> For the complete documentation index, see [llms.txt](https://www.degenbot.wiki/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://www.degenbot.wiki/understanding-alerts/consensus.md).

# Consensus

Understand agreement between signals without treating agreement as certainty.

**Consensus** is a reported alert stream; **v4** is a previously reported version label. In general, a consensus approach combines or compares multiple pieces of evidence rather than relying on one input alone.

The exact inputs, weights, thresholds, and timeframe combination for Degen's active version must be verified against the deployed method. This guide does not invent those details.

### What agreement can tell you

When an alert explains that several inputs support a direction, read which inputs it actually names. Agreement can make the explanation easier to follow, but related inputs may be responding to the same underlying price movement.

For example, two different moving averages can both reflect the same recent trend. Counting them as two signals does not automatically make them two independent pieces of evidence.

### How to read a Consensus alert

Check the market, direction, version, and horizon. Then read the stated supporting evidence and any conflicting evidence or invalidation condition.

A Consensus alert can differ from a short-term alert because the methods may use different conditions or timeframes. Compare the intended horizon before deciding that either message is wrong.

### A score is not automatically a probability

A score of 80 may mean a weighted rule total or model output. It does not mean an 80% chance of success unless the published methodology demonstrates that interpretation.

A historical 80% resolution rate also does not assign an 80% probability to a particular new message. The sample, market conditions, and method version matter.

### Use the explanation, not just the label

Ask: What agrees? What does not? How current is the evidence? What ends the scenario? If the message does not provide the information you need, it is reasonable to leave it as an observation rather than act on it.

See **Performance and methodology → How the analysis works** for a plain-language explanation of scoring and machine-learning limits.