For most of its modern history, football let a manager change three players a match, and an entire grammar of the game grew up around that number. You held a substitution back for the last twenty minutes. You gambled your final change on chasing a goal. A tiring star stayed on because the bench was a scarce resource. Then, almost without anyone treating it as a landmark, the number changed. Five substitutes, not three. What that did to the closing stages is a question the Premier League’s own goal minutes can answer, and the answer below is not the one the debate assumed.

What actually changed, and when

The shift began as an emergency measure. When football restarted in 2020 after its COVID-19 shutdown, fixtures were compressed into punishing schedules, and the game’s lawmakers approved a temporary option allowing teams to make up to five substitutions rather than three, to protect players from injury in the congestion. It was explicitly framed as provisional: a response to an unusual situation, not a permanent rewrite of the Laws of the Game.

It did not stay provisional. Competition by competition, the option was extended, and it was subsequently written permanently into the Laws of the Game as a standing allowance rather than a temporary exception. The Premier League took a detour on the way: it used five substitutes for Project Restart in June and July 2020, went back to three for 2020-21 and 2021-22, and adopted five for good only from 2022-23. The five-substitute rule is now simply how football works at the top level, and a generation of players will grow up never having known the three-sub world their predecessors managed within.

3 → 5
Substitutes allowed
3
Windows to use them
2020
Temporary, then made permanent

Five subs and late goals: a Premier League before-and-after test

Added November 6, 2026. The Premier League gives an unusually clean test of the rule, because it switched twice. It used five substitutes for Project Restart in June and July 2020, went back to three for 2020-21 and 2021-22, and adopted five for good from 2022-23. A season later, from 2023-24, its referees also began adding the time actually lost to stoppages, which made second halves longer. That leaves 2022-23 as the one season with five subs and the old added time. We timed every goal ESPN records from 2016-17 to 20 September 2026, 3,848 matches, and measured the share scored after the 75th minute.

Five subs on their own did not bring more late goals. In 2022-23, 20.0% of goals came after 75′, the lowest share of the ten full seasons in the sample. The two three-sub seasons before it had 21.9%. The difference, −1.9 points, has a 95% interval of −4.9 to +1.1, so the data rules out a large rise but not a small effect in either direction. Counted per match rather than as a share, minutes 76 to 90 produced 0.442 goals a match in 2022-23 against 0.443 in the two seasons before. The 92 Project Restart matches, with five subs, had 22.4% after 75′ against 22.8% in the three-sub seasons before the shutdown; a sample that small, played behind closed doors, says little either way.

The rise since 2023-24 is added time. Across 2023-24 to 2025-26, 24.3% of goals came after 75′, 4.3 points more than in 2022-23 (interval 1.5 to 7.1). Nearly all of it sits after 90:00. Minutes 76 to 90 carried 15.9% of goals against 15.5% (+0.4 points, interval −2.1 to +2.9), while second-half added time went from 4.5% to 8.4% (+3.9 points, interval 2.4 to 5.5), or from 0.129 to 0.252 goals a match. In every full season since 2016-17, minutes 76 to 90 carried between 15.4% and 17.6% of goals, whichever rule applied.

Premier League goals after the 75th minute by rule period, 2016-17 to 20 September 2026: 3,848 matches whose goal list adds up to the score. “After 75′” = a goal ESPN shows at 76′ or later, split into regulation minutes 76 to 90 and second-half added time (90′+N′). “Second half ends” = the average added minute in which ESPN shows the second half ending. Source: ESPN public scoreboard, /data_layer/epl_goal_times_2002_2027.json, and match summaries, /data_layer/epl_added_time_2016_2027.json, both pulled 5 October 2026.
PeriodSubsMatchesGoalsAfter 75′76′-90′90′+Second half ends
2016-17 to March 202031,4263,93222.8%17.0%5.8%90+5.2
Project Restart, June-July 202059225022.4%17.6%4.8%90+5.8
2020-21 and 2021-2237602,09521.9%16.1%5.8%90+5.2
2022-23 (old added time)53801,08420.0%15.5%4.5%90+6.1
2023-24 to 2025-26 (new added time)51,1403,40624.3%15.9%8.4%90+7.6
2026-27 to 20 September (partial)55014122.7%15.6%7.1%90+6.2
Two stacked panels by Premier League season, 2016-17 to 2026-27, with Project Restart and 2019-20 before it shown separately and the five-substitution periods shaded. Top: the share of all goals scored after the 75th minute, split into regulation minutes 76 to 90 (green) and second-half added time (gold). The green part stays between 15.4% and 17.6% in every full season, including 2022-23, the first permanent five-sub season, when the total was 20.0%. The gold part grows from 2023-24, when added time changed: 8.3%, 7.2% and 10.0%. Bottom: the second half ended on average at 90+5.0 to 90+6.1 in each season up to 2022-23 and at 90+8.1, 90+7.1 and 90+7.5 after.
Top: the share of Premier League goals scored in minutes 76 to 90 (green) and in second-half added time (gold), by season; five-substitution periods shaded, the dashed line marks the 2023-24 added-time change. Bottom: the average added minute in which the second half ended. Data: ESPN public Premier League scoreboard and match summaries, both pulled 5 October 2026; 3,848 matches; computed by our analysis script.

