Win percentage:
wins over games, and the rules that change it

A simple ratio with a few hidden choices

Win percentage looks like the easiest statistic in sport: divide the number of wins by the number of games and multiply by 100. For a league with no ties and no abandoned games, that is all there is to it. A team that won 30 of 40 matches has a win percentage of 75%.

The trouble starts when a match does not end in a clean win or loss. A cricket game can be tied, washed out by rain or abandoned. A football league may have draws. American football allows ties, and baseball does not, because games continue until someone wins. Each sport, and sometimes each competition, handles these in its own way.

The result is that two sources can publish different figures for the same team, and both can be right. The only way to compare fairly is to know which convention is being applied.

Some tournaments publish a table sorted by winning share when teams have played different numbers of matches, because a table by raw wins would punish a side that has had fewer fixtures. Others use net run rate or points as a tie-break. Reading a table well starts with its legend.

Three common conventions

1. 1All matches: games counted = wins + losses + ties + no-results. It treats every scheduled match as a game.
2. 2Decided matches only: games counted = wins + losses + ties. No-results are removed because nobody won or lost.
3. 3Ties as half: each tie adds 0.5 to the wins, so the numerator is wins + ties ÷ 2, and the denominator is the decided matches.

The third method is common in American football and some other leagues. It gives a tie the value halfway between a win and a loss.

Which one to choose depends on what you want the number to say. If the question is how often a team wins when it takes the field, count every scheduled match. If the question is how good it is when a result is reached, use decided matches. If you want a single figure that treats a shared result as exactly midway, use ties as half. State the choice beside the number, and keep to it for every team in a comparison.

Worked example: a season of 38 matches

The spread between the lowest and the highest is 5.0 percentage points for the same 38 matches. A reader comparing this team with another that was quoted under a different convention would draw the wrong conclusion about which was better.

How different sports and competitions handle it

• Baseball: no ties, so winning percentage is wins ÷ (wins + losses). Postponed games are replayed.
• American football: ties count as half a win and half a loss. A 10-6-1 record gives (10 + 0.5) ÷ 17 = 61.8%.
• Cricket: leagues vary. Many report wins ÷ matches played, others exclude no-results, and tied matches are often given half a point. Check the league's own table.
• Test cricket's World Test Championship ranks teams by percentage of points won out of points contested, with a win worth more than a draw, rather than by simple wins ÷ matches. The scoring has been revised, so use the current rulebook.
• Basketball and tennis: no ties, so the ratio is clean, though in tennis analysts also use percentage of points won.

Football leagues rank by points, not by share of wins, because a draw is a meaningful result. A team can have a modest share of wins and a high points total if it draws often.

Fantasy and rating systems often replace the straight ratio with an expected-win figure based on runs or goals scored and conceded, because margins carry information that a bare win or loss throws away. These are estimates and should be labelled as such, whereas the ratio of wins to games is a straight count of results.

How many more wins to reach a target

A common question halfway through a season is what record is required to finish at a given percentage. Work from the finish: the target times the total games gives the wins needed in all, and subtracting the wins already banked gives the wins still required.

Always round the wins needed up. 22.8 wins is not possible, and 22 would leave the team short of the target. If ties count as half a win, a tie contributes 0.5 towards the total.

Reading a win percentage with care

A percentage is also the wrong tool when the number of games is very small. After two matches the only possible values are 0%, 50% and 100%, and a single result moves the figure by half the scale.

• Sample size matters. 3 wins from 4 matches is 75%, but it is far less informative than 30 from 40.
• Strength of schedule is hidden. A team that plays weaker opponents will have a better percentage without being better.
• Home and away splits can differ widely, and a total can hide a team that wins only at its own ground.
• The ratio ignores margin. A team that wins every match by one run and one that wins by ten look the same.
• Win-loss ratio is a different thing. 22 wins to 13 losses is a ratio of 1.69, not a percentage; do not interchange them.