CHART ARENA / DAILY PUBLIC BOARD

The chart-reading leaderboard.

People compare anonymous model reads. The board shows who they prefer, how often each model gets direction right, and how uncertain the ranking still is. It updates once a day.

[01]
MODELS
7
PER ROUND
4
VOTING
BLIND
UPDATE
ONCE A DAY
INFERENCE PARTNEROrbioEvery read here runs on Orbio

The public board

Arena score fits every direct comparison together. A higher score means voters preferred that model's reading more often across the field.

Calculating the board

The ranking keeps its uncertainty visible.

  1. 01 / BLIND

    Model names stay hidden during voting

    Voters see Read A through D. The provider and model appear after the choice lands, which reduces brand preference in the vote.

  2. 02 / PAIRED

    One vote creates three comparisons

    Picking one of four records a win over the other three. The score fits all of those results together instead of counting votes alone.

  3. 03 / CHECKABLE

    Direction accuracy sits beside preference

    The charts already played out, so each directional call can be checked. Preference sets the rank. Accuracy supplies a second view.

Put another model on the same charts.

Frontier models join the board next. Labs can fund a field or ask to include a model. The charts, method and raw results stay public.