A New Day For AMMs — Impermanent Loss Has Been “Solved”
By providing users with open, censorship-free financial services, DeFi protocols have seen their assets under management, also referred to as Total Value Locked (TVL), soar:
Uniswap, a decentralized exchange that works via an automated market maker (AMM), is the most successful of all DeFi protocol through the TVL:
A Quick Primer On Liquidity Pools
On decentralized exchanges (“DEXes”) like Uniswap, liquidity providers or LPs, provide their capital to specific liquidity pools. In exchange for providing liquidity to specific pools, Uniswap LPs are rewarded with the 0.3% in trading fees generated by every trade executed via those pools. The fees are split by LPs proportional to their contribution to the liquidity pools. Below are the the 10 pools on Uniswap with the most liquidity
On the chart above you can also see the fees earned by each liquidity pool in the prior 24 hours. Multiplying those 24 hour fees by 365, gives an annualized fee estimate. And taking that annualized fee estimate, and dividing it by the total liquidity, gets you the estimated annualized yield via fees (the last row in green).
However, the annualized yield doesn’t tell the whole story. In fact, just telling investors the annualized yield is often like telling people about the night that Lincoln went to Ford’s Theater to watch “Our American Cousin”, without mentioning the fact that he was assassinated.
Impermanent Loss Can Be A Killer
When LPs provide liquidity to Uniswap pools, they must put an equal dollar value of each token in the two token pools (e.g. WBTC and ETH). However, when one of the tokens moves up or down more than the other token, there is “impermanent loss” (“IL”) incurred by the pool as it’s brought back in to 50/50 balance. The larger the moves, generally, the larger the IL. Sometimes the IL is so large, that when subtracted from the yield earned via fees, the net yield turns negative.
Nor surprisingly, IL is the bane of every AMM. Thus, every AMM has tried to algorithmically solve for, or mitigate, for IL. Those attempts have met with limited success to date.
Now, Bancor, the godfather of AMMs, has introduced Bancor v2.1, with a novel, simple, and potentially game changing “solution” to impermanent loss.
Bancor v2.1 — Shifting Impermanent Loss From LPs To The Protocol
In v2.1, Bancor made three pretty simple changes from v1, that could have a profound impact on AMMs.
The first change is that LPs only need to put one token (I’ll use a generic “TKN” symbol for this post) in to a liquidity pool. The other half of the liquidity pool is provided by the Bancor protocol and in the form of newly minted BNT (the Bancor protocol token).
The second change is that the fees don’t all go to the LPs. As the LPs only put up half of the value in the liquidity pools, they only receive half the trading fees. The other half goes to the Bancor protocol. Those fees are eventually burned when the stake is unwound.
The third, and most profound change, is that Bancor compensates LPs for any IL by minting new BNT to cover the loss.
As a result, Bancor LPs are able to earn more reliable yields on their fees and rewards, which cause LPs to be less transient with their funds given the steadier nature of returns.
It’s A Better Mousetrap For LPs, But Is It Better BNT Holders?
Bancor’s solution for impermanent loss is simply a better mousetrap for liquidity providers given the greater certainty of yield. LPs are effectively earning swap fees on their tokens without IL reducing their gains, and principal.
That said, since an initial 16% gain in BNT in the first hour following the announcement of v2.1, BNT has fallen over 48%. So there is obviously concern over v2.1’s success. The unknown, is how much more BNT will be minted to cover all the impermanent loss vs. how much BNT will be burned via fees earned.
An analysis of the fee-earning potential of Bancor’s new model is presented in a dense 38 page report commissioned by the core dev team titled “Bancor v2.1 — Economic and Quantitative Finance Analysis. ” Based on Uniswap data, the study found that “over a sufficiently long-time horizon, the fees are expected to dominate the option value.” (the authors refer to the IL insurance as an “option” — which is paid to LPs when they withdraws liquidity). If the study’s analysis is right, the protocol will create a deflationary environment for BNT supply, while providing higher aggregate net yields on liquidity pools vs. other AMMs (e.g. Uniswap and Balancer) since IL is no longer a drag on yields from fees or liquidity mining rewards.
I first wrote about Bancor in February 2018, with the title “The Pundits Were Wrong, Bancor Solves A Big Problem, and is Scaling Rapidly.” While I was correct that Bancor did solve a big problem, Uniswap ended up solving it in a way that was easier for LPs, simply by allowing LPs to use ETH in liquidity pools, instead of BNT. So to provide liquidity to the LINK token for example, LPs on Uniswap provide LINK+ETH in equal values, instead of providing LINK+BNT in equal values on Bancor. And because most LPs on Ethereum were already holding ETH, it was a way better product.
In response, Bancor’s v2.1 took it one step further, removing the requirement that LPs also provide ETH or BNT in addition to the risk asset. Now LPs can simply supply one ERC20 asset on its own (i.e., just LINK or just YFI) and collect swap fees and liquidity mining rewards.
So while Bancor’s use of BNT provided by LPs for each of its pairs was historically a burden vs Uniswap’s use of ETH, now, since it’s provided by the protocol, the use of BNT enables innovative features like single-asset exposure and insurance against impermanent loss, which appears to provide higher yields for LPs.
This could signal a new day for AMMs — one in which there is less friction, less risk and consistently higher profits for LPs. You can try it for yourself by following these directions.
The key unknowns are: 1) whether the returns on Bancor pools will be high enough to attractive LPs to the protocol in droves and 2) whether the fees generated on the protocol’s co-investments will be enough to offset the cost of insurance. Only time will tell. But it’s a bet I’m excited to make .
If you thought this was worth at least .000001 Bitcoin, PLEASE thank us by clapping below (up to 50 times). Thx!
