← All posts
1990 – 2026 · A Data Story

Sent On to Miss?

Managers gamble on fresh legs for the shootout. Across 35 years of penalties, does the substitute gamble actually backfire?

72.0%
baseline conversion (all kicks)
57.1%
specialist takers (105'+)
50.0%
dying-minute subs (110'+)
↓ SCROLL

The 119th-Minute Substitution

It's a move we see more, not less. In an analytics era chasing the highest-percentage option, managers increasingly send on a fresh, specialist taker in the dying minutes, purely for the shootout.

On 29 June 2026 the Netherlands did exactly that, then watched substitute Justin Kluivert hit the post and go out of the World Cup. England fans remember Euro 2020: Rashford and Sancho, on in the 119th minute, both missed.

There's something cruel about it. The player barely touches the game, then the whole tournament rests on his one kick.

So does the gamble work? I built a dataset of 840 shootout kicks from 88 men's international shootouts (World Cup, Euros, Copa América; 1990–2026) to find out.


How Often Does a Shootout Penalty Go In?

First, the yardstick. Across all 840 kicks, the overall conversion rate is 72.0% (95% CI 68.9%–75.0%).

Roughly three in four shootout penalties are scored. Missing is the exception, which is exactly why a missed sub stands out.

For scale: even Lionel Messi has converted about 78% of his career penalties (114 of 147, shootouts aside), not far above this baseline. The spot-kick humbles almost everyone.

72.0%
all shootout penalties
(840 kicks)
72.8%
taken by a starter
(486 kicks)
67.3%
taken by a substitute
(269 kicks)

The Cost of the Late Gamble

Line the groups up against the 72.0% baseline (dashed). All substitutes dip only slightly; it's the dying-minute subs, sent on at 110'+, who fall away.

The late groups are subsets of all substitutes and carry small samples (dying-minute subs: n=20), so read them as indicative.


Conversion by Role

Each taker is labelled by how they got onto the pitch: a starter, or a substitute grouped by when they came on. The dashed line is the 72.0% overall baseline.

RoleKicksScoredConversion95% CI
Starter48635472.8%68.7%–76.6%
Sub before 90'20313868.0%61.3%–74.0%
Sub 90–104'312374.2%56.8%–86.3%
Specialist takers (105'+)352057.1%40.9%–72.0%

A few older Copa América matches lacked published lineup data; those kicks are counted in the baseline but excluded from the role split. Lineups for the live 2026 fixtures are taken from match reports.


It's the Specialist Taker Who Pays

It's not substitutes in general who falter. Across all subs, the gap (67.3% vs 72.8% for starters) is only borderline.

The effect sits with the specialist takers, sent on at 105'+ largely to take a kick. They convert just 57.1%, and the most extreme, on at 110'+, only 50.0%.

Since the hypothesis is directional, subs predicted to do worse, the one-sided test is the relevant one; the two-sided figure is shown too.

ComparisonConversiontwo-sided pone-sided pVerdict
Starters vs all substitutes72.8% vs 67.3%0.1080.054not significant
Starters vs specialist takers (≥105')72.8% vs 57.1%0.0460.023significant
Starters vs dying-minute subs (≥110', n=20)72.8% vs 50.0%0.0260.013significant

A logistic regression agrees: being a substitute multiplies the odds of scoring by 0.77 (95% CI 0.55–1.06, p = 0.108). A substitute's scoring odds are roughly 23% lower than a starter's. For comparison, published tournament studies put substitutes near 61% against 74% for starters.

Fresh Legs Help, Until the Last Gasp

Fresh legs aren't the problem. Substitutes who come on with time to settle hold up fine, including those sent on early in extra time (90–104'). Only at 105'+, with barely a touch before the kick, does the line drop below the 72.8% starter mark (dashed).

The fresh-legs gamble looks worst exactly where managers reach for it most: the last-gasp specialist, sent on at 105'+ to take a kick.

The Specialist Takers, One by One

The closest cases to the hypothesis: every taker sent on at 105'+, often with the shootout in mind. As a group they scored 20 of 35 (57.1%).

Yet these are, on the whole, good penalty takers. In open play, the 29 with a career record convert 88% (459/523) — several with flawless records (Sarabia, Sancho, Willian) still missed here. That gap, 88% in normal play versus 57.1% in the shootout, is the point.

