − 1 comparisons, since every item except the winner has to lose once, and keeping the current best while scanning the rest reaches that minimum.8

What the queue still decides is the shape of the rest of the field. An option that arrives before the favourite can string together wins until the favourite shows up. One that arrives after meets the favourite straight away and leaves with a single loss. For the second-best option in a random queue the chance of arriving first is one in two, close to the 16 in 31 that Lewis Carroll calculated for the second-best player taking second prize in a 32-player knockout, as described in bracket vs pairwise voting.

What psychology adds

Real people are less consistent than the single-voter model, and choices made in sequence carry order effects.

In a 2009 Psychological Science study led by Antonia Mantonakis, participants tasted wines one after another and then named the best. The first wine had a large advantage. Participants who knew wine favoured the last one instead, in the longer sequences.9 The authors explained the pattern with a pairwise-competition model, in which each new wine is compared with the current favourite, the structure of King of the Hill. Simulating that model in 2014, Canic and Pachur found a strong first-option advantage only with short sequences or strong "choice inertia", a tendency to keep the current favourite.10

Speed matters as well. In Dana Carney and Mahzarin Banaji's 2012 experiment, 62% of people choosing quickly between two chewing gums took the one presented first; people who deliberated split 51 to 49.11

Contests judged in sequence show the opposite drift. Wändi Bruine de Bruin found that Eurovision scores rose with a song's place in the running order, and the same linear effect appeared in world and European figure skating championships.12

How YouPickBest runs King of the Hill

Every play on YouPickBest shuffles the options into a fresh random queue. The first two meet, the winner keeps its side of the screen, and the next option from the queue takes the loser's side. The play ends when every option has appeared once, so a ranking with 16 options takes 15 picks, and the last winner is that player's choice. A play can also begin as a bracket and switch to King of the Hill after the first matchup, and that opening result is kept.

The shuffle removes the agenda setter. Nobody chooses the order, and the order changes from one player to the next, so the advantage of arriving early or late lands on a different option each time.

Every matchup counts toward the ranking's results, pooled with the matchups played in brackets. An option's score is built mostly from its win rate across all of them. An option that has not yet won a matchup shows 0 points, which is where a challenger lands after losing its only match to a long-reigning king. Among options with wins, small samples are pulled toward the middle, so one pick does not settle a place, and a long winning streak adds a small bonus. How rankings are scored gives the full calculation.

Fifteen Zelda games, scored from King of the Hill runs and brackets together. Play it →

King of the Hill vs a bracket

King of the Hill Bracket
Picks for N options N − 1 N − 1
A losing option loses to the current king whoever it drew that round
Structure one continuous run rounds, with byes when N is not a power of two
Most wins one option can collect N − 1 the number of rounds (4 for 16 options)
Favourite of a consistent voter always wins always wins

The choice between the two is about pace. A bracket gives each round a finish line and ends in a final. King of the Hill has no rounds: the reigning option stays on screen and every pick is a new challenge to it.

Common questions

How many picks does King of the Hill take?

One fewer than the number of options. Each pick eliminates one option, and the run ends when a single option is left.

Is King of the Hill the same as the amendment procedure?

It follows the same sequence. In the amendment procedure a legislature votes on each amendment against the surviving motion; a King of the Hill voter picks between the current winner and the next challenger.

Can the order of challengers change the winner?

For a group whose preferences run in a circle, yes: the order alone can decide the result. For one voter with consistent preferences, no: that voter's favourite comes out on top wherever it starts in the queue.

Does it matter which option starts on top?

For a single consistent voter the favourite wins regardless. The order does decide how many wins the other options collect before they lose, which is why YouPickBest reshuffles the queue for every play.

More on how voting works

Sources

  1. Ryan Smith, "Street Fighter II: The World Warrior: 'Here Comes a New Challenger!'", PopMatters, 12 October 2006 — the coin-in challenge, winner stays, and the arcade queue: popmatters.com (checked 2026-09-23)
  2. Dan Ryckert, Game Informer, 13 April 2011 — Mortal Kombat's King of the Hill mode recreating the arcade queue: gameinformer.com (checked 2026-09-23)
  3. Nicholas R. Miller, "Agenda trees and sincere voting: a response to Schwartz", Public Choice (2009), doi:10.1007/s11127-009-9562-4 — quotes Duncan Black's 1948 description of ordinary committee procedure: umbc.edu (checked 2026-09-23)
  4. Robert Bredereck, Jiehua Chen, Rolf Niedermeier and Toby Walsh, "Parliamentary Voting Procedures: Agenda Control, Manipulation, and Uncertainty" (2015) — the amendment procedure, where it is used, and the Condorcet winner property: arxiv.org (checked 2026-09-23)
  5. COMAP, For All Practical Purposes, 7th edition, instructor guide to chapter 9 — sequential pairwise voting and agenda dependence: math.wisc.edu (checked 2026-09-23)
  6. Michael E. Levine and Charles R. Plott, "Agenda Influence and Its Implications", Virginia Law Review 63(4), 1977, doi:10.2307/1072445 — the flying club agenda: caltech.edu (checked 2026-09-23)
  7. Richard D. McKelvey, "Intransitivities in multidimensional voting models and some implications for agenda control", Journal of Economic Theory 12(3), 1976, doi:10.1016/0022-0531(76)90040-5: ideas.repec.org (checked 2026-09-23)
  8. Carnegie Mellon University, 15-451 lecture notes "Concrete models and lower bounds", section 2.1 — N − 1 comparisons to find the maximum: cs.cmu.edu (checked 2026-09-23)
  9. Antonia Mantonakis, Pauline Rodero, Isabelle Lesschaeve and Reid Hastie, "Order in Choice", Psychological Science 20(11), 2009: pubmed.ncbi.nlm.nih.gov (checked 2026-09-23)
  10. Canic and Pachur, Frontiers in Psychology 5:902, 2014 — simulations of the pairwise-competition model and choice inertia: pubmed.ncbi.nlm.nih.gov (checked 2026-09-23)
  11. Dana R. Carney and Mahzarin R. Banaji, "First Is Best", PLoS ONE 7(6):e35088, 2012 — 62% vs 38% under quick choice, 51% vs 49% under deliberate choice: doi.org (checked 2026-09-23)
  12. Wändi Bruine de Bruin, "Save the last dance for me: unwanted serial position effects in jury evaluations", Acta Psychologica 118(3), 2005: pubmed.ncbi.nlm.nih.gov (checked 2026-09-23)