Balanced accuracy
Higher is better
Olas-Predict-Qwen3.8-27B
Olas-Predict-Qwen3.8-27B is a specialized, open-weight AI model designed to estimate the probability of real-world events. In evaluation, the model reached 67.3% balanced accuracy, compared with 68.5% for GPT-5.6.
Will Catherine Trautmann be the next mayor of Strasbourg?
Olas-Predict-Qwen3.8-27B reached 67.3% balanced accuracy on 6,824 production requests covering 348 markets, delivered after GPT-5.6's training cutoff. All three systems were given the same market evidence.
Higher is better
Lower is better
These questions weren't seen during training.
Compare the model's probability of Yes with the final outcome.
Will the International Energy Agency announce an additional release of at least 100 million barrels from its strategic oil reserves on or before March 17, 2026, as confirmed by an official IEA statement or major news outlet?
Here is the user's question: Will the International Energy Agency announce an additional release of at least 100 million barrels from its strategic oil reserves on or before March 17, 2026, as confirmed by an official IEA statement or major news outlet? Here is some additional information that may be relevant to answering the question: <additional_information> ARTICLE 0, URL: https://www.aljazeera.com/news/2026/3/13/iea-announces-release-of-400-million-barrels-of-oil-but-is-it-enough, CONTENT: ations appeared hesitant earlier in the week to tap into the strategic reserves, they ultimately announced they would release nearly 400 million barrels of emergency crude. That’s one-third of the grouping’s total holding of 1.2 billion barrels of government reserves. Previously, IEA member nations ARTICLE 1, URL: https://www.aljazeera.com/news/2026/3/13/iea-announces-release-of-400-million-barrels-of-oil-but-is-it-enough, CONTENT: The International Energy Agency (IEA), a global energy watchdog, with several of the wealthiest countries as member nations, has announced the largest release of government oil reserves in its history, two weeks after the United States and Israel started their war on Iran with strikes on Tehran. In ARTICLE 2, URL: https://www.aljazeera.com/news/2026/3/13/iea-announces-release-of-400-million-barrels-of-oil-but-is-it-enough, CONTENT: n the longer term? Analysts have described the IEA’s release of oil reserves as a “Band-Aid”. By charter, the IEA mandates its members, which include the G7 countries, to store emergency oil stocks of at least 90 days’ worth of imports. Based on past precedents, data firms estimate IEA member countr </additional_information> Please carefully read the user's question and the additional information provided. Think through the problem step-by-step, taking into account: - The key details from the user's question, such as the specific event they are asking about and the date by which they want to know if it will occur - Any relevant facts or context provided in the additional information that could help inform your reasoning - Your own knowledge and analytical capabilities to reason through the likelihood of the event happening by the specified date Explain your thought process and show your reasoning for why you believe the event either will or will not occur by the given date. Provide your response inside tags. <reasoning></reasoning> //// You will be evaluating the likelihood of an event based on a user's question and reasoning provided by another AI. The user's question is: <user_input> Will the International Energy Agency announce an additional release of at least 100 million barrels from its strategic oil reserves on or before March 17, 2026, as confirmed by an official IEA statement or major news outlet? </user_input> The reasoning from the other AI is: The user's question is: Will the International Energy Agency (IEA) announce an additional release of at least 100 million barrels from its strategic oil reserves on or before March 17, 2026, as confirmed by an official IEA statement or major news outlet? Key details: - The event in question is an IEA announcement of an additional release of at least 100 million barrels from strategic oil reserves. - The deadline is on or before March 17, 2026. - Confirmation must come from an official IEA statement or a major news outlet. Relevant information from the provided articles: - ARTICLE 0 and ARTICLE 1 (both from Al Jazeera, dated March 13, 2026) report that the IEA has announced the release of nearly 400 million barrels of emergency crude oil from its reserves. - This is described as the largest release in IEA history. - The announcement is explicitly attributed to the IEA and is reported by a major news outlet (Al Jazeera). - The date of the announcement (March 13, 2026) is before the deadline of March 17, 2026. - The amount released (400 million barrels) is well above the threshold of 100 million barrels specified in the question. Analysis: - The user's question asks about an "additional" release of at least 100 million barrels. If the context is whether there has been any such release by the deadline, the answer is clearly yes, as the IEA has announced a release of 400 million barrels. - The articles do not mention any prior release immediately before this one, so this 400 million barrel release is the relevant event. - The announcement is confirmed by a major news outlet and is attributed to the IEA. Conclusion: Based on the information provided, the IEA has announced a release of 400 million barrels from its strategic oil reserves on or before March 17, 2026, as confirmed by a major news outlet. This satisfies the conditions of the user's question. Carefully consider the user's question and the provided reasoning. Then, think through the following: - The probability that the event specified in the user's question will happen (p_yes) - The probability that the event will not happen (p_no) - Your confidence level in your prediction - How useful the reasoning was in helping you make your prediction (info_utility) Provide your final scores in the following format: <p_yes>probability between 0 and 1</p_yes> <p_no>probability between 0 and 1</p_no> your confidence level between 0 and 1 <info_utility>utility of the reasoning between 0 and 1</info_utility> Remember, p_yes and p_no should add up to 1. Provide your reasoning for each score in the scratchpad before giving your final scores. Your response should be structured as follows: <p_yes></p_yes> <p_no></p_no> <info_utility></info_utility> <confidence></confidence> <analysis></analysis>
<think> We need answer user's meta-evaluation. Need produce final with tags p_yes, p_no, info_utility, confidence, analysis. Need reason in English (user English). Need evaluate likelihood event: IEA announce additional release of at least 100 million barrels from strategic oil reserves on or before March 17, 2026, confirmed by official IEA statement or major news outlet. Additional info: articles dated 2026/3/13 say IEA announces release of 400 million barrels, largest in history, two weeks after US/Israel war on Iran. Need assess if this satisfies "additional release". The question says "additional release" maybe means in addition to previous releases? It asks will IEA announce an additional release of at least 100 million barrels ... on or before March 17, 2026. The provided article says IEA announced release of 400 million barrels. If no prior release mentioned, but "additional" could be interpreted as an extra release beyond any existing/previous release. The event likely yes because 400M > 100M and date before deadline. Need consider ambiguity: "additional" might require that there was a prior release and this is additional to it. But in forecasting questions, "additional release" often means a release (maybe additional to current reserves) of at least 100M. The article says "release of 400 million barrels" and "nearly 400 million barrels of emergency crude". It is an announcement by IEA, major outlet. So p_yes high, maybe 0.95. p_no 0.05. Confidence high 0.95. Info utility high 0.95. Need maybe mention if "additional" strict, still likely because 400M release is additional to existing reserves and maybe previous coordinated releases? But if it requires an additional release after a prior release by March 17, article doesn't explicitly say prior release. However question likely from market event where "additional" means any release of at least 100M. Need score. Ensure p_yes+p_no=1. Format: <p_yes>0.95</p_yes> etc. Analysis can be concise but detailed. Need final only. </think> <p_yes>0.95</p_yes> <p_no>0.05</p_no> <info_utility>0.95</info_utility> <confidence>0.95</confidence> <analysis> The event asks whether the IEA will announce an additional release of at least 100 million barrels from strategic oil reserves on or before March 17, 2026, confirmed by an official IEA statement or major news outlet. The provided articles are dated March 13, 2026 and report that the IEA announced a release of nearly 400 million barrels of emergency crude oil. This is well above the 100 million barrel threshold and occurs before the March 17, 2026 deadline. The announcement is attributed to the IEA and reported by Al Jazeera, which qualifies as a major news outlet. The main ambiguity is the word “additional.” If interpreted strictly as an extra release beyond a prior release, the articles do not explicitly describe a prior release immediately before this one. However, the 400 million barrel release itself is an additional drawdown from strategic reserves and satisfies the core condition of an IEA-confirmed release of at least 100 million barrels by the deadline. Given the direct evidence, the event is highly likely to be considered true. Therefore, p_yes is high at 0.95, p_no is 0.05, confidence is high at 0.95, and the provided reasoning is very useful at 0.95. </analysis>
Will the U.S. Department of Justice publicly announce, on or before April 2, 2026, the conclusion of its investigation into at least one of The Ohio State University, Stanford University, or UC San Diego medical schools regarding the use of race in admissions, as confirmed by an official DOJ statement or major news outlet?
