Numeraire overview using the Feynman Technique
This post is one of a series of posts I am writing where I try to apply the Feynman Technique to a number of different crypto token projects. You can find the rest here. None of this should be taken as investment advice. They are just meant to be simple introductions to the projects.
Numeraire is the token used on the Numerai platform. Numerai is a platform that hosts machine learning competitions between data scientists from around the world. The Numeraire is being used on the Numerai platform to better understand each data scientist’s confidence level in their machine learning model.
Each player in a Numerai competition has the ability to bet a certain amount of Numeraire on whether or not their model will meet a certain threshold of accuracy. If their model does well in the competition, they keep their Numeraire and receive a given dollar amount as a reward.
This reward amount is determined by taking the amount of Numeraire they bet and dividing it by the players’s confidence level (a range of 0–1). If their model does not meet the minimum performance required, they lose all of the Numeraire they bet and don’t receive any award.
In order to prevent situations where a player is motivated to bet a lot of Numeraire but falsely indicates a low confidence level, to maximize their returns, Numerai pays out rewards starting with the player who indicates the highest confidence level. Each player’s model that meets the performance level required gets rewarded until the total reward amount is 0.
Therefore, if a player indicates a low confidence level, it is likely they will not receive a reward since other players with higher confidence levels will be able to receive it before them. This should motivate players to indicate accurate confidence levels.
Further reading:
