How to Auction the 3AC Wallet(s)
Maximizing total sales, transparency and price discovery
As noted by CoinDesk and others, it is impossible to know what the Three Arrows Capital (3AC) and Starry Night NFT wallets are worth. Estimates range from $3M (comically low — their CryptoPunks are worth $2.9M according to our valuations at DeepNFTValue) to a more reasonable albeit still still conservative $10-15M estimates, based largely on entry price and NFT market trends.
There’s also confusion about what is meant by the 3AC and Starry Night Capital wallets. We look at these two beauties, snapshot before any assets were transferred out this week. Before a few recent transfers, the wallets hadn’t been touched for months, pending the bankruptcy.
Thanks for reading Niko on Crypto! Subscribe for free to receive new posts and support my work.
Wallet #1: Containing 11x Punks, 3x Autoglyphs, “The Goose” and other Ringers, 31x Fidenzas, and many Pepe themed NFTs…
We are updating the article, based solely on Wallet #1 above — confirmed to be part of the 3AC/Starry Night collection. While Wallet #2 is owned by hedge fund billionaire and art collector Alan Howard. Many grails does he hodl…
In a companion piece [to be published shortly] we break down the holdings one by one, starting with the jewels of the collection (Fidenza, Ringers, Punks, Autoglyphs, Chromie Squiggles) down through the middle tier (Subscapes, Meridians). We estimate this analysis captures about 80% of the wallets’ total value, at an estimate of
~$26.4M ~$16.7M in ETH at today’s prices (12.2K ETH ~ $1,358/ETH). [Our analysis excludes both the long tail, and also small collection grails, 1-of-1s, etc — these make up a substantial minority of the wallets, by value.]
Since some of you will be burning to know how we get to 12.2K ETH and for which collections, here’s a summary. Check out the companion piece for methodology and details.
Wallet #1 (confirmed 3AC/Starry Night):
collection floor num paid value USD
Fidenza 87.83 30 3,451 6,414 $8,716,943
Punks 66.00 11 1,230 2,177 $2,958,281
Ringers 53.25 17 3,204 2,126 $2,890,093
Autoglyphs 147.59 3 788.52 716.00 $972,958
Archetype 14.20 13 628.27 365.66 $496,888
Deconstructions 21.06 6 382.90 180.05 $244,666
Subscapes 9.17 7 210.57 158.98 $216,034
Squiggles 9.84 1 70.00 121.99 $165,769
Total -.-- 88 9,966 12,261 $16,661,635
Wallet #2 (Alan Howard, unrelated — including since we posted these in the original story):
collection floor num paid value USD
Ringers 53.25 21 7,974 3,484 $4,735,329
Squiggles 9.84 39 2,483 1,424 $1,935,221
Fidenza 87.83 4 1,565 735.90 $999,999
Autoglyphs 147.59 3 1,255 509.79 $692,743
Eternal Pump 215.00 1 224.90 425.12 $577,687
Subscapes 9.17 9 1,095 285.01 $387,294
Meridian 6.79 10 536.80 210.23 $285,677
Archetype 14.20 1 125.00 62.20 $84,522
Deconstructions 21.06 1 135.00 32.04 $43,538
Total -.-- 89 15,394 7,169 $8,164,327.69
Once we break out the confirmed 3AC wallet from Alan Howard’s — we notice that 3AC are actually up in ETH terms on most collections we look into, not only Punks. The reason is timing.
Just about all of the buying by Wallet #1 took place in July/August 2021. Fidenza sales reached their highest levels a month or so afterward.
The wallet is up on Punks, at least in ETH terms. We can see this clearly from our individual valuation estimates on DeepNFTValue. All but two of these 11x Punks were bought below the current 66ETH Punks floor, even the ones with 3D glasses.
It’s a mixed bag for other collections. Some Fidenzas were bought pre-peak, and some at peak. More to the methodology of these price estimates in the companion piece — we believe that most of you will find our methods reasonable, and broadly conservative.
Having said that, these estimates come with error bars attached. Grails are “priceless” and hard to value precisely until a piece goes to auction. In the companion piece, we value these top two Ringers in the collection at 657.4 ETH and 505.9 ETH, respectively.
The first of these pieces (not The Goose) is owned by Alan Howard’s wallet. If this one too has a special name, please let us know.