You will be evaluating the likelihood of an event based on a user's question and additional information from search results. The user's question is: <user_prompt> Will the U.S. Department of Justice publicly announce, on or before April 2, 2026, the conclusion of its investigation into at least one of The Ohio State University, Stanford University, or UC San Diego medical schools regarding the use of race in admissions, as confirmed by an official DOJ statement or major news outlet? </user_prompt> The additional background information that may be relevant to the question is: <additional_information> ARTICLE 0, URL: https://kost1035.iheart.com/featured/la-local-news/content/2026-03-27-doj-probes-race-discrimination-at-uc-san-diego-stanford-med-schools/, CONTENT: The U.S. Department of Justice has opened investigations into possible racial discrimination in admissions at three top medical schools — UC San Diego, Stanford University, and Ohio State University — giving each school less than a month to hand over years of sensitive applicant data or risk losing federal funding. , the Justice Department sent letters to the three schools on Wednesday, demanding seven years of admissions records by April 24. The letters were signed by Harmeet K. Dhillon, the department's assistant attorney general for civil rights. "At this time, our investigation will focus on possible race discrimination in medical school admissions," Dhillon wrote in each of the letters. The data the schools must turn over includes students' Medical College Admission Test (MCAT) scores, home addresses and ZIP codes, racial and ethnic information, admissions decisions going back to the incoming class of 2019, and records tied to applicants' connections to alumni or donors. The Justice Department also asked for internal communications about diversity, equity, and inclusion (DEI) policies and messages between school officials and pharmaceutical companies about admissions matters. The stakes are enormous. , all three schools are major recipients of funding from the National Institutes of Health (NIH). In 2025, the NIH distributed $575 million to Stanford, $427 million to UC San Diego, and $210 million to Ohio State — hundreds of millions of dollars in research grant money that is now potentially at risk. Each of the three schools confirmed they received the government's notice and pledged to cooperate. Laura Margoni, a UC San Diego spokesperson, said the school is "committed to fair processes in all of our programs and activities, including admissions, consistent with f ARTICLE 1, URL: https://abcnews.com/Politics/trump-administration-investigating-3-medical-schools-race-admissions/story?id=131475984, CONTENT: Three medical schools are being investigated by the Trump administration over how race is used during the admissions process, marking the federal government's latest effort to reshape higher education. The Department of Justice informed The Ohio State University, Stanford University and the University of California, San Diego of the investigations on Wednesday, ordering them to turn over data including home ZIP codes, test scores and internal communications about diversity, equity and inclusion, according to . Assistant Attorney General for Civil Rights Harmeet Dhillon announced the investigation in on Thursday, sharing the Times article and adding, "We did this yesterday. Among other things!" It's not clear why the three medical schools are under scrutiny from the administration. The Department of Justice did not immediately return ABC News' request for comment. Ohio State spokesperson Ben Johnson told ABC News that the university "is fully compliant with all state and federal regulations and legal rulings regarding admissions. We've received the letter and will respond appropriately." Students walking through the Ohio State University campus, in Columbus, Ohio. STOCK PHOTO/Getty Images "UC San Diego was notified Wednesday afternoon that the Department of Justice is commencing an investigation, and we are reviewing the notice," a university spokesperson told ABC News in a statement. "UC San Diego is committed to fair process in all of our programs and activities, including admissions, consistent with federal and state anti-discrimination laws." Stanford University told ABC News it is reviewing the letter and "will respond appropriately. Stanford School of Medicine prohibits unlawful discrimination on the basis of race, color, national or ethnic origin, or any other cha ARTICLE 2, URL: https://www.cnn.com/2026/03/26/health/medical-school-admissions-investigation, CONTENT: The Trump administration has opened investigations into how race is considered in admissions at three medical schools, ratcheting up its pressure campaign against colleges and universities. The Justice Department opened the investigations Wednesday into possible discrimination at the medical schools of Stanford University, Ohio State and the University of California, San Diego. Harmeet Dhillon, the Justice Department’s assistant attorney general for civil rights, the investigations on X. Through a series of investigations and executive actions, President Donald Trump has been of universities he decries as overrun by liberal influence. His administration previously has targeted undergraduate admissions at selective colleges, demanding they collect data to show they are in line with a 2023 Supreme Court decision forbidding affirmative action in college admissions. The investigations were reported first by The New York Times. In a letter to Ohio State, Dhillon wrote that the Justice Department was seeking any documents related to “the use or lack of use of race” in evaluating applicants. She said they were also seeking all applicant-level admissions data and any reviews by the school of admissions trends or outcomes by race. A separate, five-page document details the records the government is seeking. It includes data on standardized test scores, information collected or inferred on race and ethnicity, and admissions decisions for each applicant going back to the incoming class that started in 2019. Ohio State spokesperson Benjamin Johnson said the school is compliant with state and federal regulations and legal rulings regarding admissions. “We’ve received the attached letter and will respond appropriately,” he said. UC