PPDA explained: what passes per defensive action shows

A graphic goes up at half time. One side's PPDA is 6.1, the other's is 13.4, and within a few seconds somebody has concluded that the first team is pressing and the second is sitting off. That is roughly the right conclusion reached for roughly the wrong reason, and the distance between the two is where every misuse of this number lives.

PPDA stands for passes per defensive action. The PPDA pressing metric, explained without the folklore around it, is a fraction and nothing more: the passes an opponent is allowed to make in a defined slice of the pitch, divided by the defensive actions the other team makes in that same slice. No model. No training data. No probability. Two counts and a division, which is why it spread so fast and why it is misread so often.

That simplicity is both halves of the story.

PPDA is a proxy for pressing, not a measurement of it.

A proxy is a thing you can count standing in for a thing you cannot. Pressing is a coordinated act involving several players, most of whom never touch the ball, running at particular moments towards particular opponents to make particular passes unavailable. Event data records touches. It cannot record intent, and it barely records absence. So the metric counts the residue that pressing leaves behind in the event feed, which is a lot of challenges and not many opposition passes, and infers the pressing from that.

Proxies are useful precisely because they are cheap. They fail in specific and predictable ways, and knowing those ways is most of what separates someone using this number well from someone quoting it in an argument.

What the PPDA pressing metric counts, and what it leaves out

Take the numerator first, because it is the half people forget.

The numerator is opposition passes. Not opposition possessions, not seconds of opposition control, not touches. Passes attempted or completed by the team without the ball, inside the counted zone. Whether attempted or completed passes are used differs between implementations, and the choice matters more than it sounds: counting attempts includes the pass your press forced into a teammate's shins, which is exactly the outcome you wanted.

The denominator is defensive actions. In the original formulation the set is tackles, interceptions, challenges and fouls, where a challenge is a tackle attempt that failed. Wyscout's published glossary lists fouls, interceptions, won defensive duels and sliding tackles. Some secondary write-ups add clearances. These are not the same denominator, and a team's figure moves depending on which one a provider uses, which is the first reason to distrust any PPDA number quoted without a source.

The division makes it a rate, not a total. This is the part of the design that is genuinely clever. A team that faces an opponent hogging the ball for eighty minutes will rack up defensive actions simply through exposure, so a raw count of tackles in the final third would reward the team that defends most. Dividing by the opponent's passes normalises that away. What survives is a rate: the price, in passes, that the opponent pays for each time you interfere with them. A side with a PPDA of 5 is charging the opponent an interruption every five passes. A side on 15 is letting them play.

Read it that way and the metric becomes much easier to handle. It is a tax rate on opposition passing.

There is a technical rule that follows from it being a ratio, and almost every casual analysis breaks it. To get a team's PPDA across ten matches you sum all the opposition passes, sum all the defensive actions, and divide once. You do not calculate ten match figures and average them. Averaging ratios lets a match with very few opposition passes, a wet Tuesday against a side that hit everything long, carry the same weight as a match with four hundred. The two methods give different answers, and only the first one means anything.

The same reasoning explains why PPDA is never computed for a single passage of play. A five-minute spell with no defensive actions in the zone would require dividing by zero. The metric only exists at the aggregate level, which is a hint about the level at which it should be interpreted.

Why the metric ignores two-fifths of the pitch

If you counted defensive actions anywhere on the pitch, the deepest, most passive team in the division would top the table. Every team makes tackles and interceptions in front of its own box. Doing it there is not pressing, it is defending, and the whole point of the exercise was to separate the two.

So the metric is restricted to the part of the pitch where a press can plausibly happen: the area nearest the opponent's goal, where they are building and you are the one initiating contact. In the original construction the pitch is measured on a scale from zero to one hundred running from the pressing team's own goal line, and only events beyond the forty line count. Providers usually describe the same thing from the other direction, as the opponent's defensive three-fifths.

60%40%

• Counted: the opponent's defensive three-fifths60%
• Ignored: the pressing team's own two-fifths40%

The proportions are the metric's own rule rather than a measurement. The usual definition counts opposition passes and defensive actions only in the three-fifths of the pitch nearest the opponent's goal. Some implementations set the boundary elsewhere.

