Ticker League

Smart Consensus: accuracy-weighted crowd forecasts

How a crowd forecast can beat a simple average by weighting each contributor by their difficulty-adjusted track record.

The core idea

A simple crowd average treats every forecaster equally — the analyst who has been right for years counts the same as someone predicting for the first time. Smart Consensus instead weights each contributor by how accurate they have actually been, so proven forecasters move the number more than newcomers or chronically-off predictors.

How the weighting works

Each user's weight is derived from their difficulty-adjusted mean error — being accurate on a hard, thinly-covered name counts for more than nailing an easy, heavily-covered one. The weighted mean is Σ(wᵢ · forecastᵢ) ÷ Σ(wᵢ), with weights clamped to a sensible range and any single contributor's influence capped so one account can't dominate.

Minimum history and privacy

Weighting only activates once a user has enough resolved estimates to judge them — below that threshold everyone gets a neutral weight (a plain average). Community figures are also only shown once enough independent forecasters have contributed and their answers actually differ, so no individual's prediction can be reverse-engineered.

See how the consensus estimate is built in the Earnings Season course lesson on how the consensus estimate is set.

Read the exact formula in the Smart Consensus methodology, or see where the crowd most diverges from Wall Street on the crowd-vs-Street board.