Benhida, El Morchidy, Zeghari, Enneya & Guerss (2025) · Applied Sciences 15(18), 9994. doi: 10.3390/app15189994
The first African nation to reach a World Cup semifinal. What did the data actually show? An AI-assisted study of all 7 Morocco matches at WC22 finds the same story we found via off-ball attention chemistry — from a completely different angle.
Seven matches: Croatia, Belgium, Canada, Spain, Portugal, France, Croatia (3rd place). Inputs: official FIFA post-match indicators (possession, passes, distance, sprints, pressing). Three steps:
No model, no event data, no tracking. Just public post-match indicators run through standard multivariate methods. The conclusion lines up with ours anyway.
Compact mid-block, 4-1-4-1 with Amrabat dropping. The work to not have the ball was harder than anyone else's work to keep it.
Compact organization + coordinated bursts of pressure. Defensive solidity wasn't passive — it was a metered, opponent-specific decision about when to compress.
Axis 1 (52%) — physical intensity vs. technical mastery.
When Morocco was running, they weren't passing.
Axis 2 (24%) — verticality vs. build-up.
When Morocco was attacking, they went forward fast.
The second axis is the one that maps to our finding. Morocco's score-frame attention pairs Ziyech ↔ Hamdallah (2.02×), Ziyech ↔ Aboukhlal (2.01×), Ziyech ↔ Ounahi (1.74×), En-Nesyri ↔ Ounahi (1.72×) are exactly the "verticality" axis — the attacking transitions the model attends to when scoring is imminent.
| Profile | Matches | What it looked like |
|---|---|---|
| Low-block defensive | Spain, Croatia (1) | 23–35% possession, max distance, max compactness |
| Transition-oriented | Belgium, Portugal | 27–33% possession, fast vertical recoveries — En-Nesyri's winner |
| Open / shared | France, Croatia (2) | 45–49% possession, more proactive, more defensive exposure |
Adaptability, not a fixed style. Morocco played a different game for each kind of opponent — and still kept the same defensive shape underneath.
The passing networks (Figure 8) show right-flank dominance in every critical match: Hakimi and Ziyech as the primary connection, Amrabat as the central pivot dropping in for cover.
Amrabat — >10 km every match, consistent.
Hakimi — explosive sprint profile from RB.
Ziyech — creator with variable defensive contribution.
Ounahi — dynamic midfielder, balanced output.
In our score-frame attention leaderboard, Ziyech is in five of the top 20 off-off pairs. He is the connective tissue between defensive recovery and attacking transition — exactly what the paper's passing-network diagrams show.
"Fast transition football combined with technical skill, strong defensive resilience, and rational tactical management... a synthesis of collective intelligence built on transitions, defensive solidity, and contextual adaptability."
— Benhida et al. (§4 + §5)
In other words: they didn't dominate games, they shaped them. The opponent decided how much ball Morocco would have. Morocco decided what to do with it.
| Method | What it surfaced |
|---|---|
| Benhida et al. — PCA on FIFA reports | Verticality + transition axis, Hakimi–Ziyech right flank, En-Nesyri's vertical winner |
| Our work — score-frame attention on tracking | Ziyech + Hamdallah / Aboukhlal / Ounahi / En-Nesyri dominate the top-20 off-off pairs at 1.7–2.0× baseline |
The two studies cannot agree by accident — they have no overlapping inputs. Benhida used aggregated FIFA stats; we used per-frame tracking attention. The convergence is the validation.
If two methods that share no data both point at the same attacking partnerships, the partnership is real.
Back to the site · Read the paper: 10.3390/app15189994