Earthquake risk at your fingertips

Inform the public about short-term earthquake probabilities anywhere on the globe in real-time.
Invite anyone to participate in the challenge of earthquake prediction.
Empower the media to counter fake prediction news after an earthquake happens.

RichterX organizes a global earthquake prediction competition to find out if earthquakes are predictable.
Anyone can participate in this competition by submitting predictions. These predictions are evaluated using our real-time global earthquake forecasting model.
An earthquake prediction is defined as a statement specifying the number of earthquakes above a threshold magnitude, in a circular region and within a time-window.

To keep the competition fair, each user is allowed to have only one account. If multiple accounts are detected, we will deactivate all except the initial account and none of the accounts will be eligible for rewards in that round.
An earthquake prediction can be submitted by staking 1 to 100 Earthquake Coins (EQC). The prediction time-window starts 10 minutes after submission.
The EQC balance recharges in real-time at a rate of 1 EQC every hour. Thus, if all EQCs are staked a new prediction can be submitted after an hour. The total active stake on Not to occur predictions can not exceed 100 EQC.
Predictions are evaluated every hour against new earthquakes from the actual USGS ComCat catalog to determine current scores and tentative rewards.
One hour after the end of each round USGS ComCat catalog is queried and all predictions are re-evaluated to determine final scores and final rewards.
Score is defined as [Gained EQC] minus [Staked EQC] minus [Carry over EQC].
Users with negative final scores carry over a portion of their negative scores to the next round. The carry over is fixed at 10% for scores up to -25 and increases linearly to 90% for scores of -225.

This is an ongoing experiment and thus we reserve the right to modify and update the rules as we gain further insights.

Public participation helps us improve our model by highlighting its weaknesses. More participation means more data, which means more reliable conclusions.
That's why every two weeks we distribute a reward increasing with the number of active users. The reward is increased by 1€ for each user submitting a prediction that closes during the round.
Success is defined in skill categories: We distribute 50% of the reward to all users with a positive score and a skill class C, 20% of the reward to users with skill class B. Finally, the remaining 30% is shared among users with skill class A.
Rewards in all categories are distributed proportionally to the users scores. Skill class A and B require at least 5 non-overlapping predictions
The top 3 participants donate 10% of their rewards by default to support earthquake relief efforts.
All participants can choose to donate their rewards to charitable organizations or withdraw them as cryptocurrency or gift cards. For withdrawals we use Bitcoin Cash (BCH) and Gridcoin (GRC), a cryptocurrency incentivizing volunteer computing for scientific research.