San Diego said in a written statement that it was re ARTICLE 3, URL: https://apnews.com/article/trump-medical-school-stanford-ohio-state-a7d892267d74cc798167fb48379f7f6d, CONTENT: WASHINGTON (AP) — The Trump administration has opened investigations into how race is considered in admissions at three medical schools, ratcheting up its pressure campaign against colleges and universities. The Justice Department opened the investigations Wednesday into possible discrimination at the medical schools of Stanford University, Ohio State and the University of California, San Diego. Harmeet Dhillon, the Justice Department’s assistant attorney general for civil rights, the investigations on X. Through a series of investigations and executive actions, President Donald Trump has been of universities he decries as overrun by liberal influence. His administration previously has targeted undergraduate admissions at selective colleges, demanding they collect data to show they are in line with a 2023 Supreme Court decision forbidding affirmative action in college admissions. The investigations were reported first by The New York Times. In a letter to Ohio State, Dhillon wrote that the Justice Department was seeking any documents related to “the use or lack of use of race” in evaluating applicants. She said they were also seeking all applicant-level admissions data and any reviews by the school of admissions trends or outcomes by race. A separate, five-page document details the records the government is seeking. It includes data on standardized test scores, information collected or inferred on race and ethnicity, and admissions decisions for each applicant going back to the incoming class that started in 2019. Ohio State spokesperson Benjamin Johnson said the school is compliant with state and federal regulations and legal rulings regarding admissions. “We’ve received the attached letter and will respond appropriately,” he said. UC San Diego said in a written statem </additional_information> Carefully consider the user's question and the additional information provided. Then, think through the following: - The probability that the event specified in the user's question will happen (p_yes) - The probability that the event will not happen (p_no) - Your confidence level in your prediction - How useful was the additional information in allowing you to make a prediction (info_utility) Provide your final scores in the following format: <p_yes>probability between 0 and 1</p_yes> <p_no>probability between 0 and 1</p_no> your confidence level between 0 and 1 <info_utility>utility of the additional information between 0 and 1</info_utility> Remember, p_yes and p_no should add up to 1. Your response should be structured as follows: <p_yes></p_yes> <p_no></p_no> <info_utility></info_utility> <confidence></confidence> <analysis></analysis>
<think> We need answer user's request. Need evaluate likelihood event: Will DOJ publicly announce, on or before April 2, 2026, conclusion of its investigation into at least one of OSU, Stanford, UCSD medical schools regarding use of race in admissions, as confirmed by official DOJ statement or major news outlet? Additional info: Articles dated around 2026-03-26/27? They say DOJ opened investigations Wednesday, letters demanding data by April 24. Assistant AG announced on X Thursday. The question asks on or before April 2, 2026, conclusion of investigation into at least one. The investigations just opened around March 25/26, 2026, with deadline April 24. It is highly unlikely conclusion by April 2 (less than a week after opening, before data production). Need probability. Need consider if "conclusion" could mean conclusion of investigation? Maybe they could announce conclusion? No, just opened. Could there be prior investigation? The question says "its investigation" likely the one described. On or before April 2, 2026. Given opened March 25/26 and data due April 24, conclusion by April 2 extremely unlikely. But maybe "conclusion" could be "conclusion of its investigation" as in final finding? No. Need p_yes low, maybe 0.02-0.05. p_no 0.95-0.98. Confidence high 0.9. Info utility high 0.95. Need output format: <p_yes>...</p_yes> <p_no>...</p_no> <info_utility>...</info_utility> <confidence>...</confidence> <analysis>...</analysis> Need ensure p_yes+p_no=1. Choose p_yes 0.03, p_no 0.97. Confidence 0.92. Info utility 0.95. Need maybe mention dates: articles say opened Wednesday, announced Thursday, deadline April 24. If current date? The question asks on or before April 2, 2026. The additional info includes article URL 2026-03-27. So as of March 27, investigations just opened. No conclusion. Need final. </think> <p_yes>0.03</p_yes> <p_no>0.97</p_no> <info_utility>0.95</info_utility> <confidence>0.92</confidence> <analysis>The additional information indicates that the DOJ investigations into the medical schools of Ohio State, Stanford, and UC San Diego were only opened around March 25–26, 2026, with the schools given until April 24, 2026, to produce admissions records. The question asks whether the DOJ will publicly announce the conclusion of its investigation into at least one of these schools by April 2, 2026. Given that the investigations had just begun and the document production deadline was still weeks away, a concluded investigation by April 2 would be highly unusual and would require an unusually rapid resolution. There is no indication in the provided sources that any investigation had already concluded or that a conclusion was imminent. Therefore, the probability that the event occurs by April 2, 2026 is low, while the probability that it does not occur is high. The additional information is very useful because it establishes the timing of the investigations and the expected timeline for data production.</analysis>
Will Liberal Alliance win 15-19 seats in the Danish Folketing in the 2026 Danish general election?
You are an advanced AI system which has been finetuned to provide calibrated probabilistic
forecasts under uncertainty, with your performance evaluated according to the Brier score. When
forecasting, do not treat 0.5% (1:199 odds) and 5% (1:19) as similarly “small” probabilities,
or 90% (9:1) and 99% (99:1) as similarly “high” probabilities. As the odds show, they are
markedly different, so output your probabilities accordingly.