Longer halves, not livelier ones. Before 2023-24, ESPN’s clock shows the second half ending at 90+5.3 on average; since, at 90+7.6. Counting the minute of the final whistle as the added minutes played, second-half added time produced 2.89 goals per 100 added minutes before the change (411 goals in 14,205) and 3.32 since (287 in 8,634). The difference, +0.43, has an interval of −0.04 to +0.91: a slightly higher rate, but not one the data can separate from chance. At the old rate per match, the three seasons since 2023-24 would have produced 176 added-time goals; at the old rate per added minute, 250. They produced 287. Two-thirds of the extra goals (74 of 111) is the extra minutes alone.

What the data cannot tell apart. From 2023-24 the two changes run together. There is no Premier League season with long added time and three subs, so whether fresh substitutes make those extra minutes more dangerous cannot be tested here; the small rise in goals per added minute is where such an effect would show, and it is within chance. 2022-23 is a single season, which is why its interval spans six points. Substitutions themselves stop play, and second halves already ran longer in 2022-23 (90+6.1, against 90+5.2 in both three-sub periods), but the archive does not record why. It holds no substitution events either, so this tests the rule, not the substitutes: it cannot say how many changes teams made, when they made them, or whether a substitute scored. Other things moved between seasons too: 2020-21 was played almost entirely without crowds, and 2022-23 stopped for a winter World Cup. For what the longer added time did to one-goal leads, see how safe a one-goal lead is; for how the score itself changes scoring rates late on, see game state and score effects.

How the numbers are made. ESPN’s public Premier League scoreboard gives the minute of every goal and its match summaries the minute each half ended, both pulled 5 October 2026: 3,850 matches from 2016-17 to 20 September 2026, of which 3,848 have a goal list that adds up to the score. A goal shown at 76′ or later counts as after 75′ (ESPN’s minute 76 covers 75:00 to 76:00); 90′+N′ is second-half added time; a first-half added-time goal counts before 75′. One 2018-19 match has no recorded end time and is left out of the half-length figures only. The intervals are normal approximations: two proportions for shares, two Poisson counts for rates.

The detail that matters: windows, not just numbers

The rule is not simply "five changes instead of three." To stop teams from chopping the game into fragments with constant stoppages, the additional substitutions come with a constraint: a team may make its five changes across only three substitution windows during normal time (the half-time interval does not count as one of them). Send on two players at once and you have used a window but only one of your three opportunities to stop play.

That detail shapes the tactics as much as the raw number does. A manager now thinks not only about which five players to introduce but about how to cluster them: making a double or triple change in a single window to preserve a future stoppage, or spending a window early to fix a problem and accepting fewer interventions later. The arithmetic of in-game management got richer overnight.

How it changed the game on the pitch

The first and most obvious effect is simply fresher legs in the final stretch. With five changes available, a manager can refresh an entire spine (a midfielder, a winger, a striker) in the last half-hour without exhausting his options, and can do so while still holding a change in reserve for an injury or a tactical surprise. Players who once had to be nursed through ninety minutes can be used in shorter, higher-intensity bursts, both those coming off and those coming on.

The second effect is tactical flexibility. Two extra substitutions are two extra chances to change the shape of the team in response to the state of the game. A side defending a lead can swap attackers for defenders and genuinely reconfigure rather than make a single token change. A side chasing a goal can throw on multiple fresh attackers at once. The substitutions article of the Laws became, in effect, a wider set of levers, and the managers who pull them well gained an edge over those who do not.

The third effect is the one with the longest reach: squad depth became more valuable. When you can only change three players, the quality of your fourth-to-sixth options matters, but is capped. When you can change five, the gap between your starters and your best reserves is exposed for a larger share of the match. A deep squad can now keep its intensity high for ninety-plus minutes by rotating fresh quality onto the pitch; a thin one cannot, and the difference shows most clearly when legs tire.

The competitive-balance debate

This is where the rule stops being a neutral tweak and becomes contested. The objection, raised by a number of managers at clubs without enormous resources, is straightforward: the teams best placed to exploit five substitutions are the ones with the deepest, most expensive benches. If your reserves are internationals and your opponent’s are academy graduates, then doubling the number of players you can introduce widens the quality gap precisely in the part of the game, the closing minutes, where tired legs already favour the stronger side.

