For those who are not familiar with this project, this is the community marketplace originally proposed here and is funded by the Numerai Council of Elders.
This is an interim early release with support for generic 3rd party listing of Numerai predictions, models, Signals predictions and Signals data. I would like to invite everyone (especially sellers who are currently listing their models on Gumroad) to try it out. You can get you models listed on NumerBay for free, so why not?
In the meantime, I am going to continue to work on the other various features of NumerBay, most prominent of which are on-platform listings and NMR payment support.
Please vote on features that you would most like to see in the next major release. I would prioritize features that are most in demand. Poll closes on August 20.
If you have features in mind that are not in the poll, or if you encounter any issue using NumerBay, please leave your comments here or post in the #numerbay channel on RocketChat.
Thank you and happy sailing on NumerBay
General - Sybil Robust Voting App / Numerai Analytics Dashboards / More community content on Homepage / etc.
Buyer - UX - Sort Model Listings by Numerai Rank / Rep
@nyuton Thanks for the suggestions. Both of those are in the pipeline. It is not as straightforward to implement as it seems. There are other functionalities that I need to finish before this is possible.
My subscribers often use the purchesed predictions for ensembling.
For this purpuse it would be useful to display the correlation with the metamodel on the model page.
Low correlation has higher value for the buyer.
MMC is also useful. Payout can be 2xMMC.
Sorting should be done by default on CORR+2xMMC
This puts the highest value model first, instead of the highest CORR ranking model first.
Multiple sorting and filtering metrics are coming in the next feature release (should be some time next week or earliest this Sunday). However, default sorting will be by rank. Metamodel correlation can be added after that. Thanks for the suggestions.
Instead of buyers getting prediction files from sellers, let me propose a completely different model: 3rd party staking.
The 1st party is the model maker, the 2nd party is Numerai. The 3rd party staker would have a Numerai wallet, from which he would stake NMR on the 1st party’s model. Instead of getting the prediction file, the 3rd party gets the same earns/burns (in proportion to stake amount) as the 1st party on that model.
This would require some Numerai-provided structure, for automating the eventual unstaking, and then distributing the proceeds according to the terms agreed upon.