Sports

USMNT World Cup xG review says old stories need a rethink

ESPN’s Bill Connelly used Stats Perform data to test USMNT World Cup memories against expected goals from 1990 onward.

Georgia Hale

By Georgia Hale · Staff Writer

3 min read

USMNT World Cup xG review says old stories need a rethink
Photo: ESPN.com

The USMNT World Cup xG record tells a messier story than the usual boom-and-bust memories, according to an ESPN analysis by Bill Connelly using Stats Perform data.

Connelly framed the review around a familiar American soccer mood swing: the United States men’s national team produced several encouraging performances at the 2026 World Cup, then went out with a rough loss to Belgium. He wrote that the strong spell lifted the national mood and drew tens of millions of TV viewers, while the ending revived old doubts about how far the program has really come.

Stats Perform, according to ESPN, has event data and advanced statistics for World Cup matches dating to 1966. Connelly used that database to revisit U.S. games from the modern era, starting with the country’s return to the tournament in 1990 after a 40-year absence.

What does xG say about USMNT World Cup games?

Expected goals, or xG, is a measure of the quality of scoring chances rather than the final score. In plain terms, it asks whether a team created enough good looks to deserve the goals it got.

By that lens, ESPN’s review says some famous U.S. results look shakier, while some painful exits look less bleak. The 1994 run on home soil, long remembered as a breakthrough, came with a warning label: opponents generated 7.0 xG across the tournament to the Americans’ 2.6, according to the data cited by Connelly.

The 1994 win over Colombia is one example. The U.S. won 2-1, helped by Andres Escobar’s own goal, but ESPN’s figures had Colombia ahead 1.98 to 0.90 in total xG. In the round of 16, Brazil beat the U.S. only 1-0, yet the xG gap was far wider: Brazil 3.50, USA 0.13.

The 1998 World Cup, remembered as a crash, gets a softer statistical reading. The U.S. was outscored 5-1 in France, but ESPN reported that opponents led only 4.5 to 3.3 in xG. The loss to Iran, which eliminated the Americans, was nearly level by the numbers: Iran 1.90, USA 1.67.

Which famous results look different now?

The wild 2002 quarterfinal run gets the full chaos treatment. Connelly wrote that xG suggests the U.S. was fortunate to reach the quarterfinals, then unfortunate not to go farther.

  • Against Portugal in 2002, the U.S. won 3-2 and also edged the xG count, 1.73 to 1.45, according to ESPN.
  • Against South Korea, Brad Friedel saved nine of 10 shots on target, including a penalty, as the U.S. drew 1-1 despite losing the xG battle 3.10 to 0.38.
  • Against Mexico, the U.S. won 2-0 and led xG 1.98 to 1.09, though ESPN noted officials missed a handball in the U.S. penalty area.
  • Against Germany in the quarterfinal, the U.S. lost 1-0 despite leading xG 2.00 to 1.66.

The 2006 tournament, by contrast, comes off as a real missed chance in ESPN’s breakdown. The U.S. entered with high expectations and had been ranked as high as fourth by FIFA earlier that year, but Connelly wrote that the statistical performance was even worse than 1998.

ESPN’s numbers show the U.S. created just 0.42 xG in a 3-0 loss to the Czech Republic, then played a more balanced 1-1 draw with Italy, where the xG was 1.35 to 1.01 for Italy. Ghana beat the U.S. 2-1 in the group finale and led xG 2.12 to 1.33.

Connelly’s broader point is that the scoreboard has not always matched the chance quality. For the USMNT, the World Cup record is still full of heartbreak, lucky breaks and near-misses, but the advanced numbers make the old story harder to tell in a straight line.

This story draws on original reporting from ESPN.com.