PolytoolhubWhat is Eversight and Signals APIs?
Eversight and Signals serve two different stages of research. Eversight runs an asynchronous forecasting job and returns a probability. Signals retrieves evidence associated with a market or question, including relevance-weighted material and structured market context. They can feed a research assistant without placing orders.
Who it is for
- Developers building forecasting and research assistants
- Analysts comparing model probabilities with market prices
- Teams that need traceable evidence retrieval
Features and coverage
- Available on
- REST API, Web
- Markets & venues
- Polymarket, Kalshi
- Forecast jobs
Submit a question, poll the job state and read the completed probability.
- Market-aware signals
Supply a market URL, slug, condition ID or free-text question.
- Source attribution
Inspect the material behind weighted signals.
- Historical evidence retrieval
Use as_of when a research workflow needs only information available at a past point.
- News impact
Read supported scored news signals and their direction.
- Related-market context
Explore causal market relationships and associated evidence.
- Numinous-1 model
Use the specialized forecasting model through its OpenAI-compatible endpoint and structured-output support.
Workflows with Eversight and Signals APIs
Compare a forecast with a contract
Read the exact resolution question, submit the forecasting job and wait for completion. Compare the returned probability with the market quote, retaining the evidence used.
Build a source-backed research panel
Query Signals for the contract, display linked evidence and distinguish publication time from snapshot time.
Replay historical research
Fix as_of before retrieving source material so later articles do not enter an earlier decision window.
Pricing and total cost
The APIs use metered access through supported API-key or x402 payment paths. Forecast jobs, signal retrieval and model inference are separate operations; inspect their current rate schedule before running a batch.
Getting started
Before you start: Forecast jobs are asynchronous; submitting a request does not immediately produce a finished result.
- Choose Eversight for a forecast job or Signals for evidence retrieval.
- Configure the documented API-key or x402 path.
- Start with one precise question or supported market identifier.
- For a forecast, poll until COMPLETED before reading result.prediction.
- Store source references and the query’s time boundary.
- For model inference, handle truncated output and structured-response parsing.
Useful links
What to check
- Forecast jobs are asynchronous; submitting a request does not immediately produce a finished result.
- Numinous-1 is a forecasting model, not a general-purpose assistant.
- Use either event_id or condition_id where the endpoint marks them mutually exclusive.
- A published_at time and the time a source was captured can differ.
- The model’s documented context window is 16,384 tokens; check finish_reason when output is cut short.
A closer look
Ratings & reviews
User reviews 0
The main score switches to user ratings after 5 published reviews.
A specific example helps the next person choose the right tool.
Write a review
How was your experience with Eversight and Signals APIs?
Share what you used, what worked and what could be better. We confirm your email and review the submission before publishing.
To submit a review, email info@polytradingbot.com with the tool name, your rating and your experience. Enable JavaScript to prepare a draft here.
Related tools
Browse category
Add Numinous forecasts, sourced market signals and credit-usage checks to a compatible coding or research agent.
- Language Python
Turn a market link, question or screenshot into an AI forecast and compare it with market odds and news context.
- Access Web
Start an AI research task from a Polymarket or Kalshi market URL using a Valyu-connected forecasting interface.
- Access Web
Frequently asked questions about Eversight and Signals APIs
Which endpoint should I use for supporting articles?
Use Signals and its retrieval endpoints.
How is the forecast expressed?
The completed job returns result.prediction on a zero-to-one scale.
Can I limit research to what was known earlier?
Use the documented as_of retrieval parameter.
Does it execute a trade?
These APIs provide research and probabilities; execution is a separate integration.