Skip to content
Fri 26 Jun · 02:00 UTC · 04:00 CEST · Group D

full time

Turkey3–2United States

1953+69rating1834+722ndfifa17thdown 1
A. Güler1944talismanM. Tillman1922

SoFi Stadium, Los Angeles · 70,240 seats

Missed it. It stays up.

United States to win

home 33% ✓draw 26%away 42%

sealed Thu 25 Jun · 08:57 UTC · 10:57 CEST · nothing deleted

By The DeskScout sourced the form → Analyst reasoned the call → Gates cleared → Grader settled it

locked Thu, 25 Jun 2026 08:57:48 GMT · model elo-davidson-features-v2

share the cardstorysquarefeedwide cardoutlook slide

The read

No prose on this one. The desk let the number do the talking. The call stands, and it goes on the record either way.

The ratings

The two strength reads behind this fixture: our results-based rating, the same forward-only engine the sealed call uses, read monthly, and FIFA's official table. Up is stronger on both lines; green is climbing, red is sliding. Context beside the call; the sealed number does not move.

Turkey

ProofXI rating

1953+69 this year

1884 → 1953 · 13 readings

FIFA world rank

22ndheld on the latest list

25th → 22nd · 6 lists

USA

ProofXI rating

1834+7 this year

1827 → 1834 · 13 readings

FIFA world rank

17thdown 1 place on the latest list

14th → 17th · 6 lists

the gap on our board today: Turkey by 119 points

Head to head

Turkey have had the better of this fixture.

6 prior meetings · Turkey 3, drawn 1, United States 2 · goals 10-9

  1. 2026United States 2-3 Turkey· FIFA World Cup
  2. 2025United States 1-2 Turkey· Friendly
  3. 2014United States 2-1 Turkey· Friendly
  4. 2010United States 2-1 Turkey· Friendly
  5. 2003Turkey 2-1 United States· Confederations Cup· neutral

source · international results feed

How we got hereWe make United States the stronger side, and seal the call at about 4 in 10. Open for the working.openclose

The arithmetic behind the call: a rating edge, in points, mapped to a probability.

Turkey rating edge-43
Home advantage0 (neutral venue)
Net edge-43

Davidson Elo→1X2 turns that -43 edge into 33% / 26% / 42% for home, draw, away.

Ratings only. Recent form and rest are not modelled for national sides.

How it could go

The same sealed edge, played as goals instead of straight to a result: our goals model runs the match score by score. The call above stands, this is its texture.

Turkey

1.2

expected goals

United States

1.4

expected goals

The scorelines it sees most

  • 1–113 in 100
  • 0–19 in 100
  • 1–29 in 100
the fuller goal pictureopenclose
  • 1–08 in 100

three or more goals: 48 in 100 · clean sheet: Turkey 25 in 100, United States 30 in 100

glassbox-goals-v1, our own goals model, run from the same ratings as the call. Frequencies, never a market.

The scorecard

  • United States win on the road.✗ wrong
  • No blowout: United States win by a single goal at most, if at all.✓ right
  • An open game in which both teams find the net.✓ right

The team sheets

The confirmed elevens, as named. The number beside each name is his rating on our player board, a sourced shadow signal. The sealed call above was made days before these sheets existed, and it does not move.

Turkey4-2-3-1

  1. 23U. ÇakırG1610
  2. 2Z. ÇelikD1619
  3. 15O. KabakD1763
  4. 14A. BardakcıD1605
  5. 13E. ElmalıD1587
  6. 6O. KökçüM1550
  7. 5S. ÖzcanM1846
  8. 24O. AydınM1584
  9. 8A. GülerM1944
  10. 11K. YıldızM1518
  11. 21B. YılmazF1722

11 of 11 starters carry folded match data on our board; the rest sit at the seed prior. We rate what we've seen.

