Stablecoin Cyclones: Mint & Burn Patterns
Previously we discussed how minor stablecoins looked like they might be a way to feed cash into the banks that bank crypto. Let’s dig in a little deeper.
We are going to work through, in turn, the same four minor stables as last time. They vary in how interesting they are:
- USDP reveals an interesting, and quite old, potential Binance connection.
- TUSD could not look more like it was trying to hide something.
- HUSD reveals direct connections among quite a few players adjacent to the TRON suspicion we previously raised.
- GUSD is odd, but not in the same way.
A novel charting technique is used below to illustrate what is happening. These charts are derived entirely from on-chain data and should be somewhat self-explanatory to folks who are used to that. At the bottom a more detailed technical presentation is provided.
The point here is that the mechanism we offered last time as pure speculation — that these stablecoins are used as vehicles to get USD into the “major crypto banks” — is not just plausible but consistent with a lot of on-chain data. And this data reveals direct connections among precisely the parties we would expect if that narrative is correct. It’s speculation until you find evidence it might be true; then it’s a theory.
Also special thanks to Dirty Bubble Media and Matt Ranger for wallet tagging assistance.
How Should These Stablecoins Behave?
We already established these 4 stablecoins exhibit an awful lot of burning. And, as is widely known, they are not heavily used for trading. Of course there are some exceptions — the market caps are not zero and coins are not totally useless — but in broad strokes those statements match empirical reality.
If these coins were useful we would expect to see most tokens out there floating around various protocols. We would expect lots of inter-exchange transfers and flows within DeFi. But most importantly we would expect to see long routes between mints and burns as the coins are used for real things throughout their lives. Sure, some people will mint and then redeem but that makes no sense as the main use case.
Now, we already know there is a lot of burning. So the best we can hope for is a range of circuitous routes through the ecosystem as tokens experience real economic uses before finally being redeemed.
What we do not expect to find in a useful coin is a large number of short, high volume routes from mint to burn. What economic purpose does that serve? Think of cash in the real world: how often do you withdraw a lot of money from the bank, pay someone, and they then go deposit that cash back in the bank? And what is really happening there?
USDP
Paxos runs USDP. And only a handful of wallets mint or burn USDP. An interesting Master’s Thesis entitled “Stablecoin Billionaires” from 2020 notes that with the exception of a handful of transactions in 2018 all USDP are handled by 0x5195427ca88df768c298721da791b93ad11eca65. The tiny wallet is 0xcba1766e19f32d79e3ba7a2764f04a7042324f0a and it’s been quiet for years.
But what that thesis does not say — because it’s from 2020 — is that starting in March 2021 USDP moved over to address 0xe25a329d385f77df5d4ed56265babe2b99a5436e. Now Whale Alert calls that address Binance. That is weird.
Huobi stopped using Paxos a few months later (as previously reported). And as was separately reported there is at least some connection between BUSD and HUSD beyond Paxos:
This may have come later. We don’t yet know. But Paxos switching to a wallet labelled Binance in March 2021 significantly pre-dates the trademark transfer referenced in that tweet.
Anyway, Paxos did these transactions itself. And if we expand the circle we see:
What do we find? We find tokens being minted to Binance and Huobi. And then those tokens make their way, sometimes via intermediaries and sometimes directly, to FTX/Alameda. We also see tokens minted direct to untagged wallets — in size, recall these are all large net flows —then transferred to exchanges and redeemed.
There are a few parties that look to be intermediating. And remember: USDP is not really useful on it’s own. Yes it counts as USD on FTX. But so do USD and a few other things. These graphs make a lot more sense as a way to funnel USD in than as any sort of real use of the funds. And we already found a pile of dollars that match this inflow strikingly well.
TUSD
The first thing to say about TUSD is that there are thousands of address that mint or burn tokens. Thousands each. But only a handful that mint and burn — and those tend to be incredibly unbalanced or tiny. Like minting >$200 million and burning $1k. Or minting and burning about $100k each. There are fewer than 10 of these addresses out of thousands of involved wallets and precisely none of them both mint and burn a wholesale amount of money.
And it gets more interesting. The burning is weird. The minting…let’s just say that reveals some expected parties.
Burning
TUSD are burned out of addresses that begin 0x00000000000000000000000000000000000…. That’s right: with 35+ zeros. These are ERC20 transfers not ETH ones so this does not require any special cryptography. Look at this table of top burners:
Yeah that’s weird. Oh, and those few addresses that both minted and burned?
They are the only burners that also mint.
Minting
Minting looks more natural. And wouldn’t you know, some open source reporting labels the top two wallets:
That’s right. The top wallet is cited as “Alameda FTX Depo” by a reasonable Twitter account. And the #2 address is labelled “Justin Sun” on Etherscan. Last time we said the TRON connection was not a pipe dream. Indeed.
Oh, and one of the top burning addresses is 0x00000000000000000000000000000000000020d8 which interacts very little…and almost exclusively with the Justin Sun 0x3ddf cited above. Maybe that de-anonymizes one of these.
