Rebuilding Bancor3 AUM and reputation

I’ve been thinking about how we transition Bancor3 from the current state to where we want to be. Hopefully, I can put this into a coherent post for others to discuss :slight_smile:

At the moment the majority of TKN LP’s in bancor 3 are trapped by the individual pool deficits and receive low yield from trading fees. The protocol is focused on maximizing fee income to burning vBNT / BNT to generate a positive pressure on BNT price and so reduce and eventually remove the deficit. This is being done by votrexing all Bancor 2.1 fees and 90% of B3 fees. The alunch of Carbon will also generate more fee income directed to BNT price.

This is all good, and a proper focus for the protocol at this time.

Where do we want Bancor3?

In the future we want to have large AUM in Bancor 3 generating lots of trade volume and fees for LP’s. Then we can build other features onto this to further generate income for Bancor protocol and BNT holders. However, for most of these uses, we need large AUM in B3. (Note I think Carbon AUM will be held separately so does not contribute to B3 AUM).

So, I’ve been thinking about how we get from current to future.

I think the short answer is that we need two things:

  1. BNT price to go up :slight_smile: as arbitrage will pull TKN into the omnipool.
  2. Token LP’s to deposit more TKN into Bancor3.

#1 is being worked on.
#2 is more complicated.

I’ve recently posted about DAI here. My conncertn is that when BNT passes ~ $0.53 the DAI pool will be in surplus, earning ~0.3% APY and the majority of LP’s will remove DAI as they can get >1% in Maker, AAVE or Compound. Yields will increase as AUM drops, but loss of AUM impacts trade volumes.

For me, in order for most LP’s to deposit more TKN into Bancor3 they need 3 thinks:

  1. confidence in the Smart contact :white_check_mark:
  2. confidence they can withdraw all their token in the future :slightly_frowning_face:
  3. yield in TKN that compensates for their opportunity cost :slightly_frowning_face:

The good news is we have robust smart contracts.

Confidence in full TKN withdraws has been shattered by the (entirely correct) pause in IL protection.

Yields are poor due (in part) to the 90% Vortex.

Note, I think I’m in agreement with @cryptokitty in focusing on needing AUM to work with.

TKN 7 day APY
ETH 0.04%
wBTC 0.06%
DAI 0.17%
USDC 0.14%
USDT 0.12%
LINK 0.09%
BNT 0.20%

Building confidence and TKN AUM

I don’t think that we are approaching the time where we need to start experimenting with some pools with the aim to:

  • Build confidence
  • Build TKN AUM
  • Build TKN LP income.

and that we should do this while continuing the direct the majority of the total fees to the vortex…

Note, that this may require some smart contract changes to allow individual pools to have different parameters. At the moment I’m assuming anything I want is possible :wink:. Reality may mean we need to modify plans or divert Dev time.

I’ve already posted about DAI,, to me this is a defensive strategy to defend DAI AUM, and create a playbook for USDC and USDT as BNT price increases and those pools migrate to a surplus.

What about the other TKN’s that maybe more correlated with ETH / BTC / BNT / crypto?

I think we need to start by doing two things for selected TKN’s:

  • Restore IL protection
  • Increase LP income to become more competitive.

Restoring IL protection is a big move, it returns Bancor to to one of our key value propositions. However, it also risks opening the gate to significant BNT minting (and likely reawakens some of the Bancor detractors…)

So where do we start?

My starting point was to download the Bancor3 analytics and looking at the different pools. The first filter was the surplus. If we restore ILP to tokens already in surplus, there is a small risk that IL protection will be needed.

TKN % Surplus
SHEESHA 1638807%
YFI 7799%
OCEAN 2241%
EDEN 1469%
GRT 1294%
SNX 826%
SATA 211%
RNB 178%
REN 161%
LPL 94%
DAPP 93%

I’ve take all TKN with more than 50% surplus in this table. It may be more prudent to take 200%, but 50% only adds 4 more to the screen.

My second screen would be to sort by LP TKN AUM and remove those tokens with less than $20,000. e.g. If there is a 100% surplus, but only $10 k LP AUM, then a $40 K deposit will reduce the surplus to 17%.

TKN LP TKN $ Surplus $ Total $ % Surplus
OCEAN $60,253 $1,350,089 $1,410,342 2241%
SNX $92,049 $760,069 $852,118 826%
REN $217,256 $350,814 $568,070 161%
YFI $7,012 $546,901 $553,914 7799%
EDEN $27,715 $407,061 $434,776 1469%
GRT $19,942 $258,005 $277,947 1294%
ARCONA $34,845 $159,151 $193,995 457%
DAPP $61,771 $57,298 $119,069 93%
LPL $15,824 $14,926 $30,751 94% < $20 k
RNB $6,758 $12,032 $18,790 178% < $20 k
SATA $2,742 $5,783 $8,525 211% < $20 k
SHEESHA $0 $6,808 $6,808 1638807% < $20 k

Then I would look at the trading volume, and fees:

TKN LP TKN $ Surplus $ Total $ % Surplus Fee 7 day ave LP fee 7 day volume 1 day volume
OCEAN $60,253 $1,350,089 $1,410,342 2241% 0.50% 0.37% $64,609 $9,230
SNX $92,049 $760,069 $852,118 826% 1.00% 1.93% $90,445 $12,921
REN $217,256 $350,814 $568,070 161% 0.50% 0.54% $199,178 $28,454
YFI $7,012 $546,901 $553,914 7799% 0.50% 4.10% $26,273 $3,753
EDEN $27,715 $407,061 $434,776 1469% 0.50% 0.60% $24,512 $3,502
GRT $19,942 $258,005 $277,947 1294% 0.50% 8.19% $144,840 $20,691
ARCONA $34,845 $159,151 $193,995 457% 5.00% 3.32% $23,471 $3,353
DAPP $61,771 $57,298 $119,069 93% 0.50% 46.58% $0 $0 Boosted

From this it’s obvious that DAPP is different, zero trade volume, but high yield (from Protocol liquidity incentives). As such I would remove it from consideration.

That leaves a shortlist of 7 tokens. At the point the Bancor DAO could decide to offer IL protection to all, or some based on further criteria:

  • Relationships with the TKN DAO
  • Is a project dead? (after all the TKN has under performed BNT since it was introduced…).
  • Offchain awareness (e.g. whats happening to REN following FTT…)
  • How we compare with other on chain liquidity and volume.
  • How much risk do we want from this operation, fewer pools / higher surplus reduces our exposure to Minting BNT (note we can also modify the BNT trading limit to preent a sudden TKN deposuit and price increase resulting an a deficit).

Changes we can make:

I would say that once we have identified the pools to start on we could:

  • Activate IL protection.
  • Modify the BNT trading limit (current TKN $ value including surplus???)
  • Change the pool fee to increase volume and / or income.
  • Change the vortex to increase LP income for a given pool income.

Suggested KPI’s

  • Growth in bnTKN showing an overall gain in TKN deposits for each pool.
  • Potentially increased LP income (depending on how we set the fee and vortex).
  • Positive sentiment from the crypto community for survival and recovering .

So, some questions:

  1. Do we think the time is right to start reintroducing ILP for some pools?
  2. What criteria should we select TKN’s?
  3. Can we modify all the parameters we would like to (or are they global)?

A related question, what is the right time to restart outreach to other protocols looking to add new tokens to Bancor3? (or did it never stop?)

Getting more tokens listed (even if Treasury tokens as liquidity) would be nice.


Thanks for this post which is very well organized. I will try to share some of my personal thoughts below:

I think in general people will not deposit into v3 even with liquidity protection being reenabled in selected pools due to the events of 6/19/22. Take any of the tokens that have a surplus in v3 and liquidity protection gets reenabled again, depositors in any of these pools are always betting that their token outperforms the market (they wouldn’t be holding these token(s) otherwise). If these tokens do a 10x, 20x, 30x, etc… and the pools are in deficit then BNT will always be minted at withdrawal to cover for this loss. Depending on the size of the deficit and the amount of TVL in these pools, the payment could become quite sizeable and obviously lead to large amounts of BNT being minted that will later on get sold to recoup the original tokens that these LPs deposited. The Bancor DAO will then be forced to again consider either the health of the protocol and all LPs vs minting BNT to cover any sizeable losses. I personally don’t think we should ever have to be in this position again and for that reason will not support any BNT liquidity protection in the future.

I do think that the future of Bancor is Carbon and any other financial products that gets released in the Bancor umbrella. The main value proposition of v2.1 and v3 was found to be economically unsustainable and as such while these protocols are still working they are no longer attractive to the market. Does this mean that we have to give up on some of the innovations in v3?

Maybe not, perhaps having symmetrical pools (with two or more assets) with single sided liquidity (LPs are exposed to all assets in the pool), LP tokens that auto compound fees and rewards, BNT not being a requirement in all pools, constant sum/constant product/anything in between (maybe even liquidity concentration at the price point), etc… might be something that the market is interested in and perhaps can lead to a different type of product that gets released under the Bancor umbrella.



I had assumed that everyone would want to get the omnipool back into liquidity blackhole and be generating liquidity and income.

The alternative is that as BNT price grows, we will look TKN LP’s and so AUM will shrink until it’s effectively protocol owned TKN being being liquid for fees that goto the vortex. Possibly with a cheeky LP like YFI at the moment, ~$7 k of LP token, capturing the 10% of fees from %500 K of liquidity (and an nice 8,000% surplus).

I think that the ability to set the trading Depth by adjusting the BNT trading limit was never really tested due to the events last year.

I still think that being a liquidity source for DAO’s was an excellent use case for Bancor3 although it does need control. Projects that moon drain IL, those that fail bleed BNT into the market as we become the buyer of last resort.


Another question that I’m still working on:

Why haven’t the LP’s in those 7 pools that are in surplus exited Bancor3? Are the fees (Boosted by 10% of the income for the surplus) sufficient to keep them in?

