The Five-Substitution Rule and the Final Twenty Minutes: The War of Attrition in Modern Football

In the 71st minute of a midweek Premier League fixture, I wrote a single line in my notebook: the home side had used four substitutions within seven minutes. By the 78th minute, their PPDA — the number of passes an opponent is allowed before each defensive action — had fallen from 11.4 to 8.9. The away side had made no changes. Twelve minutes later, the score moved from 1-1 to 2-1.

I am not recounting that moment for colour. I am recounting it because it has repeated often enough across the last three seasons to become a rule, and because it originates in a rule change only a few lines long in the Laws of the Game. Since the 2026-23 season, leading competitions have been permitted to allow each team five substitutions across three stoppages. The Premier League voted it through. It sounds like an administrative detail. In substance, it is a restructuring of power on the bench.

On 8 May 2026, the International Football Association Board (IFAB) approved a temporary provision allowing five substitutions, to reduce load on players as competitions faced congested calendars after the pandemic. At first it was a medical measure. In June 2026, IFAB made the option permanent in the Laws of the Game, and the Premier League adopted it immediately from 2026-23. Before that, the standard of modern football was three substitutions, accompanied by a substitute squad built on entirely different logic.

This shift changed how coaches plan an entire match, how scouts value a substitute, and how clubs allocate their wage budgets. A bench of real quality became a tactical asset to be exploited match by match, rather than an insurance policy against injury. When I track matches to build my models, I am forced to divide the game into fifteen-minute segments, because the effect of the new rule is not evenly distributed. It clusters into the final third of the match.

The 0-15 minute segment remains the original contest. Both teams are fresh, substitutions are irrelevant, and every predictive model I run assigns a low weight to the bench variable. The 15-30 minute segment is the first adjustment: a full-back pushing higher, a midfielder dropping deeper to keep the ball. The 30-45 minute segment is the zone of quiet attrition, where pressing intensity begins to fall, and that is why so many goals arrive in first-half stoppage time. The second half opens with the 45-60 minute segment at the highest tempo of the match, because both teams have just rested and have not yet hit the fatigue threshold.

Then comes the 60-75 minute segment, where the five-substitution rule leaves its clearest mark. This is the window in which the modern coach intervenes deliberately, usually introducing two or three players at once. In my tracking data, this is the segment showing the largest PPDA drop for the side making changes, and also the segment with the sharpest rise in two-way xG. The 75-90 minute segment plus stoppage is the true war of attrition, where the gap in bench quality becomes a gap in points. The key point is that the new rule does not make football faster across the whole match; it concentrates all the volatility into the final third, turning the last twenty minutes into the place where the table is decided.

Notably, the rule permits substitutions across three stoppages, plus a change at half-time. Half-time becomes a free tactical window: a coach can swap two players without spending a single window. I once tracked a Championship side using exactly this strategy for a full season, and their win rate in matches where they led at half-time ran well above the division baseline. When you can correct mistakes at half-time while preserving all three windows for the second half, you control the rhythm of a match in a way the previous generation of coaches never could.

I began paying attention to this mechanism in 2026, when the Premier League returned after the pandemic with the five-substitution rule. I tracked Liverpool across twenty matches and noticed they intensified their pressing in the 60-75 minute segment, exactly when opponents tended to make three changes at once. My data showed Liverpool's xG rising by 0.23 after substitutions. A lecturer dismissed the research direction, but my piece later caught the eye of an analyst at Burnley. That episode taught me one thing: data does not need a guarantor, it needs a correct reader. Virgil van Dijk and Mohamed Salah benefited most at the time, not because they ran more, but because their opponents ran less in the exact segment where space opened up.

Burnley at the time were a club on a tight budget. They could not buy a deep bench, so they used the rule differently: substituting earlier, aiming to preserve defensive structure rather than add firepower. Brentford later went further, building bespoke data profiles for each substitute to know who could sustain which tempo from which minute. For them, the bench is a portfolio, not a list of names.

At the other end of the financial spectrum, Manchester City won the treble in 2026-23 with a bench that could field a starting eleven capable of competing in the Champions League. Pep Guardiola routinely used all five substitutions, even when leading, because he treated them as a means of controlling tempo rather than reacting. When a side can bring Erling Haaland on in the 60th minute or withdraw him in the 75th without losing structure, the advantage lies not in any individual but in the ability to distribute load.

I learned to read pressing from an older lesson. In 2026, at eighteen, I wrote an analysis of RB Leipzig's 4-2-2-2 under coach Ralph Hasenhüttl, after a 4-1 win over Freiburg. The piece focused on how Timo Werner moved into space behind the defensive line. A journalist left a mocking comment. Instead of arguing, I re-watched fourteen Leipzig matches, counted 212 pressing actions, and published them with a heat map. Prejudice is simply noisy data the market has not yet learned to process.

By the same logic, in the summer of 2026, before the World Cup quarter-final between France and Uruguay, I predicted a France win through set pieces. I cited the number of goals France had scored from free kicks in the group stage. An editor spiked the piece. I sent an analysis of forty-seven set-piece situations through internal email. When France won 2-0, with Raphaël Varane heading in from an Antoine Griezmann free kick, the article ran and carried my name. I tell this story because it connects directly to the current topic: both set pieces and substitutions are zones where data is dismissed until the match forces people to look again.

The 2026 World Cup final is the most extreme example of the war of attrition. Argentina and France dragged each other into extra time, Kylian Mbappé scored a hat-trick, and the match ended from the penalty spot. Once a match passes the 90th minute, the decisive variable is no longer the original tactical plan, but who still has the fitness to execute it. That is precisely the zone the five-substitution rule expands: it allows a side to extend its capacity to execute by twenty more minutes, provided the bench has enough quality.

I do not prophesy. I only read data one step faster than everyone else. And the data of the last three seasons shows me a clear trend: sides with deep benches do not merely win more often late on, they actively manufacture the late phase on their own terms.

From the 2026-24 season, competitions applied stricter timekeeping for stoppage time, extending the real duration of matches considerably. A rule changes by one line, and football philosophy changes for a generation. When every match gains a few extra minutes on average, the value of a substitute rises, because the attrition window widens. This is the kind of chain reaction I always look for when reading a rule change: not what the law says, but how it shifts behaviour. In 2026, football also began trialling permanent concussion substitutions. This is the next regulatory layer that could change how clubs hedge risk. A side must now account for the possibility of losing a player to concussion, meaning it needs a bench that is not only deep but balanced across the lines.

The transfer market is a chess game in which spectators only see the pawns move. The price of a midfielder who can play the final thirty minutes has risen, because that is the profile the five-substitution rule has made valuable. Clubs now value substitutes by minutes of impact, not merely by starts. A mid-table side can sell a winger for more than his true worth, simply because his fitness profile suits the final twenty minutes. That is a valuation inflation few notice, yet it is quietly reshaping contracts.

But there is a blind spot the majority overlooks. Five substitutions do not automatically make a team stronger. They only make it more flexible, provided the coach uses them well. I have seen plenty of matches where a side burned all five changes at once in the 65th minute, then lost the ability to react to an injury or a red card in the 80th. When you spend your ammunition too early, you tie your own hands in the most important part of the match.

The second risk is broken rhythm. A side playing well can lose its structure if three incoming players do not grasp the current pressing model. Substituting by reputation differs from substituting by function. This is where good coaches separate themselves from average ones: they replace a role, not a name.

The common argument holds that the five-player rule makes football more unequal, because richer clubs have better benches. That is partly true. But it ignores how small-budget sides like Burnley or Brentford turned a limitation into an advantage by using substitutions to preserve structure rather than add firepower. Inequality lies in bench quality, and bench quality can be optimised with data, not only with money. This is where I part ways with the majority: the problem is not the rule, but how the rule is read.

There is one further layer rarely discussed: a player who knows he will be withdrawn in the 70th minute allocates effort differently. He either saves energy for the late phase, or pours it into the early phase to impress the coach. Both choices alter the match's intensity curve, and that curve is what I measure, not what I guess. When a side substitutes by the clock rather than by the flow of play, players learn to play safe for forty minutes to keep their place for the last ten. That is a psychological consequence no data table records directly, yet it surfaces in sideways passing metrics and a steady decline in duels contested.

So what deserves attention in the remainder of the season? I will look at the 60-75 minute segment of the sides fighting relegation. If a team's PPDA rises in that segment, it is losing its capacity to apply pressure when opponents make changes, and that is often an early sign of a defeat before the scoreline says so. Every number is a testimony. My job is to make sure they cannot lie.