Who Calls the Coin Toss in Sports, and What Do the Rules Actually Say?
In NFL games, including the playoffs, the visiting team’s captain calls the coin toss before it lands, and the winner chooses to receive, defer, or pick a direction. In international cricket, tradition — not a written law — has the home captain toss the coin while the visiting captain calls it; the toss winner then only decides whether to bat or field first.
A worked example with numbers
From the 2010 through 2021 postseasons, 10 of 12 NFL overtime games were won by the team that won the coin toss, and 7 of those 12 were decided by a touchdown on the very first overtime possession — a striking imbalance for an event with a 50/50 input. After the 2022 AFC Divisional Round game between the Chiefs and Bills was decided this way, NFL owners approved a rule change: starting with the 2023 postseason, both teams are guaranteed at least one possession in playoff overtime regardless of who wins the toss, removing the pattern that had let a fair 50/50 coin translate into a much larger real advantage. In the regular season, where overtime already guarantees both teams a possession unless the first team scores a touchdown, the toss advantage is smaller but still real: a team that wins the toss and elects to receive gets the first opportunity to end the game outright.
How to set the rule before the result
The general lesson applies beyond football: a coin toss can be exactly fair as an input — each team genuinely has a 50% chance to win it — while the advantage it awards is not fair at all if the resulting procedure heavily favours whoever wins. Evaluating a coin-toss-based rule means checking two separate things: whether the coin itself is unbiased, and whether the rules built on top of the toss result in a roughly even chance of winning the overall contest regardless of who calls it correctly.
Common mistakes that change the odds or the process
A common mistake is assuming a team’s recent run of losing or winning coin tosses says anything about the next toss — the same independent-trials reasoning that applies to any other fair coin. A second mistake is treating “the coin toss decided the game” as evidence the coin itself was unfair; the NFL data above show the coin was fine, while the overtime rule built around it was the actual source of the imbalance the league eventually changed. A third mistake is assuming every sport’s toss carries the same stakes: a coin toss that decides only which end of the field a team defends first, with no other consequence, does not need the same scrutiny as one that can end a contest outright, so the right response to “is this toss rule fair” depends heavily on what winning the toss actually grants.
Where this method stops being appropriate
These are the rules as documented for the cited seasons; individual leagues, tournaments, and even specific competitions revise procedures — the NFL itself changed its own rule in 2022 — so a specific team or tournament’s current rulebook should be checked before relying on any one procedure for an actual scheduled game. Local leagues, school sports, and informal tournaments frequently use simplified or entirely different toss rules than the professional examples described here, so treating an NFL or international cricket procedure as a universal standard for any coin toss would be its own kind of error.
How the random source fits into the rule
ESPN’s explainer on NFL overtime and coin-toss rules documents the visiting-captain-calls rule and the 2022 playoff overtime change; Wikipedia’s article on the cricket toss describes the home-tosses, visitor-calls convention as tradition rather than a codified law. Because one source is a governing sport’s own rules explainer and the other documents an unwritten custom, the two sources deliberately illustrate different levels of formality — a written rule that can only change through an official process, and a widespread habit that individual matches could in principle depart from without breaking any regulation.
Why a fair coin does not guarantee a fair advantage
A 50/50 coin toss is fair as a selection mechanism: each team genuinely has an equal chance to win it. What that win is worth depends entirely on the procedure built afterward. If winning the toss in a particular sport’s overtime format lets a team end the contest immediately on success, the practical advantage from that 50% coin toss can be far larger than 50% of the eventual outcome — exactly the pattern the NFL’s own 10-of-12 statistic showed before the 2022 change. Judging a coin-toss-based procedure requires looking past the toss itself to what its winner is actually allowed to do.
The same separation applies to cricket’s toss-and-choose-innings rule: winning the toss is a fair 50/50 event, but choosing to bat or field first can be a much larger advantage on a pitch expected to deteriorate over several days, which is one reason some formats and conditions are described as “toss-dependent” even though the coin itself is unbiased. The fairness of the coin and the fairness of the resulting game are genuinely separate questions.