1 or 2 may be DAO’s providing TKN liquidity with little cost to the DAO? But not all 7.


Ok, while I was thinking about the op, I did some digging into the different pools. This is a bit of a tangent into liquidity, but I thought it was interesting.

What I’ve done is look at the tokens of interest, the different liquidity on main net, the relative trade depth, and the daily volume (spot check on coingecko).


B3 “pool AUM” is double the $ value of token, so I’m treating it more like a Bancor 2.1 pool with TKN:BNT and all the trading depth available.

OK, so Bancor3 has more liquidity AUM than Uni V3, Uni v2, Sushi and Balancer for:

  • SNX
  • REN
  • LPL

This is good as it means we should capture lots of trade.

Univ2, Sushi, Balancer 50:50 pools and Bancor are all constant product AMM’s, This means that comparing AUM, gives a comparison of the actual trading liquidity at the market price. UNI v3 is different. So, we need to do something different. How I compare pools is to look at the size of trade that produces a 0.5% slippage (i.e. the price is 0.5% different to the spot price). For a $1 M constant product pool, this would be $2,530.

For example:

SNX on Sushswap and Uni v3 have $ trade size that are in proportion to their AUM, while the Uni v3 0.3% fee pool has a larger trade size (i.e. more depth) than the v2 pool, while having less AUM.

For the shortlisted tokens, I’ve noted the trade size:

The only real changes should be Uni v3. Compared to a cosntant product AMM (like B3) Uni v3 has:

  • much better trade depth for OCEAN, SNX, YFI, GRT.
  • Variable liquidity for REN

Note, REN liquidity has improved since I logged this yesterday (when it was ~$150). Looking at it’s apparent that REN liquidity has a poor profile (plots like this explain part of my dislike for Uni v3…).

In any case, looking at the available trade depth should give us an idea where on chain trades are happening and which pools Bancor would be getting Arb trades from (and where we dominate the liquidity). It also indicates that the 4 lowest AUM tokens in the screen actually have minimal liquidity on any of the main AMM’s. There could be an omission in my inital scan, further digging on coingecko or etherscan may show other AMM pools.

The final step is to look on coingecko and compare trade volume (coingecko is not perfect, as there are some gaps) Any B3 pools highlighted below are base don the Bancor app data and look high (possible due to a spike in overall trade volume earlier in the week).

From this it’s clear the Uni v3 stands out with more volume than the depth alone may suggest. I that the comparison with bancor is a combination of:

  • People just trading on the Uniswap interface.
  • People trading on trade aggregators that don’t include Bancor liquidity.
  • CEX arbitrage bots integrating Uniswap pools and not Bancor.
  • Higher fees for the Bancor pools.

Note, while Bancor has the highest TKN in our pool,for many tokens, we only dominate volume for:

  • Arcona
  • LPL

Note that the LPL pool has a 5% fee, so it is printing money for any onchain activity is likely driving the vortex.

I know that both AUM and volume can be considered vanity metrics for AMM’s. Even so, understanding where the liquidity and volume are can help the Bancor DAO decide how to manage pools.

One final note, on chain liquidity is a problem for many tokens. Many have insufficient liquidity to be safely used in Defi. If a $10,000 trade moves the price by 0.5% a $100,000 dollar liquidation is going to ruin someones day. Liquidity is Important.

Edited to add my spreadsheet


Hey @OverAnalyser
I shared most of this on DM with you, but thought I’d re-post here for visibility.

Firstly thank you for thinking through all this, and here are my my two cents. I do not think Bancor v3 is viable after the economic vulnerability rendered ILP untenable. ILP was the main offering in Bancor v3 and from my perspective it is unfixable, so I see little reason for new LPs to deposit into v2.1/v3. The main goal for B3 in my opinion should be to repair the remaining deficit via Carbon fees, and then let it sunset.

Currently I see all the energy and innovation pouring into Carbon, and I believe Carbon is the most effective way for the Bancor community to recover and thrive.

I guess sometimes you have to let everything burn to the ground and use the lessons of your failures to hit restart. What was discovered in the depths of recovering the protocol (particularly research on a dynamic fee) ultimately led to Carbon and could prove to be a better way of driving sustainable on-chain liquidity and offering users a truly valuable service.

And I don’t think Carbon’s service is only valuable for “pro traders”. Personally I wouldn’t classify myself as a “pro trader” (and I’m of course biased), but Carbon is looking very interesting for me because while I tend to passively hold tokens for the mid/long term, I do like to plan my entry and exit targets, and capitalize on my intuitions about the range in which a given token might trade. That is effectively what Carbon offers - an automated buy low, sell high “swing trading” machine, which could also be a more effective and less risky way to drive on-chain liquidity, as opposed to liquidity being wholly dependent on LPs who are exposed to IL. LPs in Carbon are motivated to provide liquidity in order to execute trades at specific prices, and don’t care about IL/fees.

Sharing some links below for more background on Carbon & Bancor. Excited to get your feedback as it evolves and the beta goes online.


More on Carbon: