A quick update on this,
There is a small work team looking at Uni v3 and the pro’s and cons for users / LP’s and the coop.
A quick update on this,
There is a small work team looking at Uni v3 and the pro’s and cons for users / LP’s and the coop.
V3 work team analysis can be found here:
and I did some number crunching on uni v2 and v3 liquidity pools for DPI:ETH
With the end of the 90 liquidity mining campaigns for both MVI and DPI, my attention is turning back to two things:
And let us not forget that liquidity mining is expensive (we currently spend 20,296 INDEX over 30 days) = $300 k @ $15 INDEX, or $1.2 M @ $60.
There are lots of moving parts/interdependencies on this (which I think is why I find it so fascinating) and we are still learning:
Over the next couple of days I’m going to do some digging and will be sharing my thoughts on a number of areas around liquidity mining and our different products. The objective is as follows:
please get involved, share your thoughts and challenge me where you think I’m wrong.
I’m going to try and split my post into chunks because I think it will be both easier to write and comprehend. I’m going to start with MVI, because I think it’s the simplest liquidity (sector product in one DEX).
Before I get to the coops considerations for LM, I thought I would capture some insights into being an LP on this pool.
On a standard xy=k AAM pool, (i.e. Uniswap v2) the LP provides equal $ values of each token to the pool and allows people to trade one for the other. As the price changes, the pool will accumulate more of the token that is dropping compared to the other. So, the LP is always at a disadvantage compared to holding the tokens in a cold wallet. This disadvantage (Divergence or impermanent loss) is countered by:
So, for a LP, they need to think in terms of the pair (MVI:ETH), the relative prices, trade volume in the pool and rewards.
note, all % are since launch on the 7th April, NOT annual
Since launch in April, staking the MVI:ETH pair had resulted in:
This 67% price drop (divergence), results in a 13.5% loss compared to buy and hold. (See loss calculator on coin gecko).
croco.finance for this pool indicates that if you were an LP since launch, the fees collected comprise 12.8% of the current value. I.e. the fees have added ~ 14% to the value held.
The INDEXcoo.com staking contract has resulted in index tokens worth ~5% of the current pool value (calculated @ $15 INDEX - claiming and selling INDEX at higher prices would have crystalised more gains.
This can be summarised as:
|MVI:ETH LP 07 April to 26 Jun 21 (80 days)|
|Divergence in price||67%|
|INDEX income (@$15 INDEX)||0.3%|
However, looking at the trade chart, it’s clear that the majority of the trading volume was early in the product life.
Looking at croco finance for the last 30 days indicates a current fee income of ~ 2.4% per year. [Daily average fee * 365 / pool value]
Overall, being an LP has got less attractive over the last 80 days.
A quick edit to say that while researching this post I’ve discovered that there is a small Uni v3 pool for MVI.
I’ve decided to add to move some of my MVI ETH liquidity to the v3 pool as a personal experiment.
Note, I think the TLV locked in v3 is incorrect, I think it is only counting the value of ETH in the pool. There is currently 520 MVI worth a total of ~$15 k to add to the $8 of ETH.
OK, MVI is 80 days old, and we launched with a 90 day campaign of liquidity mining rewards.
So, it’s time to look back at what we did, what happened, and look forward to whats next:
What have we done, the Liquidity mining campaign:
|Days||Target MVI:ETH AUM||Actual AUM at end ($ M)||INDEX rewards over 30 days||Change in # INDEX||TWAP price at start||cost at starting price|
|1 to 30||$ 5 M||6.4||3810||n/a||36.68||$139,751|
|31 – 60||$ 5 M||4.1||2743||-28%||37.81||$103,713|
|61 – 80||$ 5 M||2.6||3286||20%||35.8||$117,639|
Looking at JD’s MVI dashboard. we can see the value of MVI available as liquidity (note total liquidity will be twice this value) and that this has dropped significantly since the start of May.
At the moment, we have ~$2.5 M liquidity ($1.26 M x 2). Which is significantly less than we have been targeting since MVI was launched.
Looking at the change in LM rewards (above), there could be a decrease in liquidity (in USD terms) associated with the cut in Rewards in early May, but little sign of an increase in liquidity when INDEX tokens were increased in Mid June.
