Numeraire - worst part of NUMERAI - time to change

I was never a fan of crypto. So, when I started with NUMERAI about half a year ago I had my opinion, but I didn’t care, because I wanted to improve in ML for stock trading and prove myself in “the hardest data science tournament in the world”.

Now in the beginning of year 2023 (BTW all the best for all of you, dear colleagues :partying_face: :fireworks: :sparkler: :confetti_ball: :tada: :clinking_glasses:) when my portfolio of models has 3-mo yield 56.6% and my best model is around ww #10 in 3-mo yield, I would like to reap the result of my work and increase my dedication to NUMERAI research.

But to do that, I have to get into “the steaming pile of horse manure that is crypto” and become part of what Taleb calls the “cluster”, which brings together crypto fans, Covid-19 skeptics, climate skeptics, Putinists, and radical carnivores.

Last year brought yet another way to lose an already very risky stake in NMR: crypto exchanges would not pay you out in fiat money. This seems to me unfair: what kind of NMR risk has NUMERAI to reciprocate risk of data scientists who could lose all their stake (in real value)?

Isn’t it time to switch from NMR staking to staking in USD?

USD staking could do everything what NMR staking can do and potentially even more: it could tie performance of models with fund performance and thus fully align incentives. USD stakes would buy the same stocks portfolio as NUMERAI fund (management would decide which one) and portfolio net gains/losses would be recalculated as payout factor. The stake of underperforming models would not be “burned”, but proportionally reallocated to performing models.

What do you think?


I also think crypto is a steaming pile of garbage, and would much rather stake in USD than NMR or any other crypto.

That said, such a transition seems unlikely to me. I don’t want to make numerai’s case for them, but it’s worth considering the massive amount of value that exists in the sum of extant NMR, that created no cost-basis for numerai, and generated immense store of value for them. To transition to USD, it seems they would need to abandon NMR, a peril of it’s own (affecting stakers and hodlers), and introduce staking mechanisms that may be infeasible and certainly would cost them a great deal of money for them to implement.

I think Numerai wants to be a world-breaking, radical organization, and such changes are possible for an organization that lives up to that goal. I leave the challenge to do the right thing to them <3

In the mean time, I’m content to engage in a hustle that has made me money, without risking my livelihood. Hustle on, hustlers.


Our ultimate goal should be to stake on the US dollar. In the meantime, we can consider staking on stable coins like USDT as a way to gradually reach that goal.

1 Like

I am 100% with you. However, I think the legal hurdles are too bothersome for Numerai. I think it would also slow down their development as well, since every change to the tournament would also imply legal consequences.

It would also have to be implemented in a way where Numerai as a company does not have to spend significant amounts of money per round. For example all stakers “bet” their money in a pool, and everyone would be paid out (or burned) of that pool based on their performance, leaving net zero money in the pool every round. Maybe Numerai could throw in some money into that pool to incentivize participation.

In the end, from a price-value viewpoint the crypto solution is probably way easier/cheaper for Numerai.

1 Like

If they switch to a stable coin like USDT then legally this should not change anything because it is still a crypto currency just as NMR.
The primary reason they are using NMR instead of USD (or USDT): It is free money for them because they created it.


This is what I meant: what you call pool would be invested into stocks and thus it would be (presumably) gaining the value for data scientists. This also means that NUMERAI would not need to add any incentive of their own (although it would be highly appreciated ;-).

And to your last remark: yes, NMR is advantageous for NUMERAI, but very risky for data scientists. I call it unfair.


NMR is vital for Numerai, because they created “value” out of thin air and allowed them to pay out hugh sums. That would have been not possible with USD. They simply didn’t have that much.

Once the fund reaches some reasonable AUM, switching to USD will be an option. They will earn enough USD to pay out data scientists.

However, if the fund reaches high AUM, keeps performing well and establishes some reputation, you will be happy with your NMR.
Imagine Renaissance offering some crypto to join the fund. Would you join? I guess all of us would! Any “wannabe” trader/data scientist/investor/crypto-fan/idiot/influencer would want to prove themself in the most famous open hedge fund. That will drive NMR up!

Assuming the fund becomes a success, (Which seems very likely at the moment. Check out here: Investor Letter) NMR bet is safe. Even a good investment.