Show the numbers

The pitch PPDA looks at, and the pitch it ignores

Item	Value
Counted: the opponent's defensive three-fifths	60%
Ignored: the pressing team's own two-fifths	40%

Sixty per cent is a convention, not a law of the game, and it is worth saying plainly that providers do not all agree on it. The three-fifths line is the most common, and it corresponds to the whole of the opposition half plus a strip of your own. Other implementations restrict the zone to the opposition half, or shift the boundary to suit a particular league's build-up patterns. A figure computed on one boundary cannot be compared with a figure computed on another, and the difference is not small: moving the line by ten metres adds or removes a band of pitch through which a very large number of build-up passes travel.

The restriction also creates a hard edge with soft football on either side of it. A defensive action two metres inside the boundary counts in full. The identical action two metres outside counts for nothing. A mid-block that engages around the halfway line sits comfortably inside the counted zone, so it looks like a high press to the arithmetic. A block that engages ten metres deeper falls off the edge of the metric entirely and looks like nothing at all.

One further coding detail decides more cases than you would think. Passes are attributed by where they started, so a goalkeeper's long kick from inside his own box is an opposition pass inside the counted zone even though the ball finishes seventy metres away. The press that forced the goalkeeper to hit it long therefore adds one to the numerator it was trying to shrink.

Why a lower PPDA means more pressing, which is the bit that trips everyone up

The defensive actions are on the bottom of the fraction. That is the entire explanation, and it is still the most common error in coverage of the metric.

Work it with invented numbers. An imaginary team allows the opponent sixty passes in the counted zone and makes ten defensive actions there: sixty divided by ten is a PPDA of 6.0. Now the same team presses harder in an imaginary second match, allows the same sixty passes but makes twenty defensive actions: sixty divided by twenty is 3.0. Twice the aggression, half the number.

So a high press produces a low PPDA, and a high PPDA describes a team that is not pressing at all. Headlines get this backwards regularly, and the phrase "high PPDA" is now ambiguous enough in ordinary use that serious analysts avoid it and say "low" or "high" pressing intensity instead.

Two further consequences of the inversion are worth having in your head.

Ratios do not move linearly, so equal-sized changes in the figure do not mean equal-sized changes in pressing. Going from an invented 12.0 to an invented 6.0 means doubling your defensive actions against the same passes. Going from 6.0 to 3.0 means doubling them again, which is a far larger real change in behaviour for the same six-point move on the scale. At the aggressive end of the range, small differences in the number are large differences in football. At the passive end, large differences in the number are mostly noise.

Some analysts sidestep the whole problem by using the reciprocal, defensive actions per opposition pass. It runs in the intuitive direction, higher meaning more pressing, and it behaves properly under averaging and in regression models. It has never caught on publicly because PPDA got there first and the acronym is memorable, but where you see a pressing measure that rises with intensity, this is usually what it is.

PPDA explained through one passage of play

The arithmetic is easiest to trust when you watch it accumulate. What follows is an invented spell against an invented opponent, built to show which events land in which half of the fraction.

1. 1The zone is fixed before anyone kicks anythingOnly events beyond the 40 line, in the three-fifths of the pitch nearest the opponent's goal, will be counted. Everything the pressing side does in its own two-fifths is invisible to the metric no matter how hard it works there.
2. 2Goal kick, played shortThe opponent's goalkeeper rolls the ball to a centre-back inside his own penalty area, which is deep inside the counted zone. Numerator: 1.
3. 3A forward closes him down and never touches the ballThe centre-back's options narrow, his head goes down, and he plays the safe pass. No event is recorded. Twenty metres of hard running at exactly the right moment contributes nothing to either half of the fraction.
4. 4Centre-back to full-backNumerator: 2.
5. 5The winger slides in and missesA failed tackle attempt is a challenge, and a challenge is a defensive action in the standard denominator. The full-back goes past him and the move continues. Denominator: 1.
6. 6Full-back back to the goalkeeperNumerator: 3.
7. 7The goalkeeper hits it longPasses are attributed by where they start, so a seventy-metre kick from inside the box is an opposition pass inside the counted zone. Numerator: 4. The press has just been rewarded with a bigger numerator for forcing exactly the outcome it wanted.
8. 8The second ball is headed clear near the halfway lineIt happens in the pressing team's own two-fifths, outside the boundary. Real defending, no denominator credit.
9. 9The opponent recovers and rebuildsThree more passes between the centre-backs and the holding midfielder, all inside the counted zone. Numerator: 7.
10. 10A shirt is grabbed and the referee blowsA foul is a defensive action. Denominator: 2. As far as the arithmetic is concerned, a cynical trip and a clean interception are identical events.
11. 11The spell contributes 7 and 2Not a PPDA of 3.5. The two counts go into the running totals for the match and the season, and the division happens once, at the end, over everything.