However, when we look at the number of LP tokens, which represent people actually adding or removing liquidity from the pool. LP’s appear largely insensitive to the changes in the number of tokens issued:
i.e. neither cutting # INDEX by 28% nor increasing by 20% resulted in significant addition or removal of liquidity (~10 May and ~10 June). In addition, the wide range in INDEX price between ~$60 in mid-May to ~$15 currently seen appears to have minimal impact on the number of LP tokens.
My thoughts on this is:
Based on this, I think that the moderate changes in INDEX rewards (+/- 30%) are unlikely to change the number of LP tokens we have for MVI:ETH.
How much Liquidity do we need:
In the past we have looked at $ value required to get a certain price impact on the pool.
|Impact||Trade size||Trade size (ETH)||Trade size MVI (@$30)|
So our 2% depth is ~ 50% of what we had at the end of May.
Trade volume was massive in April, but has been much lower for the past 60 days:
The last 30 days look like this, with the average day being maybe $100 k of total volume:
Looking at the pool data from etherscan since the 27th of April, we have 2022 transactions and I’ve done a very rough manual analysis.
Of the 2,022, I’ve noted 5 trades over 700 MVI (with the largest at 2,114 MVI). and a further 15 between 500 and 700 MVI.
Personally, I would describe this fraction of trades that could have large price impacts as acceptable
As a result, I don’t see a strong reason for increasing or decreasing the number of INDEX tokens allocated for the next 30 days (i.e. 3,286 over 30 days starting ~ 10 June 2021).
However, I’m looking forward to hearing other people opinions on this matter.
Ok, the FLI products are very different to DPI and MVI, they are designed to be traded and only held for short periods.
ETH2-FLI was launched with a cap on units which was rapidly reached and the cap raised. This rapid growth of AUM and liquidity mean that the planned liquidity mining for ETH2-FLI was put on hold.
ETH2-FLI AUM and supply
An initial observation is that while the AUM shows a pretty wild ride with AUM peaking over $120 M:
The number of circulating tokens continues to increase.
There is currently significant liquidity and trade volume on both Uniswap v2 and v3. Note, I’m not familiar with Uni v3 analytics, so I’ll compare the data from Uniswap, and look and v2 using Dune
|27jun21||Unit||Uni v2||Uni v3|
|14 day average trade||Million $||1.96||4.26|
|14 day average fee||$||5,867||12,765|
|Annualised average fee income||%||46%||76%|
|ETH2-FLI in pool||# ETH2-FLI||43,375||97,530|
|ETH in pool||ETH||1,269||530|
|ETH2-FLI in pool||$ M USD||$2.4||$5.3|
|ETH in pool||$ M USD||$2.3||$1.0|
|Sell ETH for 1% price impact||ETH||9.0||20.0|
|Buy ETH for 1% price impact||ETH||9.0||24.5|
|Sell ETH for 2% price impact||ETH||22.2||54.0|
|Buy ETH for 2% price impact||ETH||22.2||51.0|
One key consideration is that, unlike v2, v3 liquidity is not symmetrical, so price impact can not be easily calculated. For example, at the moment, the v3 pool has much more ETH2-FLI locked ($5.2 ) than ETH ($1.0 M)
Even so, it’s clear that traders get better prices on v3, and LPs are (on average) getting better income from v3.
Using the flipside v3 calculator it’s possible to identify liquidity bounds that will produce returns higher than the 76$ average (but need more active management to stay in range).
Historical Uniswap v2 liquidity
While the majority of liquidity and trade volume is on v3, the v2 pool has run for the entire product life and is easier to analyse.
This shows the number of v2 LP tokens increasing until early May when it declined rapidly (when the v3 pool was launched), since late May there has been little movement in the number of LP tokens (although the number of ETH2-FLI in the pool will have increased as the price relative to ETH dropped).
Do we need to provide liquidity mining for ETH2-FLI?
The market is currently producing liquidity pools which allow ~30 ETH (~$55,000) trades at 1% price impact. and ~75 ETH trades at 2%.