Baring a major internal problem with numerai occurring, converting the current crypto staking model into a fiat staking one is completely off the table as a goal. The goals of numerai have changed significantly since its original inception and weening off crypto would be a major step backwards from those goals as described here (It’s worth giving a read if want to understand the history of numerai and how different things used to be)

The fact is that Richard’s ambitions with numerai are much higher than just having the best hedge fund, and achieving those ambitions is going to require staying in the crypto space unless another alternative comes out to realize them.

Another thing worth understanding given your screed against crypto. Every socially disruptive technological innovation that has happened throughout history has coincided with the incidence of grifters exploiting the general populace’s lack of understanding regarding the new technology. This happened with the internet in the 90s, with telecommunications in the early 20th century, with oil prospecting in the late 1800s, with railroads in the mid 1800s and even as far back as the early 1700s with international shipping. Whenever you work with something that is on the frontier of innovation, you be working with something that is simultaneously also being used for unsavory purposes. That’s just the reality of things.

1 Like

Well, here we go again. NMR is down (again) and everybody cries again. I am in this competition already for 6 years and this just keeps following the pattern. Funny that nobody complains whenever NMR reaches the ATH (again), and nobody complains whenever ppl here literally make millions of dollars in profit (again!) by combining their model performance with selling NMR at good prices with the right timing. Some ppl just don’t understand how it works, the cycles, they just buy at the ATH, and then they cry.

I understand that frustration (kind of), modelling the market is very hard and then you have to care about exposure to “bad crypto eww”. But you guys have to realize something - without NMR released as a token, this competition would never pay so much to data scientists. Ever. Not possible. There are too many participants and the fund is very small. Do you really want to have $30k in the prize pool per round? To divide between all models? This would be a realistic option related to the fund’s AUM. Honestly, I don’t.

Also, without using “crypto” for payouts it would never be possible to run global staking competition for the whole world. How would you do it with USD? With wire transfers? Paypal? The compliance burden and all the KYC nonsense would just bury the project at the very beginning. I hope that the crypto haters in this thread could see this too. Therefore I sincerely hope the NMR token with all the “crypto” aspects will stay for as long as possible!

Sure, the volatility of NMR is a problem. You have to time the buys (or wait for accumulation ranges) and sell profits using limit orders at their peaks. For me at least is this still a cost & risk I am willing to swallow. An alternate solution (payouts in USD) means negligible payouts. If this ever happens, I am gone.

For the crypto haters of this thread - I kinda understand some of your arguments (I agree that unfortunately there is a lot of scam and poor projects in the industry). Honestly, I only find BTC, XMR, ETH and NMR interesting. As much as I respect Mr. Taleb (his Black Swan book is just great), I believe he is plain wrong here (and I won’t comment further on his retarded “clustering analysis” based on a faulty generalization). Good and honest crypto projects can have value demonstrated by this very project here (I mean And I myself believe that honest “crypto” projects with prevail, Numeraire included. I accumulate / trade NMR and so it does certain private fund I am cooperating with because we believe the current NMR price is already well suited for accumulation with a long-term horizon in mind.

And if you really still believe in Mr. Taleb’s argument that the real value of any “crypto” is 0 (zero), I have a last question for you - why don’t you just short the crypto to have some skin in this argument? That would also solve your issue with exposure to crypto while buying NMR for staking (they are strongly correlated). Then you are (at least partially) hedged and you don’t need to cry in the forum anymore worry about your exposure to crypto that much and watch your model grow your precious capital.

(Yeah, I wanted to sound a bit unkind in the last paragraph (could not help it, sorry guys), but really shorting BTC / ETH is serious advice on how to at least partially mitigate the exposure to crypto if you don’t want to worry about timing NMR buys/sells.)


Read here:

It’s all about the network effect!
This is an article from 6 years ago.
Seems like, we are getting there…

1 Like

People also cry when the NMR price is high because then it is risky to get into the tournament. The current period is a great time to get in though.


Either I miss something or people miss something. I am confused to see such replies from such knowledgeable people as the matter in question is not about who (and when) cries but about a real problem. So let me repost my message from the #general chat to raise more awareness here:

You may look at it as any token (gold, coin, etc) used in online games. It’s “nothing”, but it has value anyway, as long as people keep playing the game. Our game lasts as long as the hedge fund generates profit in USD. Seems like a long game.