An entirely constructed passage of play, used to show which events are counted and which are not. No real match, team or figure is described. The pitch runs 0 to 100 from the pressing team's own goal line, and only events beyond the 40 line count.

Three things in that sequence are worth sitting with. The closing-down run that shaped the whole passage produced no data at all. The long goal kick, the clearest evidence in the entire spell that the press was working, made the numerator worse. And a foul improved the figure.

None of that is a bug in an implementation. It is what happens when you approximate an off-the-ball phenomenon with an on-the-ball event feed.

What PPDA is genuinely good at, which is more than its critics allow

The criticisms below are real, and it would be easy to finish this article thinking the metric is worthless. It is not, and the reason is sample size.

A match contains a few dozen shots between both sides, which is why single-match expected goals figures bounce around so violently. A match contains many hundreds of passes. Pressing measures built on passes therefore stabilise far faster than anything built on shots, and a season's PPDA for a team is a genuinely well-populated number rather than a rumour. It is one of the few public football statistics where a season total is not mostly noise.

Within those limits it does three jobs well.

It orders the extremes correctly. A side that sends a forward to harry the goalkeeper at every goal kick and a side that begins defending on the edge of its own box will separate cleanly and repeatedly. You do not need a metric to tell those two apart, but you do need one to rank the fourteen teams in between without watching six hundred hours of football.

It tracks one team against itself over time. This is the highest-value use and the one that avoids most of the failure modes, because a team compared with its own past shares a provider, a definition, a league and largely a squad. A step change in a club's figure after a managerial change is close to unambiguous evidence that the defensive brief has been rewritten. A slow drift upwards across a season is worth investigating as fatigue, injuries to the specific players who lead the press, or a squad quietly deciding it has had enough. The mechanics of what has actually changed live in the running, and the article on what a counter-press is actually betting on covers the phase that these numbers are trying to detect.

It is a screening tool for a whole competition at once. Ten seconds of sorting a column tells an analyst which four teams to watch first. That is not a small thing when the alternative is video review of everybody.

Effective pressing suppresses its own numerator

Here is the failure that matters most, because it is not a limitation at the edges. It sits at the centre of the metric and it points in the opposite direction to intuition.

A press that works forces the opponent to stop passing. They clear it, they hit the channel, the goalkeeper goes long from the six-yard box rather than rolling it out. Clearances are not passes in most event taxonomies, so the numerator stops growing. The ball also leaves the counted zone, so the pressing team gets no opportunity to make a defensive action there, and the denominator stops growing too.

Both halves of the fraction shrink together, and where the ratio ends up depends on which shrinks faster. Success is not guaranteed to move the number in the direction you want.

• 60Spell one, opposition passes in the zone
• 10Spell one, defensive actions in the zone
• 30Spell two, opposition passes in the zone
• 5Spell two, defensive actions in the zone

Two entirely invented twenty-minute spells against the same invented opponent. In the first the press is played through repeatedly; in the second it forces the ball long. Both come to a PPDA of 6.0. No real team or match is described.

In the invented spell one, the opponent passed freely and the pressing team scrambled after the ball, missing tackles and conceding fouls, and both counts are high. In the invented spell two, the opponent gave up trying to play out after ten minutes and launched everything, so there was much less passing to interrupt and much less to challenge. Sixty divided by ten is 6.0. Thirty divided by five is 6.0. The same figure describes a press that was being dismantled and a press that had won the argument entirely.

