Spoofs, Wash Sales, and the Coase Theorem of NFTs

Nikolai Yakovenko
7 min readNov 1, 2021

The Coase Theorem states that, given property rights, a free market of exchange, and low transaction costs, property over time will find its way to its most efficient owners, regardless of initial distribution.

Whether it’s stated as such, this is the underlying logic behind NFT “drops” where initial collections were partly given away randomly, or to the first digital addresses to “claim” a digital asset.

On June 23 2017, Larva Labs released a collection of 10,000 unique pixelated CryptoPunks. The creators kept the first 1,000 “dev Punks” for themselves, while the other 9,000 Punks were claimed, by whomever had interest in the project. Now those 10,000 JPEGs are worth an estimated 1.7M ETH — or $7.5B USD at today’s exchange rate.

The most valued Punks — and likely the most valuable NFTs to date — are the nine Aliens. Two of these sold for 4.2k Ξ each in March 2021. None have publicly traded since, but there is no reason to think the pieces are worth any less than 4.2K Ξ (ETH), given that Punks have since appreciated by at least 3x as a whole since March, in ETH terms.

However you will note in the list of sales… a recent sale for a whopping 124.46K Ξ — for a non-Alien, perfectly normal Human Punk #9998. My price estimated had this Wild White hair, green clown paint, female Punk at ~96.7 Ξ — below the median Punk value. Can this piece be worth more than all of the Aliens combined?

Obviously not. And it wasn’t.

Within minutes of the sale, sleuths found the smoking gun. The Punk was sold by the owner… to himself. The staggering amount of ETH was funded through a “flash loan” — a loan in ETH that has to be repaid in the same block to be valid.

In a sense, the sale never happened. I will not be using this transaction in updating my price estimates, and the LarvaLabs founders promised to wipe such sales from their records. Given the “attack” was via flash loan, these are easy for them to detect.

Of course none of this has stopped the media, and intellectual celebrities like Nassim Taleb, from chiming in as if the sale means something.

In practice, the spoof bids (market bids not meant to be honored) and wash sales (self sales and sales to related parties, at non-market prices) started a couple of weeks ago, with this 2.5K Ξ bid for another fairly ordinary Punk.

Whether or not this was a joke, a semi-serious offer or a publicity stunt for both the buyer and seller, this was quickly spotted as a fake bid by those who follow Punks. Given the amount was not completely insane (2.5K Ξ has been paid for a Punk, albeit not for a Human, and will be paid for a Human likely one day), a few even expert in the crypto space (but not in Punks markets) treated this bid as legit. But the community overwhelmingly saw it as a spoof.

There is still a gray area. On August 28, 2021 someone bid and bought Punk #8888 for 888.8 Ξ and also bid 888.8 Ξ for Punk #888 but that owner did not accept the bid in time. I choose to treat these bids as legit in the pricing model. But as you can see, the pieces’ attribute-based comps are not worth a fraction of those prices.

It is possible that the numbers 888 and 8888 afford extra value. But the buyer could have almost certainly bought these Punks for less, say 420 Ξ. Those would still be a high price paid for a non-Hoodie, non-Beanie Human. Still, the community seems to believe these transactions are real, if a likely overpay.

Community Based Verification

Overall, while last week’s spoofs and wash sales are not good for the legitimacy of the Punks NFT market, they go to show that the community offer high quality, quick and free validation of large suspicious bids in the network. The ETH can be traced, especially for “flash loans,” but more importantly someone will do it, gladly.

It helps, perhaps as well, to compare bids to an up to date machine-generated, unbiased price estimate, as my model aims to provide. But even absent a model such as ours, it has not been hard for the community to detect insincere bids and self-sales, and to rail against them.

When talking to smart friends about NFTs — even to crypto people who don’t know the JPEGs — a top question I get is “how legit are the sales”? I tell them very legit. This has been true, and while the recent spoofs and wash sales don’t help, we are good at detecting them.

Efficient Ownership

What does any of this have to do with Coase, and his theory of efficient asset allocation?

The NFT Coase Theorem might instead state: over time, as a project’s value appreciates, the pieces will find their way into deep pockets and diamond hands 💎🤲.

Of the original nine Aliens, eight were claimed by collectors for nothing (Alien #635 is a dev Punk). Do any of these still reside with the original collector? Not likely.

In fact we know that a single collector, “Straybits” managed to figure out the random hash and claimed seven Aliens. He sold them all, for prices of 8–12 Ξ each. A likely 1000x discount on their current dollar market value.

He did manage to hold on to an Ape, however, and it’s worth about ~3K Ξ now according to my model.

Another infamous early Punks collector, Pranksy, has bought and sold hundreds of Punks, including many Zombies. However he tends to buy these low, and flip for quick profit. He owns almost no Punks at the moment.

As the values of Punks go up, short term collectors or simply those without much capital will sell to those more comfortable owning six to seven figure $ JPEGs. Even of those who love Punks — like Richerd above — many will eventually sell. Either because the price is right, or because they no longer wish to have their identity tied to a million dollar picture.

Most will be happy they sold (for large gains), albeit I can’t imagine the seller’s remorse on some level, of a Straybits or a Pranksy, who could be sitting on $100M collections.

Collateralized JPEGs

Some collectors, of course, will not want to sell. But nor will they want to keep working in the salt mines for USD, while their NFTs are valued by the market in the six and seven figures.

Moreover, many of these owners would like to buy more NFTs, and taking out personal loans to finance new JPEG purchases, while sitting on seven figures of other JPEGs but lacking liquid funds, starts to sound a little bit absurd.

Predicting the obvious, I expect that fairly soon, there will be generally acceptable protocols for borrowing against non-fungible assets. Something like a 1–2% monthly loan on ~30% of the market value of an CryptoPunk, backed by the asset, strikes me as reasonable. Especially if the loan could be repaid any time with minor penalty. Many collectors expect to make more than 2% a month buying NFTs, and so far the market beta has proven them correct, I would imagine.

Nothing lasts forever, there will likely be an NFT bust — so say many seasoned crypto NFT enthusiasts. Those who have been around crypto since 2015 general expect an 80–90% drop peak to trough for most projects. However, being liquidated after a 70% drop, perhaps isn’t so bad if you got in at effectively zero, and that remaining 30% is all premium.

You can’t keep the Coase Theorem at bay forever — unless you lost your keys, or otherwise just didn’t know your $100 NFT has become a $100K-$1M NFT. After all, no one remembers who owned some of Van Gogh’s original paintings. They are in the museums now, or at home with very wealthy collectors. But lending protocols like this will likely keep over-leveraged “early owners” in the game a bit longer usual. Hurray for DeFi and the nerds who love the protocols.


None of this is investment advice. Entertainment purposes only.

I also have zero percent interest in creating such lending protocols, or extending credit to anyone, regardless of how many JPEGs they have on the wall, or what they use as a PFP.

I simply want to predict what is likely to happen, and suggest how unbiased, accurate prices for unique assets, may be valuable in such a world as one can imagine.



Nikolai Yakovenko

AI (deep learning) researcher. Moscow → NYC → Bay Area -> Miami