Yes and no. Online games sell tokens on their marketplace; they have a base fiat price, and then they might also have a resale value on a secondary market (but not cheaper than the official vendor price otherwise arbitrage is possible).

But in the NMR case, since the token value is not bound to USD/a stablecoin, the staking returns are highly skewed toward alpha/beta of the crypto market. That’s why we see cases when people have great models but awful ROI.

In other words, how many data scientists buy NMR to stake their models and how many crypto investors buy NMR to “diversify” their crypto portfolio because the NMR leadership is awesome (or whatever their DD reasoning is) and the token can go to the moon? What’s the ratio between people who have a direct interest in the project and crypto hoarders?

If the ratio is big, it’s not “Our game lasts as long as the hedge fund generates profit in USD.”. It’s about how many people give up due to volatility created by non-members of the project. And that’s because the majority of stake owners’ ROI would come from the crypto market alpha/beta and not from the actual models’ performance.


Now, courtesy of @nyuton ,

~1.2M NMR out of 6.1M is staked in the tournament+signals.

I have yet to look into how the smart contract issues and burns tokens, but right off the bat it can be interpreted like this: more than 80% of your ROI can be attributed to market volatility and NOT to your model’s performance. And that’s if we don’t put the TC volatility and other factors on top.
So who are we, data scientists or hoarders?


I already posted a proposal to move to a private PoA network on #general, but an even easier solution might be: if NumerAI becomes an official vendor and sets a baseline price for NMR, i.e. they are willing to buy any amount of NMR from anyone anytime at a constant price. This will set a baseline price like a stablecoin (i.e. as an investor you can’t lose more than that) without eliminating the secondary market price formation (if this is even needed, although motivation is not very clear to me).

Ideally, we would want to have it both ways: if we can buy and sell NMR at a specific price to the vendor. Thus making NMR a stablecoin as there are no arbitrage opportunities both ways, and if you don’t want to deal with NumerAI, you would be able to dispose of your tokens at any of the supported exchanges as well.

1 Like

I think they cannot do that for legal reasons. They can buy NMR on market (and most likely they do if the price is really low to provide support). But I guess they cannot really promise anything publicly - they would risk legal troubles and it could perhaps trigger discussion if the token is not a security by any chance.

1 Like
  1. Yes @numerologist, you are right! Buying tokes/shares of a company that is trying things that has never been done before is a BET. If the company/fund doesn’t work. You can loose all your money. I guess everybody understands that. Buying NMR in this sense is certainly a bet on the future of the fund.

  2. As soon as the fund starts generating profits and AUM grows, things become more predictable. Numerai can easily ensure (with the profits at hand) that the token price doesn’t fall below a certain level. I assume, we have already reached that level: Investor Letter

  3. Supporting NMR buy buying it on the open market and stating it publicly are two different things. The latter has legal consequences.

  4. On the long run NMR has a chance to decorrelate from Bitcoin and represent the true supply/demand balance of the stock-market-prediction economy.


Exactly, cheap NMR was a blessing for me, allowed me to get more serious about the tournament and start working on new models. If the issue is NMR price volatility, there were some measures mentioned on this forum, which should somehow mitigate it.

Long live NMR!


Some interesting thoughts in this thread. If someone bullish on NMR is willing to take my crypto exposure so that I can participate only in the DS tournament, please let me know. I’m sure we can find a reasonable price for compensation for the risk, and I’m open to ideas to enforce it. Perhaps it is something the hedge fund can implement for us.


I have been looking for a counterparty to swap with for over a year. The other options out there did not seem convenient :frowning:

Has there been any other discussion about a different system away from NMR?

I’m fairly new to Numerai and love the concept but NMR is by a very very long way the biggest drawback for me.

I’m UK based so I live in GBP. I’m training some models and I did my risk analysis on the cost of that with NMRGBP being in the £15-16 range. Today its at £13 which makes a massive difference to those numebrs.

Also the NMRGBP market is incredibly thin so to move any kind of volume the liquidity dries up. This means I’ll have to move all funds into USDT and then buy NMR with that, exposing me to two sets of fee’s.

The volatility is the biggest issue, along with the risk.

USD, or if it has to be crypto then USDT, for staking and payouts would be so so much better in my opinion as currently it feels like we as providers are taking on a lot of risk thats not needed.