---
pretty_name: ForecastBench Single Questions
license: cc-by-sa-4.0
tags:
- forecasting
- calibration
- real-world forecasts
- human forecasts
- cc-by-sa-4.0
task_categories:
- question-answering
- table-question-answering
- text-generation
language:
- en
configs:
- config_name: forecastbench_single_questions_2024-12-08
  data_files: forecastbench_single_questions_2024-12-08.jsonl
- config_name: forecastbench_single_questions_human_2024-07-21
  data_files: forecastbench_single_questions_human_2024-07-21.jsonl
size_categories:
- n<1K
---


# ForecastBench Single Questions

This dataset contains single-ID forecasting questions derived from the [ForecastBench project](https://github.com/forecastingresearch/forecastbench-datasets). It includes two configurations:

- `forecastbench_single_questions_2024-12-08`: Contains 429 forecasting questions with resolved real-world outcomes.
- `forecastbench_single_questions_human_2024-07-21`: Contains 473 questions with resolved real-world outcomes, augmented with human forecast probabilities from public and superforecaster groups.

## Dataset Structure

Each entry in the dataset is a JSON object with the following fields:

- `id`: Unique identifier for the question.
- `question`: The forecasting question text.
- `background`: Additional context or background information.
- `source`: Origin of the question (e.g., "manifold").
- `source_intro`: A standardized prompt prefix shown to forecasters, e.g."We would like you to predict the outcome of a prediction market. A prediction market, in this context, is the aggregate of predictions submitted by users on the website Manifold. You’re going to predict the probability that the market will resolve as ‘Yes’."
- `answer`: Binary outcome (0 or 1) representing the resolved real-world result.
- `human_public_forecast`: (Optional) Forecast probability made by the public human group.
- `human_super_forecast`: (Optional) Forecast probability made by the superforecaster (expert) human group.

## Usage

To load a specific configuration using the Hugging Face `datasets` library:

```python
from datasets import load_dataset

# Load the base dataset without human forecasts
dataset = load_dataset("Duruo/forecastbench-single_question", name="forecastbench_single_questions_2024-12-08")

# Load the dataset with human forecasts
dataset_with_human = load_dataset("Duruo/forecastbench-single_question", name="forecastbench_single_questions_human_2024-07-21")
```

## License and Attribution

This dataset is released under the [Creative Commons Attribution-ShareAlike 4.0 International License (CC BY-SA 4.0)](https://creativecommons.org/licenses/by-sa/4.0/).

It is based on the original [ForecastBench dataset](https://github.com/forecastingresearch/forecastbench-datasets), which is also licensed under CC BY-SA 4.0.

If you use this dataset, please cite the original ForecastBench project and provide appropriate attribution.

## Citation

```
@inproceedings{karger2025forecastbench,
      title={ForecastBench: A Dynamic Benchmark of AI Forecasting Capabilities},
      author={Ezra Karger and Houtan Bastani and Chen Yueh-Han and Zachary Jacobs and Danny Halawi and Fred Zhang and Philip E. Tetlock},
      year={2025},
      booktitle={International Conference on Learning Representations (ICLR)},
      url={https://iclr.cc/virtual/2025/poster/28507}
}
```

```
:contentReference[oaicite:21]{index=21}

---

:contentReference[oaicite:23]{index=23} :contentReference[oaicite:26]{index=26}:contentReference[oaicite:28]{index=28}
::contentReference[oaicite:29]{index=29}
 
```