Now push it further. Imagine the press is so effective that the opponent abandons building from the back altogether, and every goal kick is hit into the opposition half from the start. The numerator collapses towards a handful of passes, the denominator collapses with it, and the ratio becomes a small number divided by a smaller one, which is the definition of an unstable statistic. The most successful press imaginable produces the least reliable figure.

The mirror case is worse, and it is the one that should make anyone quoting the number pause. Failed tackles count. Fouls count. A team pressing chaotically, arriving late, lunging and missing, getting played through and hauling somebody down as the move goes past, is manufacturing denominator at a furious rate. That team's PPDA looks superb. On video it looks like a team being taken apart.

Intensity and effectiveness are different quantities, and PPDA measures the first while people quote it as though it measured the second.

The scoreline is doing more of the work than the tactics are

A team two goals up with twenty-five minutes left stops pressing. Not because it has become worse, but because the task has changed: the value of winning the ball high has fallen and the cost of being caught out of shape has risen. Its forwards stop making the trigger runs, the block drops ten metres, and the defensive actions migrate out of the counted zone. Meanwhile the opponent, chasing, has far more of the ball and passes it far more in their own third while looking for a way forward.

Numerator up, denominator down. The figure rises sharply and none of it is a tactical statement about how the team plays.

While trailing5.3

While level6.4

While leading by one8.1

While leading two or more11.2

Whole invented season7.2

An entirely invented season for an invented team, constructed to show the shape of the game-state effect rather than its size. The whole-season figure is the ratio of the summed counts, not the average of the four bars. No real team, season or figure is described.

Show the numbers

Worked example: one invented team's PPDA split by game state

Item	Value
While trailing	5.3
While level	6.4
While leading by one	8.1
While leading two or more	11.2
Whole invented season	7.2

The consequence for season figures is that they are partly a record of how often a team has been ahead. A side that leads in most of its matches accumulates a large slice of its minutes in the state where it presses least, and its season PPDA rises accordingly. A side that spends the season chasing games presses out of necessity in the final twenty minutes of everything, and its figure falls. Two teams with identical pressing instructions and different league positions will not produce the same number.

Any serious use of the metric splits it by game state before doing anything else. The most informative single cut is the figure while the match is level, because that is the state in which a team is doing what it actually wants to do rather than what the scoreboard obliges it to do. Broadcast graphics never make this split. Analysts always do.

Possession dependence, and why dominant teams look like pressing teams

PPDA is a rate rather than a total, which normalises away how much the opponent has the ball. It does not normalise away where they have it, and that is a different problem entirely.

Consider a team that dominates possession heavily. Its opponents rarely have the ball, and when they do get it, it is usually deep in their own third, which sits well inside the counted zone. Every time the dominant team's own attack breaks down, its players are already high up the pitch and can pile straight back onto the ball, generating defensive actions in the zone almost for free. So the numerator is held down by the opponent barely having possession at all, while the denominator is inflated by the sheer number of turnovers happening in the opposition third. The team posts a low figure. Providers have flagged exactly this confound, and it is why territorial dominance and pressing intensity are so hard to separate with this number: sides that monopolise the ball can rank as heavy pressers without pressing especially hard.

Run the logic the other way and the same confound appears with the sign flipped. A side that never has the ball is exposed to enormous quantities of opposition passing, and a fair share of those passes happen inside the counted zone, which is more raw opportunity to make a defensive action than any dominant team ever gets. Both halves of the fraction grow. Whether the ratio rises or falls depends entirely on whether the side chooses to engage up there or drop off, so a low-possession team's figure carries more genuine information about intent than a high-possession team's does.

The opponent's style is a further contaminant that is nobody's fault. Play against a side that builds patiently from the goalkeeper and your numerator fills with the twenty passes they string together in their own third every time they restart. Play against a side that goes long from every goal kick and there is barely a numerator at all. Your own behaviour has not changed between those two matches, and your figure will. This is why comparing a team's PPDA across a small number of matches is close to meaningless without knowing who the opponents were, and why the most sophisticated versions in club use adjust for the opposition faced rather than taking the raw ratio at face value.