While I’ve not done a detailed review of trades on Etherscan, it’s clear that there are a number of swaps happening in this range (e.g. 50 ETH on v2 and v3 via paraswap)
However, given that LP’s are currently enjoying rewards of ~50%, and we already have $11 M of liquidity, I would expect any campaign designed to result in a significant increase in Liquidity would be expensive in terms of INDEX tokens (+10$ M of v2 liquidity @ 50% APY for 30 days could be $410 K ~ 24,000 INDEX).
As such, I do not see Liquidity mining ETH2-FLI (on L1) as being the best use of the coops INDEX resources at this time.
However, I’m happy to hear others opinions (or read a review of the ETH2-FLI trade sizes…).
OK, BTC2-FLI is next. This launched almost at the peak of the bull market with liquidity on sushiswap (which behaves just like uni v2).
Unit supply climbed rapidly on launch and has been largely flat since then:
AUM shows a similar trend with the drop in prices resulting in AUM being lower than it was 2 weeks ago.
Looking at the sushiswap pool, we see the following:
|Liquidity AUM||Million $||5.66|
|14 day average trade||Million $||0.90|
|14 day average fee||$||2,689|
|Annualised average fee income||%||17%|
|Annualised Sushi rewards||%||24%|
|Total rewards for LP’s||%||41%|
|BTC2-FLI in pool||# BTC2-FLI||115,800|
|wBTC in pool||wBTC||90|
|BTC2-FLI in pool||$ M USD||$2.8|
|BTC in pool||$ M USD||$2.8|
|Sell BTC for 1% price impact||$||$31,300|
|Buy BTC for 1% price impact||$||$31,300|
|Sell BTC for 2% price impact||$||$62,900|
|Buy BTC for 2% price impact||$||$62,900|
Overall, ~50% of the total supply is in the liquidity pool available to trade:
Comparing the BTC2-FLI with the ETH2 liquidity a few things stand out:
I think that these are all related to the provision of $SUSHI rewards for the pool. This also leads me to think that:
An interesting question is whether a Uni v3 pool would result in more trading as arbitrage could open between the pools)
The Sushiswap Onsen programme is currently allocating 572.3 SUSHI per day to this pool ($3,800 or equivalent to 6,800 INDEX over a 30 day cycle).
Comparison with ETH2-FLI
|Market cap||$ M USD||29||5.5|
|DEX liquidity (pair)||$ M USD||11||5.6|
|trade volume||$ M USD||4.7||0.5|
|Volume / liquidity||%||43%||9%|
While ETH2 is ~6 times larger than BTC2 in total AUM, it enjoys ~x10 more trade volume.
The smaller pool for BTC2 means that price impact becomes a consideration much sooner. The current range of $31 to $62 k for 1 to 2% impact is probably sufficient. However, if the pool shrinks further (e.g. due to ONSEN rewards stopping), then improving the depth (liquidity minining or encouraging uniswap v3) will need to be considered.
(note, I’ve not done a full review of trade sizes on Etherscan so I don’t know if large high price impact trades are happening)
Do we need to do liquidity mining on BTC2-FLI?
With the continued support from sushiswap, I see no immediate need for INDEXcoop to provide rewards for this product on L1.
OK, I’ve been saving the best for last. $DPI is our largest product, with the most liquidity and has enjoyed significant liquidity mining incentives since launch. As such, it’s one of the coops largest costs with 567 INDEX being given to LPs every day ($8,505 per day = $3,100,000 per year @ $15 INDEX).