Player Team Tournament On Result Career pens Conv.
Fabián BalbuenaParaguayFIFA World Cup 2026120'❌ Missno data
Simone ZazaItalyUEFA Euro 2016120'❌ Miss4/4100%
Paulo DybalaArgentinaFIFA World Cup 2022120'✅ Scored43/4988%
Badr BanounMoroccoFIFA World Cup 2022119'❌ Saved10/1377%
Marcus RashfordEnglandUEFA Euro 2020119'❌ Post21/2391%
Jadon SanchoEnglandUEFA Euro 2020119'❌ Saved3/3100%
Jamie CarragherEnglandFIFA World Cup 2006119'❌ Missno data
RodriSpainUEFA Euro 2020118'❌ Saved3/3100%
Thomas StrunzGermanyUEFA Euro 1996118'✅ Scored4/580%
Pablo SarabiaSpainFIFA World Cup 2022117'❌ Post12/12100%
Zeki AmdouniSwitzerlandUEFA Euro 2024117'✅ Scored7/7100%
Trent Alexander-ArnoldEnglandUEFA Euro 2024114'✅ Scoredno data
Justin KluivertNetherlandsFIFA World Cup 2026113'❌ Miss7/888%
Mislav OršićCroatiaFIFA World Cup 2022113'✅ Scored9/1182%
ElanoBrazilCopa América 2011112'❌ Miss27/3090%
Marcus RashfordEnglandFIFA World Cup 2018112'✅ Scored21/2391%
Artem MilevskyiUkraineFIFA World Cup 2006111'✅ Scored3/475%
Marcus ThuramFranceUEFA Euro 2020110'✅ Scored5/683%
Nadiem AmiriGermanyFIFA World Cup 2026110'✅ Scored16/1889%
Walter GarganoUruguayCopa América 2011110'✅ Scoredno data
Ivan ToneyEnglandUEFA Euro 2024109'✅ Scored58/6294%
Xherdan ShaqiriSwitzerlandUEFA Euro 2024108'✅ Scored26/3476%
Lorenzo InsigneItalyUEFA Euro 2016108'✅ Scored45/5680%
Milan BadeljCroatiaFIFA World Cup 2018107'❌ Saved1/1100%
Henrik LarssonSweden men'sFIFA World Cup 1994107'✅ Scored36/36100%
Sylvain WiltordFranceFIFA World Cup 2006107'✅ Scored6/6100%
WillianBrazilFIFA World Cup 2014106'❌ Miss9/9100%
Federico BernardeschiItalyUEFA Euro 2020106'✅ Scoredno data
Marko LivajaCroatiaFIFA World Cup 2022105'❌ Post37/4484%
João FélixPortugalUEFA Euro 2024105'❌ Post10/1191%
Josip IličićSloveniaUEFA Euro 2024105'❌ Saved32/4080%
Gonzalo MontielArgentinaFIFA World Cup 2022105'✅ Scored8/989%
Lovro MajerCroatiaFIFA World Cup 2022105'✅ Scored16/1794%
Bradley BarcolaFranceUEFA Euro 2024105'✅ Scored0/10%
Thiago AlcântaraSpainUEFA Euro 2020105'✅ Scored1/1100%

Tip: click any column header to sort. Career pens = a player's open-play penalties (scored/taken) and conversion rate, mostly Transfermarkt/FootyStats as of ~2026 — hover for the source. These are career totals (not point-in-time before the kick) and exclude shootout penalties, which aren't tracked per player. Coverage is patchy for some (FBref was unreachable to cross-check; a few, like Bernardeschi, take penalties but have no clean figure), so read the numbers as indicative, not audited.


Baseline by Competition & Era

Conversion is remarkably stable across competitions and decades, so the substitute effect is not an artefact of one tournament or period.

CompetitionKicksConversion
FIFA World Cup30367.3%
UEFA Euro20973.7%
Copa América32875.3%
DecadeKicksConversion
1990s23179.2%
2000s14369.9%
2010s25272.6%
2020s21465.0%

Why This Isn't a Clean Experiment

A substitute is not a random starter. Part of the gap may reflect player quality, not just cold legs.

But it often cuts the other way. The late sub is frequently a designated specialist, sent on precisely because he is one of the squad's best takers, with a strong record in open play.

That even these specialists miss more often suggests the conditions (a long wait, then one cold, decisive kick) matter more than the player. Either way this is observational data: the pattern is real, but it does not prove the substitution itself is the cause.

Scope

This analysis covers shootout penalties only. In-play penalties, taken in open play rather than a shootout, convert a touch higher (around 75% historically) and carry different pressures. Everything here is shootout-specific.

Methodology & Sources

840 shootout kicks from 88 men's international shootouts (FIFA World Cup, UEFA Euros, Copa América), 1990–2026. Each taker is classified as a starter or a substitute, with the substitute's minute of entry, by joining the shootout result to the match lineups.

Two sources. StatsBomb open data (event-level data with exact substitution minutes) is the ground truth for recent tournaments. Wikipedia knockout-stage pages, which carry both the shootout box and full lineups with substitution minutes, provide the historical coverage. The Wikipedia parser was validated against StatsBomb on the 18 shootouts present in both sources, matching 100% on outcome sequences and substitute counts.

Confidence intervals use the Wilson method; group comparisons use a two-proportion z-test (one- and two-sided) supported by a logistic regression. Lineups for the live 2026 fixtures were taken from published match reports; a handful of older Copa América matches still lacked lineup data and are counted in the baseline but excluded from the starter-vs-substitute split. The 2026 World Cup is ongoing, so figures are a snapshot.

Like this report? This is what I do for a living. Same method, usually pointed at revenue data instead of goals: lead scoring, pipeline analysis, forecasting for B2B companies. If your CRM has questions sitting in it, let's talk.

Elisha Amir is an independent data consultant and a lifelong football fan. This piece is built on a custom dataset of penalty-shootout kicks assembled from StatsBomb open data and Wikipedia, validated across sources. All charts are interactive.