Read directionally, the logic is hard to dismiss: a rule that increases the impact of bench quality should, all else equal, tilt the closing stages toward squads with more of it. The counter-argument is that fresher players reduce injuries and raise the overall standard of the spectacle, and that depth has always been an advantage of richer clubs in one form or another. We should be careful not to overstate the size of the effect (matches are decided by many things, and a deep bench is no guarantee), but the direction of the incentive is clear, and it points toward the clubs that were already favoured.

What squad depth means at a 48-team World Cup

This section condenses a separate article first published on 10 June 2026.

Going into the 2026 World Cup, two more forces stacked on top of the rule. The expanded format lengthens the bracket: with 48 teams and a knockout stage that begins at 32, the eventual champion plays more games than a winner used to, often on short rest, with more accumulated fatigue and more exposure to suspensions and knocks. And a summer tournament across the region raises the prospect of serious heat for some fixtures, with one host city at major altitude: conditions that raise the physiological cost of the same running. The rule gives you the tool; the conditions create the need.

Why depth compounds in 2026
Five subs (use nearly half the outfield) × a longer bracket (the winner plays more games, on short rest) × heat and altitude (every player tires faster) = the bench decides more matches than it used to.

"Depth" is often used loosely, so it helps to split it into three components. Drop-off is how far a team's quality falls from its first eleven to its replacements: a gentle drop-off means substitutions maintain the level, a steep one means every change weakens the side. Positional cover is whether there are real options in every line, so a suspension at centre-back or a hamstring in midfield does not force a square peg into a round hole; a squad can look deep on paper and still be thin in the one position that gets injured. Rotation viability is enough quality to rest starters between matches without conceding the game, which over a long bracket keeps the best players fresher for the rounds that matter.

Most power ratings capture squad strength only indirectly, through results and chance quality, and do not natively know how a team's twelfth player compares to its first. That makes depth one of the adjustments a careful analyst makes on top of a rating, and one reason models can disagree about teams whose bench quality diverges from their headline strength. It also feeds the dark-horse calculus: a mid-tier side with surprising depth is better placed to overperform than its rating alone implies.

What is actually measurable

The honest position is that the rule’s effects are easier to reason about than to pin to a precise number, and you should be wary of anyone quoting one confidently. But several things are, in principle, measurable from public data, and worth looking at rather than asserting.

  • Late-game goal timing. If fresh legs late in matches change the rhythm, the distribution of when goals are scored across the ninety minutes is the place it would show up. Comparing the shape of that distribution before and after the change is a clean, checkable question, and we run it for the Premier League above.
  • Substitute usage and impact. How early changes are made, how often managers cluster them into one window, and how often substitutes are directly involved in late goals are all countable. The "impact sub", a player brought on specifically to change a game, is now a more deliberate tactical category than it was.
  • Minutes distribution across squads. Whether playing time has spread across larger groups of players, reducing the share carried by a small core of starters, is straightforward to track season over season.

Each of those is a study someone can run against league data rather than a figure to quote from memory, and that is the right spirit in which to treat the change.

Why it matters beyond the substitutions themselves

The deeper point is that the five-substitute rule quietly raised the value of game management as a tactical discipline. The closing minutes have always been where matches are decided, and the rule put more tools in the manager’s hands for exactly that phase, which is also the phase where transitions and counterattacks are most lethal, as tiring defences lose their shape and fresh attackers exploit the gaps. A side that can introduce a fresh, athletic fullback for the final half-hour can keep attacking down the flanks with intensity long after a thinner squad has run out of legs in the wide areas, precisely when those areas open up.

None of this is visible in the way a new offside law or a points deduction is visible. There was no single dramatic match that announced the five-substitute era. But the grammar of the closing minutes (who controls them, which clubs they favour, how a manager spends his levers) was rewritten by a rule that arrived as a temporary measure and quietly stayed. That is often how the biggest changes in football actually happen.

Reproduce it

Every figure in the Premier League section (added November 6, 2026) comes from our analysis script. It reads data_layer/epl_goal_times_2002_2027.json (ESPN’s public Premier League scoreboard, pulled 5 October 2026) and data_layer/epl_added_time_2016_2027.json (ESPN match summaries, pulled 5 October 2026) through our archive loader, which drops the 357 duplicated 2009-10 rows. It sorts every match from 2016-17 into its rule period, counts the goals after 75′ with and without added time, sums the half lengths, computes the intervals, writes every number to data/stats/five-substitutions-era-late-goals.json and draws the chart. No network, nothing hand-entered.

Sources & further reading

  • ESPN Premier League goal minutes and half lengths: the archive behind the before-and-after test (ESPN public scoreboard and match summaries, both pulled 5 October 2026).
  • The IFAB: the body that writes the Laws of the Game, including the substitutions law and its amendments.
  • FBref: match data including substitution timing and minutes played, for studying late-game effects.
  • StatsBomb: event data for analysing how substitutions change late-match intensity.
  • StatsBomb open data: free match events usable to study goal timing and substitute involvement.

Published by SoccerAnalytics from public data and the sources named in this piece. How this site is made.