The uni v3 research team has already recommended some options for DPI on-chain liquidity
Current on-chain liquidity
|27jun21||Unit||Sushi||Uni v2||Uni v3|
|14 day average trade||Million $||0.17||1.61||0.20|
|14 day average fee||$||432||4,828||594|
|Annualised average fee income||%||5.5%||5.9%||10.5%|
|LM incentives (current)||%||12.9%||12.1%||0%|
|DPI in pool||# DPI||6651||69,204||7,620|
|ETH in pool||ETH||793||8,221||232|
|DPI in pool||$ M USD||$1.4||$14.8||$1.6|
|ETH in pool||$ M USD||$1.4||$15.0||$0.4|
|Sell ETH for 1% price impact||ETH||8||58.5||9.6|
|Buy ETH for 1% price impact||ETH||8||58.0||11.8|
|Sell ETH for 2% price impact||ETH||16.3||143.0||25.8|
|Buy ETH for 2% price impact||ETH||15.9||140.0||24.0|
Note, that the uni V2 liquidity pool is significantly lower than the $55 M we have been targeting for the last 90 days. This is primarily due to the drop in both ETH and DPI prices. Looking at the number of LP tokens we see a steady reduction by about 30% since early April (28,405 LP tokens on the 4th April compared to 20,269 today)
Recent trading volumes have been lower than we have typically seen. Even so, the uni v3 LP’s are getting higher fee income than the v2 or Sushi pool, however, the average fees on v3 are not sufficient to exceed the effect of INDEX or SUSHI rewards.
In part, this is due to the liquidity being spread very wide, and a significant amount in ranges that do not cover the current price:
The flipside uni v3 calculator calculates a fee-only yield of ~60% for liquidity +/- 8% of the spot price.
Previously the focus for liquidity mining has been on the price impact of large trades. With the goal being to have a large pool that can allow large trades to be carried out with minimal price impact, and efficient arbitrage trades to keep the price close to the NAV.
However, when looking at recent trades, its clear that many large purchasers are making full use of aggregators who split the swaps over multiple pools. These aggregators also access stablecoin:DPI liquidity on 0x protocol to give the best prices.
If we look at 1inch, for large trades (100 ETH), many liquidity sources are used. In this case, less than 12% of the total volume was routed through the Uniswap v2 pool!
So, for educated whales, the size of the uni v2 liquidity pool is not of critical importance.
(Anything less than 24 ETH was routed through Uni v2, v3 and sushiswap)
Secondary effects of migration to uni v3
As discussed in the uni v3 research, moving DPI liquidity to uni v3 is expected to increase the depth of the market close to the spot price and the increased earnings for focused LP’s should allow INDEX rewards to be discontinued.
However, there are a number of secondary effects to be considered:
Personally, I think that moving DPI liquidity to v3 will be good for the coop, and token holders. There will be more work for LP’s but those that do the work will continue to receive income from providing liquidity.
The main question is how quickly we want to move, and a key consideration is the fact that the current liquidity mining for DPI:ETH on v2 ends in about 2 weeks and an IIP will required to release INDEX funds into the staking contract.
I am looking forward to hearing your thoughts.
OK, As DPI has the most options, I think some polls would be good to judge everyone’s thoughts and hopefully provoke some discussion.
So, some multiple option polls (please tick all options you would be happy with).
i.e. a continuation of the current 30 staking contract.
In addition to (/ to replace ) the uniswap v2 liquidity mining, we can look to help/encourage liquidity to move to uniswap v2.
Poll closed 06 Jul 2021 @ 15:15 UTC
Thanks for all this work @overanalyser ! I just wanted to quickly highlight a couple thoughts while they are fresh - might return to add more later if it looks helpful.
Your finding that only ~12% of 1inch volume on large trades was routed through v2 is enlightening. I have been a big proponent of the move to v3, but that just solidifies it. We need to stop incentivizing v2. It is not serving us. Even if that 12% starts moving to v3, the fees there will pick up and attract more active LPs.
The LPs that serve us on v3 will likely be very different than those on v2. And that is ok. v3 is a different game, but it is one that can serve us so much better than v2.
I think the v3 work team (myself included) needs to re-group and spec out more fully the migration so that we can get a better sense of all the ENG effort and other effort required and then we need to commit.
During one of the coop calls today the topic of the overlap of INDEX holders and liquidity providers was raised and that they could have a conflict of interest.
For the sake of absolute transparency, I would like to disclose that I have been farming the following coop tokens:
I’m also LPing (unincentivised):
In the past (but no longer) I have also LP’ed:
I plan to continue to provide liquidity with coop tokens where I see an opportunity to make money.
However, I will endeavour to contribute to the coop (PWG/ work on liquidity etc) with a focus on the coops long term growth.