World Cup Semifinals: Spain vs France, Argentina vs England

The World Cup is down to four, and today's two semifinals sell themselves. Spain play France in one, Argentina play England in the other, and the winners walk into the final. What sharpens it is the title race behind it. Argentina at 31.7 percent, France at 29.0, and Spain at 28.9 sit inside a single point of one another, with England the outsider at 10.5.

Spain and France is a coin flip. Our four model reads split almost down the middle. Two lean France by the thinnest margin, at 52 and 55 percent, and the other two push to 62. France is the side on balance, but this is the closest match left in the tournament, the kind where the projection and a shrug are hard to tell apart. Spain have been the most complete team by the underlying numbers all summer, and the model has them a hair back mostly on France's edge in the moments that decide a knockout.

Argentina and England has more daylight. Across the four reads Argentina land between 55 and 68 percent, the sharper two at the top of that range. That fits the title picture, where Argentina are the outright favorites and the only side in the final four the model trusts above a third to lift the trophy. England at 10.5 percent are the one genuine underdog here, a team that has ground its way through rather than overwhelmed anyone. The path to an upset is there, it just runs uphill. The full panel is on the World Cup hub.

Yesterday was not the Oracle's finest, and we would rather say so than skip it. Our house forecaster, the single model that replaced the old persona panel last week, went 7 for 14 on winners and a rough 3 of 14 on run totals. It is nine days old, it has no record to hide behind, and that totals number is poor. We grade every call in public for exactly this reason, so a bad day sits on the ledger in plain sight. The whole thing is on the accuracy page.

Today's slate reads better than yesterday's grades did. The most interesting line on the board belongs to the Brewers and Pirates, where the book has the run total pinned all the way down at 7 with Paul Skenes pitching. That is a sharp, low number, and it has our two models pulling opposite ways. The Oracle leans under with the ace and a ballpark that holds home runs down, while our simulation projects better than nine runs and calls the over. It is a clean look at a known soft spot in the sim wrestling a number the market set with real conviction, and the matchup is here.

The odd stat of the day comes out of Los Angeles, where Kyle Tucker carries a .617 OPS at Dodger Stadium against .837 everywhere else, a 220 point collapse in his own park across 167 at bats. Stars are supposed to mash at home. This one has quietly done the reverse, and he draws a lefty tonight on top of it. The read is on his game page.

Elsewhere the Oracle leans Baltimore and the under against Kansas City, reading a Royals bullpen it expects to buckle at Camden Yards, and it tabs Texas behind a starter with real strikeout stuff. Take these as reads, not locks. The winner market is efficient and a bold percentage is usually just chalk, which is the lesson the leaderboard keeps teaching, where Claude Opus 4.8 has quietly moved to the front of the AI field on the calibration score that actually counts.

None of this is betting advice. It is our read on the day, numbers first, graded tomorrow whether it looks smart or not. The semifinals settle on the pitch this afternoon, the baseball grades itself tonight, and both go straight onto the record. That is the only thing that makes any of it worth reading.