고강도 프레스의 엔트로피 효과: ‘압박 상황 하에서의 패스 성공률’을 통한 경기 후반 단계 확률 붕괴 해석

In the cognitive models of most football analysts, “dominance” is quantified as possession rate, shot count and xG value. This logic assumes that a match is a static cumulative probability process, holding that with superior statistics, scoring goals is merely a matter of time. However, this simplistic linear thinking completely ignores the physiological limits and tactical fragility of football as a high-intensity dynamic contact sport. A team that controls the game through high-pressing in the first 60 minutes achieves its apparent “dominance” at the cost of accelerated depletion of systemic energy.

When we examine cases of teams widely regarded as “strong sides” that suddenly concede goals or get overturned late in matches, a neglected signal emerges: the non-linear decline in Pass Success under Pressure (PSuP).

Conventional predictive logic tends to argue that if a team can pin opponents deep in their own half through high pressing in the first half, its win probability will rise over time. Yet quantitative analysis reveals the opposite phenomenon: when a team maintains an extremely low PPDA (Passes Allowed per Defensive Action, meaning high-intensity pressing) for an extended period, a distinct “efficiency breakdown” occurs in the passing connection between its backline midfielders and centre-backs between the 65th and 80th minutes.

|Analysis Dimension|Mainstream Predictive Logic|Quantitative Heterodox Logic|
| —- | —- | —- |
|Core Metrics|xG / Possession Rate / Recent Win Rate|PSuP ΔΔ (Change in Pass Success under Pressure)|
|State Perception|Good form → Higher chance of winning|High-intensity pressing → System entropy growth → Probability collapse|
|View on Conceding Goals|Accidental errors / Luck factors|Tactical structural disintegration caused by physical energy consumption|
|Prediction Focus|Linear extrapolation of outcomes|Expected value regression driven by marginal diminishing returns|

The underlying principle of this phenomenon lies in “physiology-tactics coupling”. High pressing requires players to perform frequent off-the-ball sprints within short time windows to block opponents’ passing lanes. As the match enters the latter part of the second half, players’ anaerobic endurance hits a critical threshold, creating minor gaps in the previously compact pressing shape. For a side reliant on a pressing system, the deadliest threat is not the opponent’s enhanced attacking play, but its own sharp drop in Pass Success under Pressure.

Specifically, when defensive midfielders face counter-pressing from opponents and their pass success rate falls from 85% at kick-off to below 60%, the team is highly prone to losing possession in the opposition’s midfield and forward areas during transitions from attack to defence. At this point, since the team has pushed its defensive line forward to sustain pressing, any loss of ball in dangerous areas leaves its backline directly exposed to the opponent’s fast Transition. This defensive collapse triggered by declining PSuP manifests statistically as a “drastic shift in probability distribution”: the chance of conceding remains very low for most of the match, but rises exponentially during specific windows of physical fatigue.

To translate this micro-variable into a predictive quantitative model, we need to revise traditional expected value calculations. A valid model should not focus on a team’s overall xG, but on the difference (ΔPSuP) between PSuPt1 (first 30 minutes) and PSuPt2 (final 30 minutes).

If ΔPSuP exceeds a certain threshold (e.g. 20%), and the team maintains a high Defensive Line Height (DLH), its probability of conceding in the final 15 minutes of the match will be far higher than the distribution implied by market odds. In such scenarios, the so-called “dominance of strong teams” actually becomes their biggest weakness.

In practical application, we can make predictions by observing micro details from the opening 30 minutes of a match. For example, if a key defensive midfielder commits more than two basic passing errors under pressure from opposition forwards, and his running coverage drops markedly after the 60th minute, this serves as an early warning sign of rising system entropy. At this stage, regardless of whether the score stands at 1-0 or 2-0, if the opponent adopts a low block and waits for counterattacks, the expected value of a “plot twist” in the match will rise substantially.

This analytical approach abandons vague metaphysical discussions about “form”, and breaks matches down into this chain: energy depletion → falling passing efficiency → structural vulnerabilities → probability collapse. It exposes a harsh truth in football prediction: more often than not, the more “perfectly” a team executes high-intensity pressing on the pitch, the higher its mathematical probability of collapsing late in the game.

Therefore, genuine quantitative prediction is not about identifying the stronger side, but spotting when the fatigue accumulation and structural fragility within a team’s system reach a critical threshold while it maintains its “strong” performance. When we capture the marginal decline of PSuP via real-time data, what we observe is no longer random match fluctuation, but an inevitable mathematical cycle of regression.