What changed in NFL playoff overtime, and why
Before the 2023 postseason, playoff overtime used a modified sudden-death format: the first score of the opening drive could end the game immediately, so winning the toss and driving for a touchdown ended matters without the other team ever getting the ball. The revised rule guarantees both teams a possession, converting overtime from a near-decisive coin-toss advantage into a shorter, fairer contest where the toss still matters but no longer functions as a near-guaranteed win on its own.
The change also illustrates a general option available whenever a coin-toss advantage turns out to be too large: rather than abandoning the coin toss itself, a rule-maker can change what the winner is allowed to do with it. The NFL kept the same 50/50 selection method and instead altered the procedural consequence, which is often a smaller, more targeted fix than replacing the random method entirely.
The same principle is useful outside sports: if a household or group finds that a coin-flip or wheel-based tie-break keeps producing an outcome that feels unfair, the honest diagnostic question is usually not “is the coin biased” but “does whoever wins the tie-break get an outsized benefit.” Adjusting what a random win is worth — splitting a task, alternating who benefits over successive draws, or adding a guaranteed minimum for the loser — often resolves the complaint without touching the random method at all.
Was the NFL’s 10-of-12 pattern likely to happen by chance?
Testing the null hypothesis that winning the coin toss carries no real advantage in NFL overtime — so a toss winner should win the resulting game about half the time — against the observed 10 wins out of 12 postseason overtime games from 2010 through 2021 gives a probability, under that null, of P(10 or more of 12) = [C(12,10)+C(12,11)+C(12,12)] ÷ 2¹², which works out to about 1.93%. That is well below the conventional 5% threshold commonly used to call a pattern statistically unlikely to be chance, consistent with — though on its own not proof of — the idea that the old sudden-death overtime format really did hand the toss winner more than a 50% chance of winning the game, exactly the imbalance the 2022 rule change was built to remove.
How fast the evidence would have mounted with more games
The same calculation shows why patterns like this get taken seriously the longer they persist. If the identical 10-of-12 rate had simply continued for another 12 postseason overtime games, reaching 20 wins out of 24 for toss winners, the probability of that happening under a true no-advantage coin would fall to about 0.077% — from roughly 1 chance in 52 down to roughly 1 chance in 1,297. Doubling the sample while keeping the same rate does not double the strength of the evidence against the null hypothesis; it multiplies it by roughly twenty-five, which is one reason a governing body facing a suspicious pattern usually does not wait for an extreme, all-but-impossible statistic before acting.
The “due for a win” mistake in toss commentary
A broadcaster noting that one team has lost the coin toss five games running, with an implication that the team is “due” to win the next one, is describing a real but irrelevant fact: the probability of losing five independent 50/50 tosses in a row is (1/2)⁵ = 3.125%, uncommon but far from impossible across an entire league’s worth of games each season. The sixth toss remains an independent trial at essentially 50% regardless of the streak, the same reasoning error covered elsewhere on this site for coin-flip and dice streaks — a losing run changes a team’s feelings about the coin, not the coin’s actual probability on its next flip.
How many tosses it would take to actually check a coin for bias
Detecting a genuine 1-percentage-point bias in a specific coin or a specific referee’s tossing method, with 95% confidence, needs the same roughly 9,604 recorded tosses used throughout this site’s coin-flip guides. Even at a generous estimate of 20 or so games a season for one team, each contributing a single toss, accumulating that many toss records from one team alone would take on the order of hundreds of seasons — one reason nobody has ever mounted a real statistical case that a specific NFL coin, as opposed to the procedure built around it, is unfair.
Where the numbers stop and a factual dispute begins
None of this arithmetic can settle a disagreement about what actually happened at one specific toss — whether the call was made before or after the coin left the referee’s hand, or whether an audio replay genuinely shows what a team captain said. Those are questions of evidence and timing, not questions a probability calculation can resolve, and they come up in real disputed tosses far more often than any argument about whether the coin itself was fair. A coin can be a perfectly unbiased 50/50 device and a specific toss can still be honestly disputed, because the two disagreements — is the device fair, and what happened this one time — are answered by completely different kinds of evidence.