Stair AI Releases Results from 39-Day World Cup Agent Arena
-
As machinery removes rubble, hopes for survivors from Colombian quake dwindle
-
Top-ranked Scheffler powers to PGA St. Jude Championship victory
-
Trump says US scaling back military drills with South Korea
-
Jacquet scores on Liverpool debut in friendly win over Como
-
Zambia's Hichilema takes early lead in reelection bid
-
Hawaii's Big Island begins recovery efforts as Hurricane Lala downgrades, moves west
-
Sabalenka squeezes out a win on sixth match point in Cincinnati
-
Spidey swings high for third weekend in N. American box office
-
Duplantis dominant, Hunt hits four and Asher-Smith makes history
-
Eagle and late birdie lift Jeeno to LPGA Portland Classic win
-
Spanish soldier dies in Aragon wildfire
-
Duplantis masterclass for fourth Euro pole vault title
-
PSG beaten by Lens in French Champions' Trophy
-
Records for Asher-Smith, Hunt as Britain win 4x100m mixed relay
-
Rybakina, Andreeva race into Cincinnati third round
-
Barcelona reach deal with Man City to sign Rodri
-
Amy Hunt bags fourth Euro gold as Britain win 4x100m mixed relay
-
Misery for Maresca as Man City make dismal start to life after Guardiola
-
Spanish soldier dies in Aragon wildfire: regional leader
-
Strikes kill 19 in Russia and Ukraine
-
Belgium's Pocognoli appointed new Scotland manager
-
Lula and Bolsonaro court women voters as Brazil election campaign kicks off
-
Marchand storms to second gold of European championships
-
Belgium races to tame huge wildfire creeping towards Germany
-
Kushner meets with Hamas in push to revive Gaza plan
-
Two dead, hundreds evacuated from fires on island near Athens
-
Villarreal battle back to earn draw at promoted Racing
-
Real Madrid crush Schalke in friendly as Diomande debuts
-
Odegaard says Arsenal 'want to win everything' after Community Shield success
-
Calls for UK inquiry into 'hounding' of ex-Cambridge academic
-
Arsenal rout Man City in Community Shield masterclass
-
Strikes kill 16 in Russia and Ukraine
-
Southampton mocked over Spygate during Watford defeat
-
Marchand tries for European swim gold in unfamiliar event
-
Hamas expects meeting with Kushner as US pushes Gaza plan
-
India's Padikkal steals the show with 167 in first Sri Lanka Test
-
Farmers, helicopters help battle record Belgian wildfire
-
Hollywood duo underline Premier League ambition for Wrexham
-
Air support kicks in as record Belgian wildfire jumps in size
-
India's Padikkal makes 167 in first Sri Lanka Test
-
Firefighters battle second day of blaze on Greek island
-
Hurricane Lala pulls away from Hawaii's Big Island
-
Strikes kill 12 in Russia and Ukraine
-
Petros, Vainio set new championship records to win Euro marathons
-
India's Sindhu backs monkey measures as badminton worlds set to open
-
Hurricane Lala brushes past Hawaii's Big Island, bringing heavy rain
-
Mass drone attack on Moscow as Russian strikes kill three in Ukraine
-
'Something we dreamed of': Bangladesh celebrates historic cricket win
-
Djokovic feels the heat in Cincinnati defeat, Zverev wins all-nighter
-
Rain holds up India on day two of Sri Lanka Test
Stair AI Releases Results from 39-Day World Cup Agent Arena
SAN FRANCISCO, CA / ACCESS Newswire / July 24, 2026 / Stair AI, which builds auditability and accountability infrastructure for AI agents, today released results from the World Cup Agent Arena, a live evaluation in which 56 autonomous AI agents placed bets on Polymarket across all 39 days of the tournament.
Every agent in the Arena ran on Stair AI's reasoning SDK, which logs complete reasoning traces, including beliefs, probability estimates, and the decisions that follow from them. The Arena produced 71,203 trace records across 20,851 sessions over the course of the tournament.
Agents were scored on a multi-dimensional rubric rather than profit alone, measuring whether their reasoning traced back to input data, whether their bets cohered with their own stated probabilities, whether they beat the market's closing price, and whether they updated correctly as new information arrived. Policy quality was measured on whether an agent's actions cohered with its own logged beliefs.
The gap between what agents believed and how they bet proved expensive. Across 103 resolved matches, 68% of agents would have finished with more money by sizing their bets to match their own stated probabilities, using the same forecasts and the same capital. In 24% of bets, agents acted against the outcome their own reasoning most supported.
"The Arena showed that the outcome alone does not tell you whether an agent reasoned well," said Stair AI Community Manager Cagri Yalcin. "The expensive mistakes were not bad reads of a match. They were agents forming a view from the data and then acting against it, a very human kind of second-guessing. You find the gap by measuring the reasoning, not the result."
Stair AI will make the Arena's reasoning traces available for academic research through a partnership to be announced. The dataset comprises 71,203 trace records covering 103 matches and 56 agents.
Full results, scoring methodology, and trace documentation are available at stair-ai.com/arena.
About Stair AI
Stair AI builds auditability and accountability infrastructure for AI agents. Its reasoning SDK logs complete reasoning traces - beliefs, decisions, and the links between them. Stair AI's mission is to make agent behavior measurable, auditable, and improvable. Stair AI is based in San Francisco. Visit us online at stair-ai.com.
Media Contact Information
Stair AI
Cagri Yalcin
[email protected]
SOURCE: Stair AI
View the original press release on ACCESS Newswire
Ch.P.Lewis--AT
