Cost of POL for INDEXcoop

Product
1

This is a brain dump I wrote a couple of days ago…

Providing liquidity on DEX whether by individuals or protocols has costs. However, the headline $ value lost can have multiple (overlapping causes):

  • Market price exposure
  • Impermanent loss on a constant product ratio (xy=K) DEX
  • Accelerated IL on uni v3

Market price exposure (for anything other than stable coins).

  • In the last 180 days we have seen the following (coingecko spot prices):
    • ETH -32%
    • INDEX -537
    • DPI -50%
    • MVI -54%
    • DATA -53%
    • GMI (since launch 113 days) -62%

So, any portfolio (including POL) constructed from these assets could be expected to have lost 30 to 50% of its $USD value over the last 6 months (which is the market that the liquidity pod has been working in).

If we work in $ETH terms, and use MVI as a proxy, then MVI has lost 31% vs ETH, so a 50:50 (cold wallet HODL) portfolio would be down 15% in ETH terms over 6 months.

AUM allocated

Coupled with the large AUM needed for DEX liquidity ($5 M of MVI:ETH would give a 1% price impact trade ~$25,000. [with another 0.3% LP fee for the trader]).

  • $5 M of assets losing 30 to 50% in USD over 6 months is painful.
  • 1,156 ETH of assets ($5 M 180 days ago) losing 13% is also painful (~150 ETH lost on the MVI + ETH HODL).

Impermanence loss

Token pairs on xy=k constant product DEX (uni v2 etc) also suffer IL. However, it’s not as big a deal as some think. An x2 price divergence results in a 5.3% loss vs HODL.

For the MVI ETH case above a 31% drop of MVI vs ETH creates a 1.7% IL.

So, If you LP’ed $1,000,000 MVI ETH on uni v2 for the last 180 days you would have lost:

  • 16% due to ETH price action (-$160,000)
  • 27% due to MVI price action (-$270,000)
  • An additional 1.7% due to IL (-$17,000)
  • Total = -44.7% (-$447,000)

Note: You would have earnt some LP fees.

If the bear market had been reversed, then the IL loss would still happen, but the price gains would drown it out.

The key point for indexcoop:

For a constant product DEX (xy=k), IL is not great, but price action dominates liquidity pod PnL.

One other point is the ILis only crystalised when you exit the position, if MVI gains vs ETH in the future, then a passive LP position will see IL reduce.

Concentrated liquidity

Concentrated liquidity is great because it lowers the AUM needed to give a specific market depth. $100,000 of liquidity can have the same trade experience as $1,000,000 of xy=k.

However, IMHO it has three downsides:

  • v3 LP’ing is a competitive game, a more concentrated and active LP will capture an above-average share of fees (see the Big Friendly Whale in the icETH pool below).
  • Active management costs attention and Gas
    • Or use of a 3rd party manager service with smart contract risk / fees.
      • Note managers take a share of the fees, but suffer no IL.
  • Concentrated positions concentrate the IL. So if the prices diverge over time, the LP has to sell the underperforming asset to buy the other (sell low buy high).
    • This works both ways: if a pair diverge 30% and then reverts to the original ratio.
      • The V2 position has zero IL.
      • The active v3 position has crystallised some IL each time they reposition (whether manual or using a manager such as G-UNI or VISR).

1508×622 45 KB

Some additional reading:

I think that my blog differs in that it looks at the costs of repositioning an LP position, I’m not sure the others tried to do this (but they are either wide all wallet reviews or a mathematical prediction)

INDEXCoop concentrated liquidity
So, by using concentrated liquidity, we save capital, and so reduce the total $ value loss in a bear market. However, the IL may well be higher % relative to $ deployed (and possibly higher in $ terms than if we had much more AUM in a constant ratio pool).

I suppose you could consider concentrated liquidity as leveraged liquidity, more depth = more exposure to fee income. There is no risk of liquidation. However, forced periodic rebalancing will bleed NAV decay just like a FLI product (an analogy that probably only works for 6 people in the world…).

Note for Yielding products paired with their parent token, then the divergence should be predictable and one direction so it becomes easier (I’ll be writing a blog on my icETH positions).

Obligatory Bancor comment (:wink:)

For a protocols governance token, depositing their native (i.e. $INDEX) treasury tokens into Bancor effectively solves POL:

  • Uses idle treasury tokens
  • Provides liquidity
  • No need to buy other token ($ETH) so can be deep at little cost to the protocol.
  • No IL
  • Fully passive
  • THe impact of price movement is identical to holding in the treasury multi-sig.

However, for coop products, liquidity pod would need to buy the products, so it would have a significant capital cost (and is a constant product so needs more capital).

There is also the small point that Bancor doesn’t currently whitelist structured products like DPI. but I’m working on this. :wink:

10 Likes
2

@ElliottWatts It would be great to see the POL IL and depreciation by product in the financial reports. It’s one of our largest product costs and financial risks but it’s not shown anywhere.

5 Likes
3

We will integrate this into the report for this month. The loss is easily calculated once a position is exited and the loss crystallised. The challenge comes if we want live tracking on IL across all positions. The liquidity pod and analytics are working on a solution to this cc @MrMadila @anthonyb.eth

6 Likes
4

Thanks @overanalyser , always learn from your liquidity posts and looking forward to the one on icETH positions!

3 Likes
5

@JosephKnecht I Know that @MrMadila, the analysis team, and the finance nest have been looking at quantifying the costs of POL. However, it is a fiendishly complex beast. [Which is why I did a much simpler model above].

One of the key points I was trying to make is that it is easy to see the coops POL as a financial disaster. However, a significant part of this is due to a combination of two things:

  1. the coop does all finances in $USD
  2. We are in a bear market.

My hypothesis is that if we had been in a bull market and our total POL was going up in USD terms then most people would be content with a +ve headline number.

Don’t get me wrong, POL requires lots of capital and it exposes that capital to market prices, IL and concentrated liquidity can crystalise IL.

6

OK, I think I’ve been nerd sniped. :roll_eyes:

Another anecdotal data point to consider.

I’ve been LP’ing icETH since it launched and can look at 3 positions since 5th April 2022:

  1. HODL icETH (not done, but can be calculated)
  2. LP from Spot to +1% 217675
  3. LP from 1% above Spot to +2% 215782

*Note: the idea was that the 1st LP position will slowly go out of range as icETH increases vs ETH, and so the WETH can then be used to buy icETH to reposition in a 1% range above the LP 2… *

HODL

image
image720×121 4.03 KB

LP 1 (1% above starting spot)

image
image717×112 4.49 KB

LP 2 (from 1% above initial spot to 2% above spot - currently out of range)

image
image716×112 4.32 KB

Comparison of profitability
image

Cost of repositioning: If I exited LP 1 and swapped to 100% icETH and placed a new LP position, I will incur a 0.05% pool fee, and spend~0.08 ETH on gas (~ 1% on gas). So LP 1 is not yet break even vs HODL after gas.

Cost of LP’ing:
If I exit either LP position today, I would realise a 13% loss (in $ terms)

Where I think that this is relevant to INDEXcoop POL is that even though the LP positions are 13% in loss, I’m quite happy to stay in these positions (even if that means I’m forced to exit at a $ loss).

1 Like
7

I appreciate that estimating unrealized Uni v3 PnL is very difficult and that Anthony and Andrew are working hard on it and even made a recent breakthrough. I’ll defer to Finance Nest whether it’s better to use current estimation methods or wait until we have an accurate calculation.

One partial solution to the risk of holding POL is to pay contributors partially in a fixed number of product tokens at contributors’ preference. This would transfer the product risk/reward to contributors, increase TVL and holders, reduce INDEX sell pressure, create a strong collective incentive to design better products, and reduce gas costs and slippage to whatever extent we swap product income for ETH and stables. Even paying only 5-10% of contributor salary in a fixed amount of product tokens would reduce the risk of POL substantially.

@ElliottWatts Wen product tokens for salary?

3 Likes
8

Finance Nest are looking at implementing the option of offering product tokens for salary. We have in recent months implemented this for the launches of icETH and JPG. We will provide an update before the next round of rewards.

4 Likes
9

Like many, I would welcome the opportunity to take part fo my coop rewards in product (or LP) positions.

As I see it, it would give the coop a wider LP base, and lets contributors support our products (and hopefully earn some fees).

However, everyone should be aware of the costs and risks and the benefit of staying in positions long term. Personally speaking, a main net LP position < $5,000 USD, isn’t worth entering. So it would take me many months to accumulate something decent if I only directed a fraction of my rewards to it each month.

4 Likes