Reviewers: @afromac, @allan.g
As described in the May 4 Leadership Forum, the Index Coop Product Nest is actively working on resolving the economic issues with composite index products. This post provides an update on those efforts.
Composite index products face several major economic challenges including high costs and financial risks for Index Coop, intense effort to manage rebalancing and recomposition, product asset decay, and negative overall profitability. Here we define ‘composite index product’ to mean any product made up of a diverse basket of tokens. These challenges are specific to composite index products because of the high turnover (i.e., rebalancing) and the lack of high correlation with the other token in the Liquidity Provider (LP) pair. This lack of correlation is important because it reduces the attractiveness of LPing and hence increases the need for Liquidity Mining Incentives (LMI) and/or Protocol-Owned Liquidity (POL).
The profitability challenge for composite products has been highlighted previously in several posts:
The profitability of the composite products can be found in these dashboards for Cost Revenue Breakdown and Product Profitability, as well as IC’s quarterly financial reports.
In total, the economic issues with composite products include the costs of LMI; the financial risk, capital requirements and impermanent loss from POL; and the gas costs, asset decay, and intense effort required for rebalancing and recomposition. Higher AUM does not resolve these issues since rebalancing costs also scale approximately with AUM. Moving to Layer 2 is not a solution either due to liquidity fragmentation, worse underlying liquidity, the remaining exposure to asset decay, and the limited market. Launching with exchange issuance only and no dex liquidity is commercially unproven and still does not address the issues with rebalancing.
Ongoing effort and next steps
The following efforts are underway to resolve the issues mentioned above.
||Continue to avoid
||Hedge with borrowed ETH or short position
|Rebalancing and recomposition
||Stop or curtail. Automate and access deeper underlying liquidity available on CEX’s through KeeperDAO’s Rook Protocol and/or Managed Balancer Pools.
It’s expected that KeeperDAO’s Rook Protocol and/or Managed Balancer Pools will be required to address the issues with rebalancing and so we will not be able to resume composite product launches until one of these are integrated into our technology stack. It’s not expected that either of these will be ready for composite product deployment before Q1 2023. This is a tentative estimate that reflects the early stages of research and development that we are in.
We still believe composite products have significant potential even though they are not economically feasible at this time. Research and development efforts are actively underway to resolve the costs and financial risks of composite products. We’ll report regularly on our progress and we look forward to resuming composite index products if and when they can be made financially viable.
Thanks for the clear update here. I’m incredibly proud to be working at Index Coop where we are taking these problems very seriously and are proactively researching how we can improve our products. Excellent work by you as usual!
Is there something about DPI that makes it so much more profitable than the other indices? Is it simply a function of relatively high TVL? It seems like GMI could be on a similar trajectory if TVL increased.
DPI has high underlying liquidity and relatively low turnover, but is very unprofitable after considering LMI. The Dune profitability dashboards above don’t include LMI costs or POL depreciation. No composite index product has ever been profitable even excluding contributor reward costs.
Thanks @JosephKnecht @afromac @allan.g ,
Appreciate the important work happening here.
I’m still not fully understanding the profitability/scaling forecasting (excuse any dumb questions). I’ve read through the previous posts and forecasting by @JosephKnecht and @prairiefi, but still having trouble grasping. If we remove the LMI, significantly reduce or temporarily remove the rebalancing, does this mean the products have the potential to start scaling efficiently in upward trending market conditions ? And the main remaining burdens are the POL IL and depreciation?
Is there a way to better understand the impact of POL on our products? I would be really grateful for a quick presentation from Product - dumbing down and helping us better understand POL and the impact of POL on profitability (or loss). I know you guys are busy, but something short/simple would be valuable to help us explain to others in the community.
What is the transition plan? Removing Liquidity incentives and rebalancing frequency are understood, but are there any ideas on how we hedge or transition existing products to the new pools? Are we going to test with an existing composite product and update the composition, or look to create a new test product/prototype?
Where is the overlap with Set V3? Is it possible for us to get a roadmap from Set and see how they’re thinking about solving the above? Would love to see that partnership continue and be able to launch new composite products with them.
Thanks in advance for response - no rush if anything above needs to be answered at a later point with the continued updates you mentioned.
Yes. The most promising strategies are to borrow the ETH against stables as finance has done for icETH; have an inverse hedged LP like @MrMadila did with a iETH-ETH2x Balancer Pool on Polygon; or, more experimentally, a fully delta-neutral LP using an adjustable short hedge; or launching with exchange issuance only and no dex liquidity.
The POL costs are not widely appreciated by the community because they’re not separated out on our financial reports nor on any IC Dune dashboards even though these costs are very large and variable. I think that’s a gap. The Liquidity Pod is working on tracking the POL costs but it’s technically a hard problem. I’ve asked finance to specify the realized POL costs in the financial statements. @MrMadila is perhaps best-placed to give a crash course.
There’s no intention to transition launched products to a new platform. We will however continue to work with external methodologists to revisit their rebalancing methods.
TokenSets V3 may have some features that will improve rebalancing. We’re in regular contact with Set and look forward to building together.
Hi @JosephKnecht , @afromac , @allan.g ,
I am pretty new to the community and have a lot to learn, but I read your update here and think I could offer some thoughts around rebalancing, from my professional experience. Improving transaction cost structures and market liquidity are definitely top of mind when it comes to aligning incentives with investors and minimizing operational expenses for Index Coop. I’d like to understand something better though, related to Index Coop’s Composite Indices:
From Product Performance - Surfacing Gas Costs, the post states:
“A single MVI or DPI rebalance typically requires 100 - 200 trades costing 20,000 to 40,000 USD in ETH. In total, automated & composite index rebalances have cost [more than $350k over the past 12 months] with the trend continuing upwards.”