So what happens when we hunt for involved addresses and plot them?
Here we see a few cycles that are generally consistent with the idea that these things funnel USD into USDT. Tokens are minted to Binance and redeemed via FTX. And the other way around. Tokens are also minted to Justin Sun and then knock around a closed system for a bit before being redeemed.
And, perhaps most interestingly, FTX and Binance share a redemption address. Remember these are special addresses TrueUSD created that start with 35+ 0s. They are ERC20-only and not the result of a random process (because physically that’s impossible). So either TrueUSD made a mistake and sent the same address to these two rival exchanges or they are coordinating TrueUSD on some level. We will note that TrueUSD does not look particularly error prone on this front. Again: we have more data than this.
HUSD
This one is the juiciest. First we found nearly a thousand addresses and not a single one of them minted and burned. But both sides look to be proper addresses. We were able to tag the largest minter (0x5f64d9c81f5a30f4b29301401f96138792dc5f58) as Huobi and the 3rd largest minter (0x83a127952d266a6ea306c40ac62a4a70668fe3bd) as Alameda.
Also some of these wallets — like 0x5586a235b4c5eff57b8b65f77a97780d9347fa47 — look like staging wallets off of exchanges (Huobi in this case). All they do is receive tokens from Etherscan-tagged exchange wallets and then burn them. One imagines somewhere an operations person is checking the tokens arrived, redeeming the real USD, and then burning the tokens. Maybe it’s automated — it doesn’t really matter precisely how it works.
The largest burner (0x4bd0244d26df7780a56df2f20c53b507e35cb373) looks like a staging wallet out of the one we found above tagged Binance (0xE25a329d385f77df5D4eD56265babe2b99A5436). Now recall we already know everyone involved banks at the same handful of places with the suspiciously-USDT-like cash balances.
So what do we see here? We see Huobi, Alameda and Binance minting and redeeming a lot of these things out of different wallets. And it’s not like the wallets just rotate over time — many many wallets are used for years on both sides. These look very much like individual exchange-adjacent client wallets, minting and redeeming out of different sources.
This is the heart of the flow and it’s all cycles. These are billions of dollars in tokens that are not used for trading flowing from source, through a pair of exchanges in close proximity, and then back to sink.
GUSD
This one is different. The Gemini Dollar does not look to interact much, or directly, outside the NYC area. Here are all net flows > $50 million within 3 hops:
There is a little interaction with curve.fi. But really it’s stuff involving Genesis and BlockFi. In a literal circular flow of funds. Whatever. We can cross this one off the list as far as “major source of inflows” is concerned. If there is anything interesting happening at all it is within Gemini.
Now, we can have a discussion about whether there is much real economic activity or not. But given little of the minting and burning escapes the tri-state area whatever is happening it looks, um, contained.
Last time we decided to ignore the elevated churn at BUSD and just add up all activity at 4 minor stablecoins. The idea was to assume whatever oddness may or may not exist at BUSD nets off against whatever slice of action in those minor stables was not “odd.” GUSD total burn is around 10% of total BUSD burn. So our assumption looks reasonable — most of GUSD can be purposeful and still we need only 5% or so of BUSD burns to offset this Gemini->Genesis->BlockFi->Gemini circle. Note that, as we have continually noted, what look like “guesses” are really conclusions drawn from evidence you’ve not seen yet.
Huh?
This could not look more like an effort to obfuscate capital flows if it tried. Yes it is plausible this is all bog-standard economic activity. But recall, again, what else we know: USD at the “crypto money center” banks tracks USDT well, and USDT-on-TRON tracks these stablecoin burns well.
Now we know that the former head of TRON is involved directly in the process. We also know these burns are happening out of places that bank at the major banks — and that freshly minted tokens often make their way there before returning home to the null address.
Does this prove beyond all doubt these are conduits for USD to get redeemed at Signature/Silvergate and then back USDT? No. Does it show how closely the empirical evidence matches a world where that is happening? Yes. Yes it does.
If anyone else has a plausible theory for what is going on please let us know. We’ve got lots of data and tools to test theories.
A Note On Methods
Last time we introduced a charting technique that made plain some weirdness in stablecoin flows. This time we dug a bit more deeply on-chain. This is just a quick summary of how we built the plots above.
Token mints and burns come out of, and go in to, the 0x000…0 “null address.” These charts consolidate all flows among wallets within a few hops of null. And we drop small net positions to keep the charts (relatively) clear.
This means if two wallets swap a billion dollars back and forth but are net flat we don’t bother to show it. This is about finding loops that flow from null, out in to the world, and back in to null. Further, we did not run a contrived search to fit a theory. As we have been hinting all along this is a multi-part story. And any foreshadowing refers to results already on hand. But it is important to work is carefully and in steps.
Finally, and once again: apologies this is not an ELI5. This is raw primary source work with novel methods we’ve developed and built ourselves. It’s a bit ugly…but that is how sausage is made. Incrementally releasing this story has yielded a lot of helpful information. So we will continue to focus on getting information out there over producing attractive charts and tables. Better graphics are coming in the recap.