These pieces were bought for 2100 ETH and 1800 ETH, at the peak of the Ringers boom in August/September 2021. Both are likely worth less today, in ETH terms much less in USD — however we doubt that you’ll could find a credible Art Blocks collector who would suggest that “The Goose” will fetch less than 1,000 ETH at auction. Yet our model conservatively prices it at 505.9 ETH.
Price discovery on these grails, is reason alone to run an auction.
The creditors of 3AC would likely have a few objectives in an orderly disgorgement:
collect the maximum total sum for the NFT assets
sell off all of the assets “cleanly” — leave nothing remaining
complete the sales in a timely manner
sell the assets in a fair, open and legally defensible manner
minimize fees to auction houses, etc
We don’t see selling the wallets in one lot as value maximizing or defensible. The lot — at an approximately $20M value [including items we didn’t rate] — while a small part of a $1B+ bankruptcy, is still too lumpy for more than a few parties to bid on. As we’re written about for the “Sotheby’s 104 Punks” auction there is no case for a bundle of NFT assets being worth more than the “unbundled” sum. Bundling destroys value.
The buyer of the single lot will demand a discount, get it, then later selling off a portion of the lot at higher unbundled prices, making the auction proceedings look like they did not maximize value. Meanwhile many smaller parties would like to bid on individual items, or a subset of the collection like “all of the Subscapes,” yet are excluded from the price discovery process in a single lot sale.
Let us to propose an auction mechanism for maximizing value, increasing access to smaller collectors, and maximizing price discovery.
Of course this will likely not happen.
But let’s dream a little…
Level 1: Partial Un-bundling
Auctioning a single NFT is easy, and it happens every day. It’s enough to start an auction on OpenSea, Fractional Art or on AAVE — and let the community know on crypto Twitter. These platforms take low single digit auction fee on the final price: 2-5%.
However, in this case the creditors need to auction a collection of hundreds of NFTs:
177 88 in large collections that we scored, and 30+ other assets that are likely to fetch 10-200 ETH each [not counting the other 3AC wallets we may be missing in this analysis]. For example the wallets owns one Drifter “Where My Vans Go” piece, bought for 46 ETH and likely worth more than that today.
Running 100+ auctions in parallel would cause pandemonium. It would be bad for extracting maximum value, bad for price discovery, and may be even worse than selling the entire collection as a single lot, to be quite honest — as some lots go unnoticed or unsold. Leaving simultaneous auctions open for a long time does not really make things much better. Bidders’ ETH is tied up until an auction completes, and all of the serious bidding takes places close to the deadline.
Inevitably some lots will go cheap, and get flipped days or weeks later, once the market has had time to digest this tsunami of assets.
The creditors’ goal should be to sell off all of the pieces, in an orderly and timely manner, while maximizing total price, and somewhat minimizing fees.
A simple partial solution would be to form a committee of NFT experts, and carve up the entire collection into a few categories:
grails, to be sold individually — such as “The Goose” Ringer, the Zombie CryptoPunk, and the Drifter photograph shown above
common pieces from “established” collections, bundled by collection — for example the ten non-Zombie CryptoPunks can be sold as a single lot [we value the non-Zombies at around 1,000 ETH total]
remaining miscellaneous items — organized by type (photography, memes, etc)
For example the wallets include a number of Pepe the Frog items, some bought for frothy prices. The frogs don’t fit into any of the top 6-10 collections but can be bundled into a single lot — attractive for a meme focused collector.
There are no Bored Apes in the collection, but there’s a golden Bored Ape Kennel Club dog with a banana on his head. This can be a single lot, or it could be a jewel of sorts bundled alongside a bunch of lesser odd lot non-art, non-meme NFTs.
We see this committee, and a smart splitting of the collection into grails and theme-based bundles, as an improvement, and also somewhat likely to happen.
Presumably one could think about the ordering of the lots, to maximize value, and excitement. The order in which items sell, will certainly affect prices. Closing an auction is also great, so focus can shift to the next grail on display.
We would suggest auctioning the bundles first, the grails second, while oscillating between collections and slowly building toward the best grails, finishing with the odd lots and the Pepes. Think something like the college football New Year’s Bowl Games or the World Series of Poker.
This would be pretty cool, and the auctions could last a fortnight.
But we can do better…
Level 2: Group Estimates and Pre-Pricing
Imagine we already have several groups of NFT collectors, itching to bid on the single lot — or on big chunks of the collection. Instead of producing one winner and a number of losers, we propose that these whales work together, pre-price all of the lots in the collection — with minimal bundling, except for the long tail and the odd lots.
