Submitting Forecasts

Once a forecasting question is active, forecasters can submit probability estimates. Hinsley supports forecasts from registered users, guest forecasters invited by email, and AI models. All forecasts are aggregated into a consensus view that updates in real time.

How to Submit a Forecast

Navigate to a forecasting question and click the forecast link to open the forecasting form. The form displays the question text, background information, and resolution criteria to help you make an informed estimate.

Binary Questions

For binary (Yes/No) questions, you see two probability inputs - one for Yes and one for No. Adjusting one automatically updates the other so the two always sum to 100%. You can type a number directly or use the range slider.

Multinomial Questions

For multinomial questions, you assign a probability to each answer option. All probabilities must sum to 100%. If your entries do not add up, use the “Normalize to 100” button to proportionally adjust all values.

Rationale

You can optionally provide a written rationale explaining your reasoning. Rationales help other forecasters understand different perspectives and can improve the quality of the overall forecast.

Who Can Forecast

Forecasting questions accept contributions from several types of forecasters:

  • Registered users - Anyone with a Hinsley account who has access to the analysis or question
  • Guest forecasters - People invited by email who submit forecasts without logging in. They provide their name and email address with their forecast.
  • AI forecasters - Hinsley’s AI models, which generate forecasts automatically on a configured schedule or on demand

If a question has been published, anyone with the link can view and submit forecasts without authentication.

Viewing Results

The question detail page provides several views for understanding the current state of forecasting:

Summary Tab

The summary tab shows the aggregated consensus forecast - the combined probability estimate across all forecasters. It includes:

  • Consensus probability - The current aggregated estimate for each answer
  • Trend graph - How the consensus has changed over time as new forecasts are submitted
  • Distribution chart - A histogram showing how individual forecasters’ estimates are spread
  • Aggregate rationale - An AI-generated explanation of the consensus, including a bottom-line-up-front summary, base rate information, key arguments for and against, and a summary of recent changes

Current Forecasts Tab

Shows the most recent forecast from each forecaster. This gives you a snapshot of where each participant currently stands.

Forecast History Tab

Shows all historical forecasts with timestamps, letting you see how individual forecasters have updated their estimates over time.

Sources Tab

Displays any source documents connected to the question, providing context for the forecasting topic.

AI-Generated Forecasts

When AI forecasting is enabled for a question, Hinsley’s LLM models periodically generate their own probability estimates. AI forecasts follow the same format as human forecasts - they assign probabilities to each answer option and include a detailed rationale.

AI forecasts are generated asynchronously. When a generation is in progress, the question page shows a loading indicator. Each AI forecast rationale includes:

  • The Bottom Line - A brief summary of the AI’s assessment
  • Base rate information - Historical context relevant to the question
  • Arguments for and against - Key factors supporting and challenging the estimate
  • Recent changes - What has changed since the previous AI forecast

AI forecasts are included in the aggregated consensus alongside human forecasts. The aggregation method can be configured to weight different forecaster types differently.