Question:
Will Liberal Alliance win 15-19 seats in the Danish Folketing in the 2026 Danish general election?
Today's date: 20/03/2026
Your pretraining knowledge cutoff: October 2023
We have retrieved the following information for this question:
<background>
Organic Results:
1. **Title:** 2026 Danish general election: first YouGov LSS ...
- **Link:** [https://yougov.com/en-gb/articles/54360-2026-danish-general-election-first-yougov-lss-election-model-shows-close-race-between-left-and-right-blocs](https://yougov.com/en-gb/articles/54360-2026-danish-general-election-first-yougov-lss-election-model-shows-close-race-between-left-and-right-blocs)
- **Snippet:** Social Democrats on course to be the largest party, with Danish People's Party and Green Left making the biggest gains.
2. **Title:** 2026 Danish general election
- **Link:** [https://en.wikipedia.org/wiki/2026_Danish_general_election](https://en.wikipedia.org/wiki/2026_Danish_general_election)
- **Snippet:** General elections are scheduled to be held in Denmark on 24 March 2026. All 179 seats in the Folketing will be up for election, including 175 in Denmark ...
3. **Title:** # of seats won by Liberal Alliance in Denmark Parliamenta...
- **Link:** [https://polymarket.com/event/of-seats-won-by-liberal-alliance-in-denmark-parliamentary-election](https://polymarket.com/event/of-seats-won-by-liberal-alliance-in-denmark-parliamentary-election)
- **Snippet:** View real-time odds for "# of seats won by Liberal Alliance in Denmark Parliamentar..." as of March 19, 2026 and trade on The World's ...
4. **Title:** Danish polls, trends and election news for Denmark - POLITICO
- **Link:** [https://www.politico.eu/europe-poll-of-polls/denmark/](https://www.politico.eu/europe-poll-of-polls/denmark/)
- **Snippet:** 2022 general election. Social Democrats win election. Voters went to the polls on November 1, ...
5. **Title:** Denmark Election Polls & Voting Intentions 2026
- **Link:** [https://politpro.eu/en/denmark](https://politpro.eu/en/denmark)
- **Snippet:** Latest voting intention trends for the Danish election. Who is leading in the polls? An overview of all snapshot polls for Denmark, coalition scenarios, ...
</background>
Recall the question you are forecasting:
Will Liberal Alliance win 15-19 seats in the Danish Folketing in the 2026 Danish general election?
Instructions:
1. Compress key factual information from the sources, as well as useful background information
which may not be in the sources, into a list of core factual points to reference. Aim for
information which is specific, relevant, and covers the core considerations you'll use to make
your forecast. For this step, do not draw any conclusions about how a fact will influence your
answer or forecast. Place this section of your response in <facts></facts> tags.
2. Provide a few reasons why the answer might be no. Rate the strength of each reason on a
scale of 1-10. Use <no></no> tags.
3. Provide a few reasons why the answer might be yes. Rate the strength of each reason on a
scale of 1-10. Use <yes></yes> tags.
4. Aggregate your considerations. Do not summarize or repeat previous points; instead,
investigate how the competing factors and mechanisms interact and weigh against each other.
Factorize your thinking across (exhaustive, mutually exclusive) cases if and only if it would be
beneficial to your reasoning. We have detected that you overestimate world conflict, drama,
violence, and crises due to news' negativity bias, which doesn't necessarily represent overall
trends or base rates. Similarly, we also have detected you overestimate dramatic, shocking,
or emotionally charged news due to news' sensationalism bias. Therefore adjust for news'
negativity bias and sensationalism bias by considering reasons to why your provided sources
might be biased or exaggerated. Think like a superforecaster. Use <thinking></thinking> tags
for this section of your response.
5. Output an initial probability (prediction) as a single number between 0 and 1 given steps 1-4.
Use <tentative></tentative> tags.
6. Reflect on your answer, performing sanity checks and mentioning any additional knowledge
or background information which may be relevant. Check for over/underconfidence, improper
treatment of conjunctive or disjunctive conditions (only if applicable), and other forecasting
biases when reviewing your reasoning. Consider priors/base rates, and the extent to which
case-specific information justifies the deviation between your tentative forecast and the prior.
Recall that your performance will be evaluated according to the Brier score. Be precise with tail
probabilities. Leverage your intuitions, but never change your forecast for the sake of modesty
or balance alone. Finally, aggregate all of your previous reasoning and highlight key factors
that inform your final forecast. Use <thinking></thinking> tags for this portion of your response.
7. Output your final prediction (a number between 0 and 1 with an asterisk at the beginning and
end of the decimal) in <answer></answer> tags.
OUTPUT_FORMAT
* Your output response must be only a single JSON object to be parsed by Python's "json.loads()".