The general shape of the problem is familiar from other proxy statistics. A metric that describes a team also describes everyone it has played, in the same way that a possession percentage says as much about the opponent as about the team holding the ball.

Pressing high and pressing hard are not the same thing

The counted zone is fixed. Anything inside it counts equally, whether it happened on the edge of the opponent's six-yard box or a stride into the opposition half. PPDA therefore measures how often you engage inside a wide band of pitch, and says nothing about where in that band the engaging happens.

Where the side engages	How hard it engages	What PPDA reports
In the opponent's third	Hard	A low figure. The one case the metric reads correctly.
Around the halfway line	Hard	A low figure too, barely distinguishable from the row above.
In the opponent's third	Passively, holding a high line without closing down	A high figure, correctly.
In its own third	Hard, an aggressive low block	A high figure, because almost nothing it does is counted at all.

Rows one and two are the confusion. A team pressing ferociously from the halfway line and a team pressing ferociously in the opposition box are doing different things, want different outcomes, and expose themselves to different risks. The metric puts them in the same place.

Row four is the blind spot. A team defending deep and aggressively, winning the ball with violent challenges thirty metres from its own goal, generates almost no counted events and reads as passive. The way a compact deep block is actually organised is invisible to this number by construction, which is fair enough given what the number was built for, provided nobody claims it measures defensive aggression in general.

The fix is not a better ratio. It is a second number, and the standard one is defensive action height: the average distance from your own goal at which your defensive actions occur. Height and intensity together are far more informative than either alone, and plotting a league on those two axes is one of the genuinely useful things you can do with public data.

What PPDA says about pressing structure: nothing at all

A press is an organised act. Who goes first, which passing lane is cut as he goes, who takes the near option and who takes the far one, which touch or body shape or backwards pass sets the whole thing in motion, how many players stay behind so that being played through is survivable. That structure is what distinguishes a coached pressing side from eleven players chasing.

None of it appears in this metric. The numerator and the denominator are agnostic about which player made the action, whether he was supported, whether the press was collective or a lone forward's private initiative, and whether the moment was a good one to press at all.

The last point is the sharpest. Pressing at the wrong moment is a serious error, and it produces a defensive action just as pressing at the right moment does. A midfielder who charges at a centre-back who has time, space and three options open, breaking his line's shape to do it, contributes exactly the same one to the denominator as a forward who arrives at the precise instant a bad first touch makes the ball unprotectable. The metric cannot see the difference. The work on which moments are worth attacking is set out in the piece on the cues that tell a side when to go, and the value of the resulting turnover depends heavily on the shape left behind, which is the subject of what a team does in the seconds after winning the ball back.

Nor does PPDA distinguish a settled press against an opponent's goal kick from a counter-press against an opponent who has just intercepted. Those are different phases with different geometry and different odds. Both leave identical fingerprints in the event feed.

The measures that fix parts of the problem

No single number replaces PPDA. A short stack of them, each aimed at one of the failures above, gets much closer.

Defensive action height answers the where. Take the average position of a team's defensive actions along the length of the pitch and you have a direct measure of the line at which it chooses to defend, uncontaminated by how hard it presses once there. Tracking data extends this to the height of the defensive line itself, which is a separate and equally useful quantity.

Zone-split PPDA answers a milder version of the same question by computing the ratio separately for the opponent's defensive third and the middle third, rather than smearing one number across the whole band.

Pressure events answer the missing off-ball problem. Some providers record an event whenever a defender closes to within a short radius of an opponent receiving the ball, whether or not any contact follows. That turns the invisible closing-down run from the flow above into recorded data, and counting pressures rather than only tackles and fouls removes one of the metric's largest blind spots.

Opponent pass completion under pressure answers the effectiveness question. If your press is working, the opponent's completion rate should fall below their own baseline. Comparing what a team's opponents actually completed against what they would be expected to complete given the pressure applied separates a press that disrupts from one that merely engages. Providers publish variants of this alongside PPDA precisely because PPDA cannot do it.