The bench

USA4-1-2-3

  1. 1M. TurnerG1581
  2. 23J. ScallyD1733
  3. 22M. McKenzieD1732
  4. 12M. RobinsonD1606
  5. 6A. TrustyD1677
  6. 8W. McKennieM1837
  7. 14S. BerhalterM1685
  8. 7G. ReynaM1690
  9. 21T. WeahM1754
  10. 9R. PepiF1817
  11. 11B. AaronsonM1861

11 of 11 starters carry folded match data on our board; the rest sit at the seed prior. We rate what we've seen.

The bench

source: api-football.com · fetched Fri, 26 Jun 2026 01:31:11 GMT

The timeline

The goals, assists and substitutions as they came, read straight off the match feed, by the clock. Sourced facts; the call above was sealed days before any of them.

  1. 3'A. Trustygoalassist S. Berhalter
  2. 10'A. Gulergoalassist B. Yilmaz
  3. 31'B. Yilmazgoalassist O. Kokcu
  4. 49'S. Berhaltergoal
  5. 58'C. Pulisic for T. Weahsub
  6. 76'S. Dest for G. Reynasub
  7. 77'A. Zendejas for B. Aaronsonsub
  8. 77'A. Freeman for J. Scallysub
  9. 84'C. Soyuncu for Z. Celiksub
  10. 84'C. Uzun for K. Yildizsub
  11. 86'M. Tillman for W. McKenniesub
  12. 88'K. Ayhan for O. Kokcusub
  13. 90+2'M. Muldur for O. Aydinsub
  14. 90+2'I. Kahveci for B. Yilmazsub
  15. 90+8'K. Ayhangoal

source: api-football.com · fetched Fri, 26 Jun 2026 06:50:09 GMT

Full time

Missed it.

Turkey 3-2 United States

Verdict
wrong but well reasoned
Claims right
2/3

The debrief

We sealed United States to win at 42%, about 4 in 10, before kickoff. Full time said Turkey 3-2 United States: the Turkey win, an outcome we had at 33 in 100.

The call missed, but the reasoning mostly held. That is what a probability is, the chance we sealed for what happened is exactly how often this should hurt. It stays up. The claims went 2 of 3.

What we got wrong, by name

  • United States win on the road.

One match never grades the model, calibration is judged over the whole record, and this one is now in it.

How it moved our numbersThis result nudged our live ratings for both sides. Open for the signed moves.openclose

On the live national Elo, this result moved Turkey +27 and United States -27.

On our player board, the squads moved Turkey +115 (16 appearances) · United States -87 (16 appearances).

Our own ratings, computed from the result, a shadow signal beside the call, never one of its inputs. The sealed forecast above does not move.

THE DESK AT WORK

  1. ScoutSourced the inputsPulled both sides' ratings.
  2. AnalystWeighed the edgeNet edge -43 after home advantage; drafted 3 checkable claims.
  3. ModelMapped it to a distributionDavidson Elo→1X2 → 33/26/42 (home/draw/away).Favourite: United States · 42%
  4. ScribeFiled the readPrepared the read behind the call.
  5. GatekeeperCleared the gatesNo betting markets · Sourced facts only · No fabricated names · Falsifiable claims, all clear.
  6. The DeskLocked the callLocked Turkey v United States before kickoff.

A replay of how this call was built. Real steps, paced to read. The model owns the favourite; the desk sources and explains it.

Behind the glass

The pipeline

  1. Researchinputs sourced
  2. Forecastlocked before kickoff
  3. Drafthouse template
  4. Gatesall checks passed
  5. Publishshadow
  6. Gradescored vs the result

The checks

  • No betting markets

    The call is built from football strength alone, never a bookmaker's price, as an input or anywhere on the page.

  • Sourced facts only

    Every player, squad or result fact must trace to a dated source under seven days old, or it is blocked before publish.

  • No fabricated names

    A registry blocks invented or mistaken player and team names from ever reaching the page.

  • Falsifiable claims

    Each claim on the scorecard must be checkable against the final result, no vague hedging that can't be graded.