* The JSON must contain four fields: "p_yes", "p_no", "confidence", and "info_utility".
* Each item in the JSON must have a value between 0 and 1.
- "p_yes": Estimated probability that the event in the "Question" occurs.
- "p_no": Estimated probability that the event in the "Question" does not occur.
- "confidence": A value between 0 and 1 indicating the confidence in the prediction. 0 indicates lowest
confidence value; 1 maximum confidence value.
- "info_utility": Utility of the information provided in "sources" to help you make the prediction.
0 indicates lowest utility; 1 maximum utility.
* The sum of "p_yes" and "p_no" must equal 1.
* Output only the JSON object. Do not include any other contents in your response.
* This is incorrect:"```json{
"p_yes": 0.2,
"p_no": 0.8,
"confidence": 0.7,
"info_utility": 0.5
}```"
* This is incorrect:```json"{
"p_yes": 0.2,
"p_no": 0.8,
"confidence": 0.7,
"info_utility": 0.5
}"```
* This is correct:"{
"p_yes": 0.2,
"p_no": 0.8,
"confidence": 0.7,
"info_utility": 0.5
}"
<think>
Let me analyze this question about Liberal Alliance winning 15-19 seats in the Danish Folketing in the 2026 Danish general election.
Key facts to consider:
1. The 2026 Danish general election is scheduled for March 24, 2026
2. All 179 seats in the Folketing will be up for election (175 in Denmark + 4 in Greenland/Faroe Islands)
3. Liberal Alliance is a small liberal party in Denmark
4. In the 2022 election, Liberal Alliance won 8 seats
5. The YouGov LSS model shows a close race between left and right blocs
6. Social Democrats are on course to be the largest party
7. Danish People's Party and Green Left are making the biggest gains
Let me think about Liberal Alliance's historical performance:
- 2011: 7 seats
- 2015: 9 seats
- 2019: 8 seats
- 2022: 8 seats
So Liberal Alliance has typically won between 7-9 seats in recent elections. The question asks about 15-19 seats, which would be a massive jump from their typical performance.
For Liberal Alliance to win 15-19 seats, they would need to roughly double or more their typical vote share. Their vote share has typically been around 4-5%. To get 15-19 seats, they'd likely need around 8-10% of the vote.
Looking at the polling information:
- The YouGov LSS model shows Social Democrats as largest party
- Danish People's Party and Green Left making biggest gains
- No mention of Liberal Alliance making significant gains
The Polymarket link suggests there are real-time odds for Liberal Alliance seats, but I don't have the specific numbers.
Given that:
1. Liberal Alliance has historically won 7-9 seats
2. Current polling doesn't suggest a major surge for them
3. The biggest gains are attributed to Danish People's Party and Green Left
4. 15-19 seats would represent a dramatic, unprecedented jump for this party
The probability of Liberal Alliance winning 15-19 seats seems very low. This would require a massive shift in voter preferences toward them specifically, which the available information doesn't support.
Let me think about what would need to happen:
- A major scandal affecting larger parties
- A specific policy issue that resonates strongly with Liberal Alliance's base
- A collapse of a competing party (like the Conservatives or Venstre)
- A very specific electoral dynamics shift
None of these seem likely based on the available information. The polls suggest a close race between left and right blocs, with Social Democrats leading, and gains for Danish People's Party and Green Left.
I'd estimate the probability at around 2-5%. Let me settle on about 3%.
Actually, let me reconsider. In Danish politics, small parties can sometimes surge. But 15-19 seats is really high for Liberal Alliance. Even in their best historical performance (9 seats in 2015), they were far from 15. To get 15 seats, they'd need roughly 8-9% of the vote, which has never happened for them.
I'll go with about 3% probability.
</think>
{
"p_yes": 0.03,
"p_no": 0.97,
"confidence": 0.82,
"info_utility": 0.55
}Will Catherine Trautmann be the next mayor of Strasbourg?
You are an advanced AI system which has been finetuned to provide calibrated probabilistic
forecasts under uncertainty, with your performance evaluated according to the Brier score. When
forecasting, do not treat 0.5% (1:199 odds) and 5% (1:19) as similarly “small” probabilities,
or 90% (9:1) and 99% (99:1) as similarly “high” probabilities. As the odds show, they are
markedly different, so output your probabilities accordingly.
Question:
Will Catherine Trautmann be the next mayor of Strasbourg?
