Haha, I was just about to ask about impermanent loss until I got to the last paragraph. Looking forward to that. I'm not fully tuned to the maths, but my best guess is making LP leverage levels dynamic.
Questions:
1. What is the best stablecoin (in terms of robustness to censorship and also robustness to market volatility) that is currently out there, in your opinion? Why?
2. I suppose one issue with the above stablecoins is fungibility. There is USD1, USD2, USD3 etc. If liquidity is unified (kind of like the ill-fated Bancor) then that prevents specialisation... Is there a way to get around this?
3. How big do you think the decentralised stablecoin market can be as a fraction of the total stablecoin market (including fiat backed stablecoins). Why?
1. DAI is most uncensorable, but USDC is more liquid.
2. If one of the stablecoins becomes less valuable than $1, say $0.5, and that stablecoin is taken as "USD" in a pool, the price of the token in that pool will adjust via arbtrage assuming there are ways to trade those tokens in Uniswap, etc. The stablecoins generated by a LAMM using a depreciated stablecoin will then mirror the value of the input stablecoin. I'm not sure what kind of situation you are trying to get around?
3. I remember working this out, and I think the theoretical limit is the amount of ETH outstanding times 1 minus the margin rate. So, if there are $100B ETH outstanding, and the margin rate is 20% (ie, leverage is 5x), you could have $80B in stablecoins created maximum. Given practicalities about margins, it would be less than that (ie, everyone targets an excess margin ratio of 40%). I'd have to get back to you on that...I think one issue is that unlike with traditional banks, no one in the system is given a partially unsecured loan, so you do not get the standard money multiplier where $1 in reserves generates $1/(1-reserve ratio) of M1.
1. Have you looked at LUSD (probably more censorship resistant than DAI as it's only ETH backed. It uses a chainlink oracle for liquidations, but has Tellor oracle as backup) and RAI (also only backed by ETH, uses chainlink oracles)?
2. I am thinking there is a LAMM for WBTC-USD1, and one for ETH-USD2. USD1 and USD2 are not exactly fungible. The liquidity is fragmented, which is not ideal. There's a version of each USD for every liquidity pool...
3. Good point. I wasn't thinking of theoretical limit though. I'm curious what you think this percentage can be (percentage uncensorable as a fraction of total stables)? What I'm trying to get at is how big you think the demand can be for uncensorable stables? And why you think that potential demand is big or small?
I haven't really looked at LUSD and RAI, but they seem ok. I don't like the oracle dependence, as that can be censored.
2. There are trade-offs. A single USD from multiple pools would generate more acceptance than multiple ones, but then this single stablecoin would be subject to the risk of the weakest pool (here, WBTC) that would be easier to manipulate. Arbitrage should keep USD1 and USD2 at equivalent values. Alternatively, you could create a USD1 and USD2 pool with low fees so if people really like one they could just use that.
3. With ETH at $150B it would be easy to get $30B in stablecoins out of it, maybe $100B. There's $150B in stablecoins now, though a lot of that is UST and BUSD (about $90B), which are not trustless, but they are useful accounting fictions that enable trading. In the long run I think this is the way, because as mentioned, I don't trust oracles like Chainlink in the long-run (regulators know who to lean on to blacklist various dapps); USDC, UST, BUSD are centralized and not transparent. There's $20 trillion in M1 and $2 trillion in currency outstanding, so there's clearly a lot of potential demand (add in other big fiats probably make that market 3 times larger).
BUSD is the very same as USDP (Paxos). It is pretty transparent. It is regulated in New York. You can find the reports online.
USDC isn't as strongly regulated as USDP or BUSD, but it's a lot more transparent now than it was back when it was backed partially by commercial paper.
Clearly there is a $100B+ demand for stablecoins, but this is nearly all fiat-backed stables. My question is whether and why a decentralised coin (like LUSD or RAI or your proposal) could take over from BUSD/USDP/USDC?
The benefit I see to decentralised stables is that the regulatory steps of one jurisdiction do not extend to other jurisdictions (e.g. if OFAC ban a USDC address, Circle blacklists and everyone is blocked. If OFAC blocks an LUSD address, non-US people can still redeem their LUSD for ETH.) I'm not sure if this benefit is substantial enough though for decentralised US stablecoins to dominate the overall stablecoin market though...
So, it seems to me that decentralised stables will always be a small fraction of total stables, but I am interested in strong arguments against that.