Why are there 100-200 trades per monthly rebalance if the basket is composed of < 20 tokens? I know there is at least one thing I must be missing here, perhaps related to contracts or the structure of the index, so I will do my research and intend to fully understand that. My immediate next thought would be: could management of the tokens be consolidated, if it’s currently fragmented and causing a high volume of trades to be executed per rebalance?
I would really appreciate it if you could point me in the right direction to documentation on rebalancing methodologies or share more about how model selection and rebalancing is calculated and tested, beyond the methodology information published on the product websites. I ask because I am wondering how many different/which rebalancing methodolodies the Product & methodoligists groups (there are likely others involved I’m not omitting on purpose, I’m still learning my way around the Coop) have backtested to assess not only transaction costs but also impact to basket performance and its expected return and volatility distribution.
Considering a quarterly rebalance for Composite indices like DPI or MVI, with an additional condition of rebalancing if there is a one way 10% or 20% deviation (as an example) from model weight (calculated daily in a test environment) could yield not only transaction cost savings but also potentially better risk-adusted excess returns due to profit-taking and rotating out of high volatility tokens on a more opportunistic basis. Another idea to complement that would be to introduce a realized vs. implied volatility component to the model.
I would love to bounce ideas and debate on these topics if that would be helpful. I’m here as a personal investor and also someone with a tradfi background that sees the massive potential and believes in Index Coop’s mission & values. I’d like to be part of building something bigger and better than ourselves. I’d love to contribute whever possible but at the same time have a lot left to learn about Index Coop and its members so I don’t mean to overstep by sharing thoughts right away on the forum.
Welcome to the forum. We’re delighted to have you here. We always welcome thoughtful comments and it’s not ‘overstepping’ in the slightest.
The rebalances are done through 100s of small trades of ~100 bps tradesize each, instead of 1 large trade with maximal price impact. This is done in order to minimize the net price impact. If the token is arbitrageable across exchanges then each small trade will be arbed back to its base price and the net price impact will be lower than had one done 1 large trade.
Here are some video tutorials:
You might need to request access.
The answer is very little if any. I’m only aware of this analysis looking at different rebalancing periods for MVI. I’ve done internal calculations showing that the rebalancing premium tends to be negative for thematic indices even before the asset decay and gas costs of rebalancing. This result is not surprising considering that the components in thematic indices are very highly correlated but have very different long-term returns. If rebalancing is required for some reason, then you’re correct that there are probably benefits to looking at tolerance bands instead of calendar-based rebalancing.
Sounds great. I’ll dm and we can take it from there.
Why are we considering LMI when discussing $DPI or $MVI profitability. We haven’t done and LM on either product for over 6 months!
Yes, we allocated a huge number of INDEX to LM on $DPI (1% airdrop and 9% LM). However, that was specified in the genesis agreement for INDEXcoop in Septemner 2020.
There was no input from any of the current members of the coop, liquidity pod, product nest, Finance Nest or the index council when that 1,000,000 INDEX were assigned to $DPI.
Part of the goal of that distribution was to great massive sector dominating AUM, and attract a community to build $INDEXcoop.
Maybe we should assign all that token spend to community nest
Is this a reverse sunk cost fallacy?
Set labs and Defi Pulse decided to distribute significant tokens in the past as part of our genisis / to ensure credibly neutral token distribution therefore, we will constantly count those tokens against the profitability of the product.
$DPI generated ~$1,000 income yesterday with no cost to the coop.
$DPI generated ~$1,000 income the day before yesterday with no cost to the coop.
Why do we continue to count the expenditure of the past when we should learn those lessons and be looking at now and the future profitability?
I fully agree there is significant marketing value to providing LMI to boost TVL, even if it’s at a significant loss. I think we’re approaching the question from the same perspective which is, Based on what we’ve learned how can we set up our future products for success? It’s probably fair to say we’ve learned that LMI is very expensive and POL is very risky. We’re now working to address those. I hope that’s all sensible.
What I’m trying to push back against is the message that $DPI and $MVI are [currently] loss-making [because we have historically overspent a lot / on them].
The only costs (other than general marketing) the coop has wrt either is gas for rebalances, which are currently on hold. Obviously, this is something we need to solve, but there is no reason we can’t solve it.
It sounds like we agree that DPI and MVI are profitable on an ongoing basis but are deeply loss-making on a cumulative basis when including LMI. I think a clear PnL per product would help clarify these points in the future.
@JosephKnecht Thank you for the warm welcome and your helpful follow up. As soon as I started reading your answer re. rebalancing, a lightbulb went on in my head - thank you for the explanation there, minimizing price impact is very sensible and makes a lot of sense. That has not been a concern for tokens I tend to look at with significantly higher market cap & daily trading volume criteria, so that was my own anchoring. Appreciate you clarifying. Are you using TWAP over a specific period?
Thanks as well for sharing the resources and following up via DM. The Methodologist Tutorial I requested access for via my gmail, and the rest I’m checking out now.
Look forward to our next chat!
I assume you’re referring to the price/mcap determination for rebalancing. I think each product tends to have a different method, see ‘Determination Phase’.
Thanks, sorry for being vague - I meant if you use TWAP (Time Weighted Avg Price; there’s also VWAP which takes into account trading volume) or other methods for determining how you split up the trades during rebalancing events to minimize price impact. Here’s a helpful article, we can chat more when we meet in 30 min