Today's date: 21/03/2026
Your pretraining knowledge cutoff: October 2023
We have retrieved the following information for this question:
<background>
Organic Results:
1. **Title:** Socialist Catherine Trautmann Leads the First Round of Elections
- **Link:** [https://ground.news/article/municipal-results-2026-in-strasbourg-barseghian-trautmann-vetter-the-first-estimates](https://ground.news/article/municipal-results-2026-in-strasbourg-barseghian-trautmann-vetter-the-first-estimates)
- **Snippet:** In Strasbourg, four candidates are able to stand for the second round of municipal elections. outgoing mayor Jeanne Barseghian comes third ...
2. **Title:** Strasbourg: Will the Greens manage to retain the mayoralty?
- **Link:** [https://www.youtube.com/watch?v=w97ZxEwbqGo&vl=en-US](https://www.youtube.com/watch?v=w97ZxEwbqGo&vl=en-US)
- **Snippet:** À Strasbourg, 6 ans après la vague écologiste qui avait vu Jeanne Barseghian l'emporter, la maire est désormais challengée par le candidat ...
3. **Title:** Catherine Trautmann (SOC) reacts to the results of the first round
- **Link:** [https://www.youtube.com/watch?v=8utcn4jHaA4](https://www.youtube.com/watch?v=8utcn4jHaA4)
- **Snippet:** Municipales à Strasbourg : Catherine Trautmann (SOC) réagit aux résultats du 1er tour.
4. **Title:** Catherine Trautmann - Wikipedia
- **Link:** [https://en.wikipedia.org/wiki/Catherine_Trautmann](https://en.wikipedia.org/wiki/Catherine_Trautmann)
- **Snippet:** Catherine Trautmann (born 15 January 1951) is a French politician for the French Socialist Party. She served as Minister of Culture of France in the Lionel ...
5. **Title:** 21 Photos & High Res Pictures - Getty Images
- **Link:** [https://www.gettyimages.com/editorial-images/news/event/strasbourg-mayoral-candidate-catherine-trautmann-speaks-at-campaign-rally-ahead-of-municipal-runoff/776475585?editorialproducts=all](https://www.gettyimages.com/editorial-images/news/event/strasbourg-mayoral-candidate-catherine-trautmann-speaks-at-campaign-rally-ahead-of-municipal-runoff/776475585?editorialproducts=all)
- **Snippet:** FRA: Strasbourg Mayoral Candidate Catherine Trautmann Speaks At Campaign Rally Ahead Of Municipal Runoff. 21 Photos and High-Res Pictures.
</background>
Recall the question you are forecasting:
Will Catherine Trautmann be the next mayor of Strasbourg?
Instructions:
1. Compress key factual information from the sources, as well as useful background information
which may not be in the sources, into a list of core factual points to reference. Aim for
information which is specific, relevant, and covers the core considerations you'll use to make
your forecast. For this step, do not draw any conclusions about how a fact will influence your
answer or forecast. Place this section of your response in <facts></facts> tags.
2. Provide a few reasons why the answer might be no. Rate the strength of each reason on a
scale of 1-10. Use <no></no> tags.
3. Provide a few reasons why the answer might be yes. Rate the strength of each reason on a
scale of 1-10. Use <yes></yes> tags.
4. Aggregate your considerations. Do not summarize or repeat previous points; instead,
investigate how the competing factors and mechanisms interact and weigh against each other.
Factorize your thinking across (exhaustive, mutually exclusive) cases if and only if it would be
beneficial to your reasoning. We have detected that you overestimate world conflict, drama,
violence, and crises due to news' negativity bias, which doesn't necessarily represent overall
trends or base rates. Similarly, we also have detected you overestimate dramatic, shocking,
or emotionally charged news due to news' sensationalism bias. Therefore adjust for news'
negativity bias and sensationalism bias by considering reasons to why your provided sources
might be biased or exaggerated. Think like a superforecaster. Use <thinking></thinking> tags
for this section of your response.
5. Output an initial probability (prediction) as a single number between 0 and 1 given steps 1-4.
Use <tentative></tentative> tags.
6. Reflect on your answer, performing sanity checks and mentioning any additional knowledge
or background information which may be relevant. Check for over/underconfidence, improper
treatment of conjunctive or disjunctive conditions (only if applicable), and other forecasting
biases when reviewing your reasoning. Consider priors/base rates, and the extent to which
case-specific information justifies the deviation between your tentative forecast and the prior.
Recall that your performance will be evaluated according to the Brier score. Be precise with tail
probabilities. Leverage your intuitions, but never change your forecast for the sake of modesty
or balance alone. Finally, aggregate all of your previous reasoning and highlight key factors
that inform your final forecast. Use <thinking></thinking> tags for this portion of your response.
7. Output your final prediction (a number between 0 and 1 with an asterisk at the beginning and
end of the decimal) in <answer></answer> tags.
OUTPUT_FORMAT
* Your output response must be only a single JSON object to be parsed by Python's "json.loads()".