Haha, I was just about to ask about impermanent loss until I got to the last paragraph. Looking forward to that. I'm not fully tuned to the maths, but my best guess is making LP leverage levels dynamic.
Questions:
1. What is the best stablecoin (in terms of robustness to censorship and also robustness to market volatility) that is currently out there, in your opinion? Why?
2. I suppose one issue with the above stablecoins is fungibility. There is USD1, USD2, USD3 etc. If liquidity is unified (kind of like the ill-fated Bancor) then that prevents specialisation... Is there a way to get around this?
3. How big do you think the decentralised stablecoin market can be as a fraction of the total stablecoin market (including fiat backed stablecoins). Why?
1. DAI is most uncensorable, but USDC is more liquid.
2. If one of the stablecoins becomes less valuable than $1, say $0.5, and that stablecoin is taken as "USD" in a pool, the price of the token in that pool will adjust via arbtrage assuming there are ways to trade those tokens in Uniswap, etc. The stablecoins generated by a LAMM using a depreciated stablecoin will then mirror the value of the input stablecoin. I'm not sure what kind of situation you are trying to get around?
3. I remember working this out, and I think the theoretical limit is the amount of ETH outstanding times 1 minus the margin rate. So, if there are $100B ETH outstanding, and the margin rate is 20% (ie, leverage is 5x), you could have $80B in stablecoins created maximum. Given practicalities about margins, it would be less than that (ie, everyone targets an excess margin ratio of 40%). I'd have to get back to you on that...I think one issue is that unlike with traditional banks, no one in the system is given a partially unsecured loan, so you do not get the standard money multiplier where $1 in reserves generates $1/(1-reserve ratio) of M1.
1. Have you looked at LUSD (probably more censorship resistant than DAI as it's only ETH backed. It uses a chainlink oracle for liquidations, but has Tellor oracle as backup) and RAI (also only backed by ETH, uses chainlink oracles)?
2. I am thinking there is a LAMM for WBTC-USD1, and one for ETH-USD2. USD1 and USD2 are not exactly fungible. The liquidity is fragmented, which is not ideal. There's a version of each USD for every liquidity pool...
3. Good point. I wasn't thinking of theoretical limit though. I'm curious what you think this percentage can be (percentage uncensorable as a fraction of total stables)? What I'm trying to get at is how big you think the demand can be for uncensorable stables? And why you think that potential demand is big or small?
I haven't really looked at LUSD and RAI, but they seem ok. I don't like the oracle dependence, as that can be censored.
2. There are trade-offs. A single USD from multiple pools would generate more acceptance than multiple ones, but then this single stablecoin would be subject to the risk of the weakest pool (here, WBTC) that would be easier to manipulate. Arbitrage should keep USD1 and USD2 at equivalent values. Alternatively, you could create a USD1 and USD2 pool with low fees so if people really like one they could just use that.
3. With ETH at $150B it would be easy to get $30B in stablecoins out of it, maybe $100B. There's $150B in stablecoins now, though a lot of that is UST and BUSD (about $90B), which are not trustless, but they are useful accounting fictions that enable trading. In the long run I think this is the way, because as mentioned, I don't trust oracles like Chainlink in the long-run (regulators know who to lean on to blacklist various dapps); USDC, UST, BUSD are centralized and not transparent. There's $20 trillion in M1 and $2 trillion in currency outstanding, so there's clearly a lot of potential demand (add in other big fiats probably make that market 3 times larger).
BUSD is the very same as USDP (Paxos). It is pretty transparent. It is regulated in New York. You can find the reports online.
USDC isn't as strongly regulated as USDP or BUSD, but it's a lot more transparent now than it was back when it was backed partially by commercial paper.
USDT is definitely less transparent. More here: https://pinotio.com/p/what-is-the-safest-dollar-backed (btw, UST is US Terra, no longer around).
Clearly there is a $100B+ demand for stablecoins, but this is nearly all fiat-backed stables. My question is whether and why a decentralised coin (like LUSD or RAI or your proposal) could take over from BUSD/USDP/USDC?
The benefit I see to decentralised stables is that the regulatory steps of one jurisdiction do not extend to other jurisdictions (e.g. if OFAC ban a USDC address, Circle blacklists and everyone is blocked. If OFAC blocks an LUSD address, non-US people can still redeem their LUSD for ETH.) I'm not sure if this benefit is substantial enough though for decentralised US stablecoins to dominate the overall stablecoin market though...
So, it seems to me that decentralised stables will always be a small fraction of total stables, but I am interested in strong arguments against that.