Checking sentiment on impermanent loss insurance.
Let’s say we can offset impermanent loss by adding insurance – which uniswap pairs should be included?
Will collect pairs mentioned in this thread and create a multiple choice voting in ~24 hours.
UPD: Since not that many submissions I just put the list together:
General:
DeFI:
Other:
Eth/usdc & eth/dai I think are the obvious pairs
Renbtc - eth pair would also be exciting! Ampl/ETH pair coverage could create an insane arbitrage opportunity if we can find a way to make it profitable (cautious of negative rebase and fund distribution for team and ecosystem).
So I think if we’re doing this as a Yearn product, it makes sense to target most of the high-liquidity and high-volume pairs that we consider legit.
Then it gets a bit trickier. There’s some other good options for neighboring projects such as PICKLE/ETH and FARM/USDC. While I think it’s beneficial to work with other projects in the space, my only concern would be based on how the insurance works. Since these are still relatively new projects, price swings (and subsequent IL) could be large, so that might be something to consider.
Perhaps other big DeFi players would be good as well, and also have high volume:
Lastly, DPI/ETH has had good volume for the past week and seems to be a cool new instrument but I’m not sure if we would want to include it in the first round.
My other consideration is a basic lack of understanding on how this would work. Is it a zero-sum game, where more pairs offered means less benefit to the other pairs? Or is it purely additive? If it’s the latter, then I would probably suggest throwing in a few more pairs—I’m just not sure what kind of target range for number of pairs we’re shooting for.
I think this is a great idea. How would this work in practice? Someone offer X amount of insurance for a particular pair struck at a particular ratio for some fixed period of time or does the insurance only cover losses beyond a particular barrier? For example I want to insure my position only beyond a down 50% or up 100% scenario? Obviously insuring the entire position for all losses may not be economical because particularly on stable coin pairs your are likely to run a few percent loss and your revenues from fees are in the low single digits to begin with (excluding UNI incentives)
I am for this and think it will attract a lot of AUM if we implement IL strategies and how to mitigate the downside of said strategies.
Other than the obvious ETH/WBTC, ETH/USDC, ETH/ USDT, ETH/DAI pairs we could add every major DeFi token: AAVE, YFI, UNI, SNX and LINK.
Including the DPI/ETH insurance as suggested above might also be in high demand.
this would be a great addition along with the introduction of IL strategies.
Collecting sentiment for now. Minimalistic readme is available in repo.
I think a phased rollout here makes sense with prio given to the most liquid LP’s (i.e. Uni farming). YFI-ETH would be a logical follower.
Ladies and Gents,
we keep in mind :
The increase in usage and participation in automated market makers has led to a vast set of new scoring rules and pricing mechanisms.
Analyzing these mechanisms, which range from LMSR style market makers and CFMMs to scoring rules for rates, from the perspective of optimization provides insight into why specific tools are more popular than others and work well in practice.
In particular, we show that CFMMs provide an easy optimization problem for arbitrageurs to synchronize off-chain and on-chain pricing data, along with several right conditions that often hold in practice, which imply that CFMMs are likely to be very well behaved.
This generalization encompasses all live CFMMs and guides designing CFMMs that are better for specific asset types and volatilities.
Thank you all.
WBTC/ETH and ETH/USDC pairs. Would support all pairs that have UNI farming, however.
WBTC/ETH has the lowest level of IL due to its relative low volatility for the pair (always subject to change).
+1 for renBTC/ETH
+1 for SNX/ETH
+1 for UNI/ETH
Apparently, more ETH is sent on-chain to uniswap and yfi amm pools per day than all CEX combined.
Best ones for LP and uniswap bots,
no brainer, we as LP have more skin in-game and take most of the risk.
Offset impermanent loss by adding insurance, you have my vote.
In my opinion, the obvious pairs:
ETH - renBTC
ETH - USDC
ETH - DAI
ETH - USDT
Defi:
ETH - UNI
ETH - YFI
ETH - AAVE
ETH -SNX
ETH - yUSD
Others:
ETH - PICKLE
ETH - LINK
Got it thanks. This is a much needed product and would be absolutely amazing. My only concern is that particularly for the stable coin pairs, it is going to be really expensive. The BTC/ETH pairs and the like will be much cheaper due to higher correlation between them. If insuring the full amount it is unlikely any insurance payout would always be zero because the pairs are unlikely to end exactly where the insurance was enacted. If you get the point where you see live pricing and its expensive consider adding a barrier. I think most LP providers would accept a few % IL loss, but want to avoid the 70-80-90% moves that cause exponential slippage. For example if you insured greater than a net 40% downside or net 66% upside (a 3% IL loss) on exchange rate would bring the cost down materially and make it look more like insurance where 100% of the premiums end up being profits to the insurance providers sans real “insurance events” in the case of extreme exchange rate moves.
Also If i’m taking on the full IL risk as an insurance provider why should I accept anything less than the return the LP provider is getting. I’m taking all the risk and the LP provider is basically arb’ing me. Only difference is my collateral. He is 50% token A / 50% token B and I am 100% some other token.
@YFI-Cent Your last paragraph is key and says it all. What @milkyklim essentially proposes is a put option with dynamic strike price. If we bring something crazy complex like this on the market, we can just as well program our own options market. 
What happend to the idea with Stable Credit and single sided liquidity? I think that is a way better solution for multiple reasons.