Germany's 83% historical shootout win rate. England's 14%. Real pattern or small sample noise? Here's what 60 years of World Cup penalties actually tell you.
England have a 14% historical World Cup penalty shootout win rate. Germany have 83%. Italy have around 50%. France have a strong record; Spain a weak one.
Some of these patterns feel deeply causal. Germans are clinical; the English crack under pressure; Italians have steel. The narratives have been told for decades. They're embedded in how every commentator describes a knockout match heading toward extra time.
But here's the question worth asking: are these patterns real signal, or are they the small-sample noise of an inherently random event?
Our model has to make a decision about this. So does any tournament forecaster. This post is about what the data actually says, what we do, and why even the best answer is unsatisfying.
A single penalty kick is roughly a 75-80% conversion rate at the elite level. That's the average across thousands of penalties from competitive matches.
The variance around that 75% is influenced by:
So a shootout penalty is roughly a coin flip with the coin biased toward the taker β about 65-70% conversion in tournament settings.
A typical shootout takes 5 kicks per side minimum, often more. With each kick being an independent-ish event, the outcome of a shootout is the product of 10+ individual probabilistic events. The variance on the total is huge.
Across all World Cup shootouts since they were introduced in 1982, here's what we see:
These numbers are stable. They give a baseline for what shootouts look like generically.
What they don't reliably tell us is whether specific teams have systematic shootout edges.
Germany have won 6 of their 7 World Cup shootouts. England have won 1 of their 4.
Germany's record is striking. So is England's. But seven shootouts spanning 40 years is a tiny sample for a probabilistic event. Run 7 fair coin flips and the chance of getting 6+ heads is about 6%. Not impossibly rare. Across all the national teams that have been in shootouts, somebody was going to look streaky just by chance.
The harder question: is there real causal structure?
The case for real structure:
The case against:
The honest position is that there's probably some signal β penalty preparation and culture probably do affect outcomes by a few percentage points β but it's smaller than the headline records suggest, and the rest is noise that the brain has organised into a story.
We have to make a choice. Three options:
Option 1: Assume penalty shootouts are pure 50/50. Treat them as coin flips, ignore historical records. This is the most conservative approach.
Option 2: Apply small adjustments based on long-run records. Give Germany +5%, England -5%, France +3%, etc. This honours observed patterns while being humble about small samples.
Option 3: Model shootouts based on current squad composition. Track individual penalty conversion rates for the players in each squad, predict shootouts based on those rather than national records. Most theoretically rigorous, but data-hungry.
We use a hybrid of options 1 and 2. Specifically:
This means our predictions have low confidence on shootouts. When we say "if it goes to a shootout, this team is 53%," we mean it. Not 80%. Not 65%. Just barely better than half.
Some plausible factors we deliberately leave out:
Goalkeeper specialty. Some keepers are reportedly better at penalty saves. Marcin BuΕka, Diego Alves, etc. β specialists with measurable above-average save rates. Including this would require player-level data we don't currently maintain. The improvement would be small (a percentage point or two), and we don't think it's worth the complexity.
Order effects. Going first matters; the player taking the 5th kick to win/lose matters. We could model these effects but the sample data is noisy.
Recent shootout experience. A team that just won a Euros shootout might carry confidence into the World Cup; a team that just lost one might carry baggage. We don't model this. The psychological effects are real but the data is too sparse.
We may add some of these in future iterations. The question is always whether they add real predictive power or just give the model more knobs to be wrong with.
About 25% of World Cup knockout matches go to shootouts. Across a tournament with 16 knockout matches, you'd expect 4 to be decided on penalties.
If shootouts were pure coin flips and we treated them as such, our tournament-winner probabilities would have a specific shape: any team that needed to win multiple shootouts to lift the trophy would have a meaningful penalty-related drag on their probability.
If we applied strong national adjustments (Germany +20%, England -20%), our probabilities would shift toward Germany and away from England in any path that involves potential shootouts.
Our middle approach means our probabilities are somewhere in between. England's tournament probability isn't penalised heavily for shootout history, but it's not ignored either. The model says: shootouts are mostly random, but we acknowledge a small residual signal in long-run records.
If this seems like a frustratingly grey answer β that's because the underlying reality is grey. There isn't a clean truth here, and pretending there is would mean either over-reading the records or ignoring obvious patterns.
The 2026 World Cup will probably feature 4-6 shootouts. They'll either confirm or push back on the patterns the long-run data suggests.
If Germany lose a shootout to a team they'd be slight underdogs to in regulation, the "Germans are unbeatable on penalties" narrative will take a hit. If England win a shootout from a losing position, the "England can't do penalties" narrative softens.
We'll publish how our model did. If we under-predicted England's shootout probability and they progressed, we owe you an honest write-up. If our 53% / 47% calls were 50% calibrated across many shootouts, that means the model was honest about how unpredictable they are.
Penalty shootouts are mostly coin flips. The patterns that look like skill or psychological edge are probably partly real and partly noise, in proportions we genuinely cannot disentangle from the data we have.
Anyone who tells you they can confidently predict a shootout is fooling themselves. Including the bookmakers; their shootout-related odds are wider than their match-result odds for exactly this reason.
When you see a knockout match heading toward extra time and the commentator tells you "this team's record favours them in shootouts" β they're probably overstating the effect. The real answer is: it's nearly a coin flip. With a slight nudge from preparation and culture, but mostly a coin flip.
That's not a satisfying answer. It's the honest one.
OddsIQ provides AI analysis, not financial or betting advice. Past performance does not guarantee future results. Gamble responsibly: BeGambleAware, GamCare, GamStop.