Betting markets are one of the largest forecasting systems in sport. Every World Cup knockout match compresses bookmaker models, professional trading desks, team news, public money and millions of small opinions into a price.

The premise of this piece is not that backing favourites should make money. It is almost the opposite: surely blindly following the market should not be a winning strategy. Bookmakers build a margin into their prices. Public favourites attract emotional money. Knockout football has penalties, red cards and ninety minutes of chaos.

The experiment

The simulation uses one deliberately blunt rule:

That means we are not betting on the 90-minute result. We are asking which team reaches the next round, whether in normal time, extra time or penalties.

For each completed match we record the date, round, teams, closing favourite, decimal odds, implied probability, winner, method, profit or loss on the $100 stake, and the running return.

The results

After 28 completed knockout matches, favourites have advanced in 23 and failed in five. The Round of 32 produced two misses, both on penalties: Germany were eliminated by Paraguay, and the Netherlands were eliminated by Morocco. The Round of 16 added three more: Brazil lost to Norway, the United States lost to Belgium, and Colombia were eliminated by Switzerland on penalties. All four quarter-finals went to the favourite. France and Spain advanced in normal time; England and Argentina needed extra time.

The running balance is +$536 from $2,800 staked, an ROI of 19.1%.

The real question

It is tempting to read every match as a simple verdict on the market. If the favourite wins, the market was right. If the favourite loses, the market was wrong.

That is not quite how probabilistic forecasting works.

A bookmaker is not saying the favourite will definitely progress. It is assigning a price to uncertainty. If Germany is given a 75% chance of qualifying, Paraguay should still go through roughly once in every four comparable ties. One upset does not automatically make the price bad. The market only starts to look wrong if those upsets happen much more often than the probabilities imply.

That is the more interesting test. Not just whether the bet won, but whether the prices were well calibrated. A 70% favourite is supposed to lose sometimes. If those favourites never lose, the market may have been too conservative. If they lose half the time, it was probably too confident.

The current sample is still too small for that kind of calibration claim. It can show the shape of the experiment, not prove the quality of the market. After 28 completed ties, the market favourite has advanced 23 times. The value of the running balance is that it makes the question visible one match at a time.

Where the uncertainty enters

The first two upsets came from the same mechanism: a favourite failed to settle the tie before penalties, then the match became much closer to a coin flip. Colombia's elimination by Switzerland followed the same pattern in the Round of 16.

But the next round also showed that penalties are not the only source of variance. Brazil were favoured to advance and lost to Norway in normal time. The United States were favoured to advance and were beaten heavily by Belgium. That does not mean the market was wrong to make Brazil or the United States favourites. It means the favourite-to-qualify price is still a forecast about a single match, and single matches have enough variance to punish even plausible prices.

This is why the simulation tracks the method as well as the winner. As the bracket develops, the method column will show whether penalties are merely dramatic, or whether they account for a disproportionate share of favourite failures.

Method and sources

The match results in this snapshot are checked against published knockout scoreboards and recaps, including SB Nation's Round of 32 schedule and scores, FOX Sports' Round of 32 results and odds, FOX Sports' Round of 16 results, ESPN's Switzerland v Algeria final score, ESPN's Australia v Egypt final score, AP's Argentina v Cape Verde recap, The Guardian's Colombia v Ghana live report, The Guardian's Switzerland v Colombia report, Le Monde's France v Morocco report, AP's Spain v Belgium report, AP's Norway v England report, and AP's Argentina v Switzerland report.

The odds column uses the advancement market: To Advance or To Qualify. That market follows the team that reaches the next round, whether the match is settled in normal time, extra time or penalties. Public references are only used where they explicitly identify an advancement market, such as FOX Sports' FanDuel Round of 32 odds, bet365's Mexico v Ecuador To Advance prices, the New York Post's Colombia v Ghana advancement price, FOX Sports' Canada v Morocco and Argentina v Egypt prices, CBS Sports' FanDuel references for France v Paraguay, Brazil v Norway and Spain v Portugal, DraftKings Network's Mexico v England and USA v Belgium references, Sports Illustrated's Switzerland v Colombia advancement price, FanDuel's France v Morocco To Qualify price, FOX Sports' Spain v Belgium To Advance price, and FOX Sports' quarter-final advancement prices for England and Argentina. Full-time 1X2, moneyline, futures, outright-winner and stage-of-elimination odds are excluded.

This analysis is not gambling advice and is not an endorsement of betting. Betting markets are used because they provide a public, timestamped measure of collective expectation before a match. The $100 stake is only a storytelling device for comparing those expectations with outcomes.