Improvement to payout factor

Based on the idea of a Progressive Tax, I think the payout factor could be improved for smaller stakes, on average the payout factor could remain the same as is, so that Numerai would still pay the usual amounts based on current payout factor, but it could have different values for each individual stakes. I didn’t investigate the numbers but maybe slightly reducing the payout factor as stakes get bigger, it could greatly improve payouts for small stakes. The main idea is that whales will still have huge payouts but it will improve payouts for the other participants.

A regressive payout factor could be like:

stakes pf multiplier
< 1 NMR 400%
< 10 NMR 200%
< 100 NMR 100%
< 1000 NMR 95%
> 1000 NMR 90%

The actual tiers and multipliers could be determined based on historic data or stakes and/or performances of each individual rounds.

Not Sybil resistant. By splitting stakes into smaller amounts and distributing them across multiple accounts, one could get nice 400% profit with just ten lines of code with no benefits to the system.


I truly hate the idea. As communism survivor I am allergic to anything smelling of socialist nonsense.

“Whales” will not have “huge payouts” under your scheme. Percentagewise it will be meager and you are penalizing them for what and why? Why help smaller stakes? This is skin in the game tournament and if you do not believe in your models, you should not be extra rewarded. Only merit counts. Or should count.

… and as @ssh points out it just opens the way for free riders and parasites who know only one thing: to game the system for their advantage. This is disastrous for any society. Believe me. I lived thru it.


@svendaj Totally agree. It makes zero sense to favor small stakes over high stakes.

Well, “atol_m12” had a payout close to 1550 USD on round 679 and at 90% that’s still 1400 USD, in many countries that’s like 3 or 4 month’s wages, and I think that’s a huge payout

Also based on Progressive Tax, it is known that flat tax rates increase inequality so what’s the point of huge payouts to whales? For them to have even bigger stakes? That reduces the payout factor and payouts for everybody and, from what I understand, their contributions to the meta-model are already capped at 1%. And about their percentage, like I mentioned before, it could be analyzed what’s the real additional contribution of whales’ predictions against the rest to maybe get to a proper percentage, 90% is an example

About “free riders”… they already receive 100% payout, right? so changing to 90% or 1000%, payouts and burns will increase accordingly, free riders remain free riders

On that note, there are many reasons to help smaller stakes

  • mainly, increasing stakes of good models so that their contribution to the meta-model increases faster
  • people with good ideas but no money, like discussed in the last fireside chat
  • many countries, crypto is really expensive which causes smaller stakes
  • many countries, NMR is not even available on local exchanges which requires extra hoops and paying more transactions fees, which further decreases stakes