Pre-pricing provides two benefits:
a conservative but reasonable lower bound on the value of each lot in the collection
the whales’ involvement and support for the auctions
A canny whale will not wish to declare his maximum price for a lot, especially publicly. However we can still involve them in a process of price discovery, by asking them to declare a price privately. Then we will run a full auction afterwards, starting with the “second highest private bid” as the reserve price.
The full procedure looks like this…
Pre-qualified whales submit private bids for all* of the lots in the collection.
These bids are semi-binding. More in a moment.
The whales are able to submit a lowball bid in private, however if they bid significantly more than their private bid in the final auction (say over 50% more) they will be forfeit for the rewards that they would have received on that lot.
The whales will be rewarded — both for participating (winning auctions) and for pre-price discovery (submitting high private bids for lots they don’t ultimately end up winning).
An independent auditor will have access to all of the private bids, examining the process in real time — with the authority to reveal private bids if warranted. No funny stuff.
All the whales’ rewards will be publicly announced, but not the private bids, unless the auditor deems it warranted.
The auctions take place in three stages:
Whales submit private bids. The second highest of these private bids determines the reserve price of the auction.
All non-whale participants are invited to submit binding private bids in phase two of the auction. They need to bid above the reserve price, to qualify for the final, live stage of the auction.
Live bidding will begin, open to all whales, and to all non-whales who submitted a qualifying bid in the second stage. No one else can participate. The auction has a brief time limit, which is extended by 5 minutes by every new high bid. Each new bid must be 1% higher than the previous high bid, and programatic bidding is allowed.
Auctions will be staggered — 2-5 auctions proceeding simultaneously, like multiple channels on NFL Sunday. Each lot will take several minutes to several hours, depending on how many bidders make it to the final stage. The actions should be friendly both to programatic bidders — setting a private maximum and increasing their bid automatically during the extension period — as well as to humans who want to bid live and will have several minutes to react to each development.
Starting with the second price of the whales’ private bids allows for two things
a guaranteed* minimum sale on all items
a splitting of the sale price into baseline and premium — without resorting to heuristics like floor prices
Take Punks for example. Any well capitalized whale can submit a bid of the current Punk floor, plus or minus epsilon. Being forced to buy a Punk at floor, or a little below floor, is not that big a deal. More likely, the Punk will sell for a bit more than the floor, anyway.
Having said that, even every case it is useful for the creditors to know that they will get at least the floor price for every Punk — and similar for other collections where floor-based heuristics are less reliable.
It also makes little sense for the creditors to pay an 18% auction house fee for selling a Punk at floor — when you can list it on the Punks marketplace and pay nothing but gas. But it does make a bit of sense for the auction participants to “earn” something like an 18% fees on the “premium” above a lower bound estimate, when such a premium is extracted by the auction process.
Lots generate fees for the whales in two ways:
If a whale buys a lot, he receives a refund on the premium that the lot fetches over the reserve price. This fee is up to 16% of the premium, but ratcheted down if the price paid is much more than the whale’s initial private bid.
All whales who submit an estimate will split a 2% “price estimate fee” of the premium when an above-reserve sale is generated — in proportion to their proximity to the auction price.
If the reserve price is not met, the highest bidding whale is required to buy the item (at the second highest price — ie less than his bid) — and will receives a 5% refund on the total price. No other fee is generated.
In other words, whales are compensated for submitting high private bids, and also for buying lots. In return they are encouraged to bid on every lot, to bid as much as possible, and these bids are semi-binding.
There is scope for collusion between whales, but this is why we have a trusted auditor to check for funny business. At worst, the auction still sells every lot, paying a combination of a 18% on the “premium” above a lowball estimate of the item, and a 5% fee when nobody else bids above the whales’ reserve estimate.
The 3AC wallets contain a lot of grails. Assuming that price uncertainty above floor is something on the order of 30-50%, we estimate that this fee structure would generate 5-10% of the total sales price of the lots in fees — even if the whales end up buying most of the lots in the collection. Not bad, all things considered. And every time a whale buys a lot, he gets a discount, which aligns vibes and incentives. While at the same time providing an opportunity for anyone outside the cabal of whales to bid, encouraging broad participation and higher prices.
Of course in a fortnight long, multi-auction process, we can not hold whales to their original private bids, especially as they buy up lots, depleting ETH reserves. We would allow the whales to alter their private bids, up until the time that a lot opens.
We do ask the whales to submit initial bids for all lots at the start, if anything to order the lots correctly. But the bid that is binding, can be modified up until the very last minute.
Summary stats will be published at the end of the auction, declaring not only which whales collected how much in fees, but also how close they were to submitting “ballpark” bids for the various auctions. Whales who lowballed all of the lots will be exposed and shamed, so to speak, while collecting very little in fees.
Of course this is all pretty unlikely to happen. Selling off ~$20M in NFTs as part of a billion dollar bankruptcy, will likely be an afterthought. However we hope something like this does get implemented. It would be nice for price discovery and price transparency, selling the NFTs one lots one at a time, in the right order, with reasonable reserves and minimal bundling.
The Art Blocks platform charges a fee for projects listed on the platform of 10%. At the time a token is purchased, 10% of the purchase price is transferred to the Art Blocks address with the balance immediately transferred to the project creator.
— from the Art Blocks website
Another reason that an auction may need to happen, is that ArtBlocks in particular, generate hefty artist royalties upon resale. It is possible to avoid such fees via private transfer, but it is not in a collector’s interest to antagonize Tyler Hobbs or Snowfro by avoiding such fees. When fees need to be generated, it becomes imperative to price each item via individual sale. Or at least to bundle items by collection, for proper artist attribution. But once we do that, we are halfway to a process like the one suggested..
Given that the wallets are largely ArtBlocks by value, it may behoove the ArtBlocks team to advocate for such an auction process, and supporting the organization of a process through a portion of their fees. If the ArtBlocks part of the collection sells for at least $10M, that’s a $1M fee, split between the artists and Snowfro.
It does not seem unreasonable to suggest that a party collecting $1M in fees, could help pay for some administrative costs and perhaps even the whales’ first class tickets to Singapore for a live auction of “The Goose.” A well publicized and orderly auction will almost certainly generate higher total sales, and higher fees than a free for all, especially when there’s a guarantee on all items. And whatever fees this sale generates, will be 100% higher than the fees one could have expected, had 3AC not went bankrupt, and held on to The Goose and other grails for a long time, as we’re sure that they intended.
Veering further into opinion, we think that a well-run, “grand event” of an auction might be good for the NFT space.
Anyone who’s been an vintage car auction in Vegas knows it’s exciting to see a 1969 Mustang get sold to the highest bidder. You need those highlights, but the event moves hundreds of other cars, from rebuilt 1920s pickups to 1970s Chevy Chavelles, to modern monstrosities.
August 2021 was Art Blocks szn; volume hit stratospheric heights. H/t Richard Chen’s cryptoart leaderboard.
There was a “call to action” buying into Art Blocks then — these collections were in the zeitgeist. You can buy a Ringer or a Fidenza right now. Several are for sale at reasonable prices. But there’s no call to action. There’s no reason to start researching which ones you like, or a reason to think these might be offered at a better deal.
There is no Art Blocks “free agency season” to use another sports analogy — a downside of the 24/7 crypto art market. Until now.
A well designed, well publicized auction, brings attention to these collections. We would known that 30x Fidenzas are about move, and a bunch of Ringers. This gives the public time to do their research, and take a bit of interest. There’s still probably 90% of potential Fidenza owners, who haven’t bought one… and here’s a chance to get a Denza (one of 30) that will, absolutely, definitely sell — in a fair and honest time-limited process. Not dripped out over weeks by a dealer.
While $20M bags is good chunk of change to drop on a fragile market, it about what QQL generated in a 14ETH/NFT mint last week. And it’s hard to imagine that much of the QQL interest is coming from those who don’t already buy high end NFT art.
But also for an auction to proceed smoothly, it helps to have reasonable reserves. I had several people reach out, concurring that a “serious bids only” auction would run more smoothly — auto auctions for vintage cars tend to start with a reserver. In this case, given there’s a bankruptcy proceeding — you do need a cabal of whales to guarantee some prices to generate reserve prices for all items — unless one were to sell the whole lot first, then auction off from there.
A number of you have reached out in DMs — feel free to drop a public comment here as well. Glad to get the conversation going on how to best sell or distribute these grails. Interested to hear what you think.
Thanks for reading Niko on Crypto! Subscribe for free to receive new posts and support my work.