The stake outcomes graph shows how many of EQCs the user staked in each prediction.

• 'To occur' and 'Not to occur' predictions are shown as seperate stacked bars.
• The color of the bars shows the outcome: true, false or open.

The distribution of stakes map shows the locations of the user's predictions.

• The shading indicates the amount of staked EQCs: darker shades for many, lighter shades for fewer EQCs.
• 'To occur' predictions are shown as up arrows, 'Not to occur' as down arrows.

There are three rewards categories:
1) Initial 125€ shared among all users with a positive score and skill class C 
2) Additional 50€ for users with skill class B  and at least 5 non-overlapping predictions.
3) Additional 75€ for users with skill class A  and at least 5 non-overlapping predictions.
Skill classes of users with open predictions are shown as A, B, C or D because they may change.

Rewards increase with the number of active users. 1€ is added to the rewards for each user submitting a prediction that closes during the round. This sum is distributed in the following categories:

1) 50% of the rewards are shared among all users with a positive score and skill class C 
2) 20% is shared among users with skill class B  and at least 5 non-overlapping predictions.
3) The remaining 30% is shared among users with skill class A  and at least 5 non-overlapping predictions.
Skill classes of users with open predictions are shown as A, B, C or D because they may change.

Information ratio (IR) is defined as Success rate(fraction of true predictions among all predictions) divided by the Average RichterX Probability (mean of forecast probabilities assigned to all predictions).
Users on the diagonal line (IR=1) perform as good as the RichterX model expects, steeper lines mean higher IR values.
Users with less than 5 predictions are shown as , those with 5 or more are shown as
Skill classes are determined by information ratio and significance defined as (100% - p-value). P-value is the probability of obtaining a larger or equal IR score by random chance assuming that the RichterX model is correct.