Directness and width of the opponent's play answer the same question from another angle. A press that is working pushes the opponent's actions towards the touchlines and forces them to move the ball forward faster than they would like. Both are measurable, and both rise when the ball is being hit long under duress, which is the outcome PPDA registers as a larger numerator.

Packing-style measures answer the value question. Counting how many opponents a pass or a carry takes out of the game gives you a number for how comprehensively a press is being played through: being repeatedly bypassed by a single pass that removes five players is a specific, diagnosable failure. The construction of the metric that counts opponents taken out of the game is the attacking mirror of everything described here, and reading the two together tells you whether the engagements are achieving anything.

Tracking-derived pressing intensity answers all of it, at a price. With every player's position ten or twenty-five times a second you can compute how long it would take each defender to reach the ball, how much space the man in possession genuinely has, how many players moved towards the ball at the moment it was played, and the accelerations and decelerations that make up the physical cost of the whole exercise. These are the measures clubs actually use internally, and what optical and GPS tracking data makes visible is the reason the public conversation about pressing is a decade behind the private one. The data is expensive, it is rarely published, and the models built on it are proprietary, which is precisely why a free ratio from an event feed remains the most quoted pressing number across the professional game.

How analysts actually use PPDA

Nobody serious treats it as a verdict. It is used as a filter, and the workflow is dull in a way that is worth describing because it is the opposite of how the number appears in public.

The figure is computed per team per season from summed counts, split by game state, and where possible adjusted for the opponents faced. It is then plotted against defensive action height, so that pressing hard and pressing high are separated. That plot identifies teams that sit somewhere surprising: a side with a low figure and a low height is pressing hard from a mid-block, which is a different plan from the low-figure, high-height sides around it.

Then somebody watches the video. That is the part the metric exists to make efficient, not the part it replaces. The number identifies which twenty minutes of which four matches are worth examining, and the examination establishes what is actually happening: whether the press has triggers, whether the second line follows the first, whether the fouls are cynical stops or mistimed lunges, whether the opponent has found the free man every time and the figure is flattering a defensive failure.

Two other uses come up regularly. In opposition scouting, a coarse read on how much pressure to expect shapes whether a side plans to play through it or over it, and the figure is a reasonable starting hypothesis to be confirmed on video. And in recruitment, the figure describes a team rather than a player, so it can never be used to judge an individual's pressing. What it can do is characterise the system a player is arriving from or going to, which matters when assessing whether a forward's defensive numbers are his own work or his manager's instructions.

The honest summary of professional practice is that PPDA is used to decide where to look, never to decide what is true. It is at its most useful in exactly the situation where nobody bothers to quote it, which is a single club tracked patiently across two or three seasons.

Reading a PPDA figure honestly

When one appears in front of you, six questions restore the context it was stripped of on the way to the graphic.

Whose definition is it? The pitch boundary and the list of defensive actions both vary between providers, and figures from two sources are not comparable in either direction.

Over how many matches? A single match's figure is contaminated by one opponent's style and one match's scoreline. A season's is a genuine number.

Against whom? Ten fixtures against patient build-up sides and ten against long-ball sides produce different figures from identical pressing.

At what game state? If the split is not available, ask how often the team has been leading. Successful teams post higher figures for reasons that have nothing to do with their pressing.

Where were the actions, not just how many? Without defensive action height alongside it, the figure cannot distinguish a press that starts at the opposition goalkeeper from one that starts at the halfway line.

Did anyone watch it? The metric counts engagements. It has no opinion on whether they worked, whether they were coordinated, or whether they were a good idea, and a low figure produced by a team being played through while flailing looks identical to one produced by a team suffocating its opponent.

None of that makes the number worthless. It makes it a proxy that knows it is a proxy, which is a better position than most football statistics occupy. Used as a shortlist it is excellent. Used as a ranking it is arguable. Used as a conclusion it is a way of being confidently wrong about a team that never once had the ball where the metric was looking.

The number tells you how often a team interfered with its opponent in the part of the pitch where interfering is expensive. That is a real fact about a season, and it is worth having. What happened when they interfered is a separate question, and it is still answered by watching.