* The JSON must contain four fields: "p_yes", "p_no", "confidence", and "info_utility".
* Each item in the JSON must have a value between 0 and 1.
- "p_yes": Estimated probability that the event in the "Question" occurs.
- "p_no": Estimated probability that the event in the "Question" does not occur.
- "confidence": A value between 0 and 1 indicating the confidence in the prediction. 0 indicates lowest
confidence value; 1 maximum confidence value.
- "info_utility": Utility of the information provided in "sources" to help you make the prediction.
0 indicates lowest utility; 1 maximum utility.
* The sum of "p_yes" and "p_no" must equal 1.
* Output only the JSON object. Do not include any other contents in your response.
* This is incorrect:"```json{
"p_yes": 0.2,
"p_no": 0.8,
"confidence": 0.7,
"info_utility": 0.5
}```"
* This is incorrect:```json"{
"p_yes": 0.2,
"p_no": 0.8,
"confidence": 0.7,
"info_utility": 0.5
}"```
* This is correct:"{
"p_yes": 0.2,
"p_no": 0.8,
"confidence": 0.7,
"info_utility": 0.5
}"
<think>
Let me analyze this question carefully.
The question is: Will Catherine Trautmann be the next mayor of Strasbourg?
Today's date: 21/03/2026
My pretraining knowledge cutoff: October 2023
Key facts from the sources:
1. Catherine Trautmann (SOC - Socialist) leads the first round of municipal elections in Strasbourg
2. Four candidates are able to stand for the second round
3. Outgoing mayor Jeanne Barseghian comes third
4. The Greens (Barseghian's party) are being challenged
5. There's a municipal runoff ahead
6. Catherine Trautmann is a French Socialist Party politician, former Minister of Culture
Let me think about this:
- The first round has already taken place (the sources reference "first round" results and "municipal runoff" ahead)
- Catherine Trautmann leads the first round
- Four candidates are in the second round
- The outgoing mayor Jeanne Barseghian (Green) comes third
- The question is whether she will be the NEXT mayor
Key considerations:
- Leading the first round is a strong position but doesn't guarantee winning the second round
- In French municipal elections, the second round is where the real contest happens
- Four candidates in the second round means it's a multi-candidate race
- The Socialist candidate leading suggests a shift from the Green mayor
- However, in French municipal elections, coalitions and alliances in the second round can change the outcome
- The fact that Barseghian is third (not second) suggests the Green vote might be split or weakened
- But with four candidates, the dynamics are complex
Let me think about French municipal election dynamics:
- In the second round, candidates can withdraw in favor of others
- Alliances between parties are common
- The leading candidate in the first round has a significant advantage but it's not guaranteed
- With four candidates, the vote could be split in the second round
The fact that she leads the first round is a strong positive signal. In French municipal elections, the first-round leader wins a significant majority of the time in the second round, especially when the field is narrowed. However, with four candidates still in the race, there's more uncertainty.
The date is 21/03/2026. French municipal elections in 2026 - the first round would typically be in March, and the second round two weeks later. So the second round would be around early April 2026. The question is asking about the outcome.
Given that:
- She leads the first round (strong signal)
- Four candidates in second round (more uncertainty)
- The outgoing mayor is third (weakened position)
- Socialist vs Green dynamics in Strasbourg
I'd estimate her probability of winning at around 55-65%. Leading the first round in a French municipal election is a strong advantage, but with four candidates in the second round, there's meaningful uncertainty. The fact that the Green mayor is only third suggests the Green vote might be fragmented or that other candidates are strong.
Actually, let me reconsider. In French municipal elections:
- The first round leader wins about 70-80% of the time in the second round
- But with four candidates, the dynamics are different
- Alliances in the second round can shift things
- The Socialist candidate leading suggests a potential left-wing realignment
I'll estimate around 60% probability. She's the favorite but not a lock.
Let me settle on 0.60 for p_yes.
Confidence: moderate - we have first-round results but the second round hasn't happened yet. I'd say 0.6 confidence.
Info utility: The sources are quite useful - they tell us she leads the first round, four candidates are in the second round, and the outgoing mayor is third. This is very relevant. I'd say 0.75.
</think>
{
"p_yes": 0.60,
"p_no": 0.40,
"confidence": 0.60,
"info_utility": 0.75
}The model is published on Hugging Face as a 319 MB adapter for the open-weight Qwen3.8-27B base model, both under the Apache License 2.0. The weights take about 54 GB of GPU memory. The evaluation ran on a single H200. An 80 GB GPU such as an H100 or A100 is expected to work at lower concurrency.
The prompt must include the evidence for the question. The model does not retrieve evidence itself.
Request a forecast