The Mid-Tournament Klement Scorecard: How Right Is the Model So Far?
After one week of group-stage football, Klement's WC 2026 model has 9 verifiable calls in. We tally the hits, the misses, and the three stories still in play.
The Mid-Tournament Klement Scorecard: How Right Is the Model So Far?
One week into the 2026 World Cup, Joachim Klement's model has nine verifiable predictions on the board. Here's the running scorecard — what he got right, what he got wrong, and the three calls that still decide whether his perfect winner streak survives to 4-for-4.
- Verifiable group-stage calls so far
- 9
- Cleanly right
- 5
- Cleanly wrong
- 2
- Mixed (right thesis, wrong detail)
- 2
- Net model accuracy (weighted)
- ~67%
- Klement's claimed model accuracy
- ~55%
The Scorecard
Each row below maps a Klement prediction onto an observed result from the first nine days of group-stage football. Verdict definitions: right = prediction and result aligned; wrong = prediction contradicted; mixed = thesis intact but detail off; open= call still active, evidence pointing in Klement's direction.
| Match / Call | Klement said | Actual | Verdict |
|---|---|---|---|
| 🇪🇸 Spain vs 🇨🇻 Cape Verde | Spain 99% top-2 | 0–0 draw | open |
| 🇧🇷 Brazil vs 🇲🇦 Morocco | Brazil group winner | 1–1 (Brazil still in) | open |
| 🇳🇱 NL vs 🇯🇵 Japan | NL group winner, JP second | 2–2 (both advancing) | open |
| 🇺🇸 USA vs 🇦🇺 Australia | USA best-third (3rd) | USA top of group | wrong |
| 🇹🇷 Türkiye Group D run | Türkiye top-2 (59%) | Eliminated after 2 matches | mixed |
| 🇩🇪 Germany vs 🇨🇼 Curaçao | Germany routine win (93%) | 7–1 Germany | right |
| 🇸🇪 Sweden vs 🇹🇳 Tunisia | Sweden best-third route | 5–1 Sweden | right |
| 🇵🇹 Portugal vs 🇨🇩 DR Congo | Portugal Group K winner (77%) | 1–1 draw (Ronaldo started) | mixed |
| 🇨🇮 Ivory Coast vs 🇪🇨 Ecuador | Ecuador #2, Ivory Coast best-third | Ivory Coast 1–0 (Diallo 90') | mixed |
Pattern 1 — The Model Is Right About Tier Boundaries
Klement's framework distinguishes four tiers: elite, strong, mid, weak. When matchups cross a tier boundary (Germany vs Curaçao, Sweden vs Tunisia, Spain vs Cape Verde at the extreme), the model is consistently right about who wins. The Spain vs Cape Verde draw is an outlier on a single result, not on the group-stage trajectory.
Pattern 2 — The Model Is Noisy About Same-Tier Ordering
Where Klement is consistently off: ranking similar mid-tier teams against each other. Ecuador vs Ivory Coast in Group E. Australia vs Türkiye in Group D. These are coin-flips dressed up as deterministic calls. The bracket survives them — both teams typically advance somewhere — but the implied confidence is overstated.
Pattern 3 — Host Advantage Is Under-Priced
The biggest single miss is the “host advantage diluted” assumption. USA on top of Group D, Mexico leading Group A, Canada through from Group B. All three co-hosts are outperforming Klement's structural projection. The model treats co-hosting as ⅓× the normal host effect; the evidence after one week suggests it's closer to ⅔× or even full.
The Three Calls That Still Matter
The model's 4-for-4 winner streak depends on three downstream predictions, none of which the opening week has resolved:
- Japan beats Brazil in the R32. The headline upset. Both teams are positioned to make the R32. The 2–2 NL-Japan result is structurally encouraging for Klement — Japan can play with top-7 sides. Read the full analysis →
- Netherlands win the trophy. The Group F draw narrows the path. A win against Sweden is now mandatory to lock first place. Read the winner case →
- Portugal reach the final.Klement's most contrarian call survives a 1–1 with DR Congo, but only if Martínez benches Ronaldo as the tournament progresses. Read the Portugal analysis →
The Honest Confidence Update
Klement himself frames the model as ~55% predictive, ~45% luck. After one week, the running accuracy is approximately 67% — better than the long-run rate, but a small sample. Two clean wrongs (USA top of Group D, Türkiye out) and two mixed verdicts are the kinds of errors a ~55%-accurate model is supposed to make. The fact that none of the three downstream winner-path calls has been broken is the more important signal.