Token Terminal Smart Beta Index

Thank you for your comments. Below are answers to the most recent questions raised during the weekend:

1. Token Terminal includes protocols that are not ERC20, but you have not explained why certain ERC20 protocols were included but not others (eg. SUSHI)?

Section 3.2 of the proposal includes an initial set of criteria sufficient to launch the proposed index, and is subject to additional requirements (if any) from Index Coop and/or Set Protocol.

Regarding SUSHI, it is among the assets to be included in the index. The reason it is not included yet, is due to the early stage of the project (see also answer to question #2 below).

2. What does it mean that the protocol must be recognised as having a high-quality product and team? Recognized by whom?

This requirement aims to ensure that the protocols included in the index are developed by a full-time team and/or community. Here, the Index Coop is able to step in to exclude assets that display undesirable characteristics or have yet to reach a sufficient maturity in their development.

3. What does it mean that the determination phase takes place during the third week of the month?

This refers to the monthly process of determining the necessary rebalances to the index. Technicalities regarding specific dates should be specified together with the Index Coop and Set Protocol to ensure best possible execution of necessary rebalances to the index.

4. I don’t see any criteria about what has to be done with sales revenue to be included here. Do all sales have to go into protocol treasury or distributed to token holders? MANA by Decentraland has pretty high sales (idk exact P/S) but all fees earned are used to burn MANA tokens, would this token be eligible to be included in TTI? Would ETH qualify once EIP-1559 is implemented? Are there non-DeFi tokens you guys are considering adding?

The price to sales ratio (P/S) is a top-level metric and is not affected by how the protocol decides to use its revenues. In general, early-stage protocols tend to reinvest the majority of their revenues into growth, often by subsidizing the use of the protocol’s service. The current focus on Ethereum-based DeFi assets is motivated by the index’s aim to find fairly valued and widely used protocols.

Although, we don’t see a reason not to include non-DeFi assets later on, should they satisfy the rest of the criteria in the proposed index.

5. Why use fully diluted market cap instead of 1 year forward market cap?

In our view, a fully-diluted market cap serves as a more accurate comparison point between different protocols. The fully-diluted market cap determines the price that an investor ultimately pays for a given asset.

Hopefully these answered your questions.

6 Likes

So you are using an estimate of 1 year forward revenues across protocols rather than attempting to do a full discount of all the future value flows?

It seems to me using fully diluted market cap makes this inconsistent because different protocols have different dilution schedules.

An example of would be two protocols that have the same fully diluted market cap but one of them comes to full dilution next year while the other comes to full dilution over 10 years. In the case of the latter the revenue per token would be greater one year out assuming the same value flows. Even if they had the same revenues over a 10 year period the latter would still be the more valuable protocol because of having more revenues per token in the earlier years which compounds into a larger amount.

Hi All,

Please find attached an updated dashboard of the TTI. The dashboard includes data on the historical portfolio weightings and the underlying components (market cap, revenue, and the price to sales ratio) that are used to calculate the weightings and index value over time.

During the past few weeks we’ve moved our product to a more scalable infrastructure, which allows us to add new assets to Token Terminal more efficiently.

Looking at the backdated asset weightings in the TTI, we can already see how a fundamentals-based index gradually starts to diverge from a market cap-based index over time. As the DeFi space grows and new assets are listed to Token Terminal, this differentiation should become more clear over time. The appendix section includes a list of assets (in no specific order) that we’re looking to add to Token Terminal during Q1 2021.

Let us know if you have any questions.

https://terminal.tokenterminal.com/dashboard/TTI

8 Likes

I’m a little confused by this point - Compound and Sushi have liquidity mining (minting), CRV has locking, etc.; really these seem like tokenomic primitives, how do you decide if such a pattern is disadvantageous to passive holders?

1 Like

thanks @TokenTerminal

I guess overlap with DPI is something we should think about as a community. While I think that the overlap, over-time, will be significant, I’m less worried about this now compared to when this index was first proposed. The main reason being the following:

  • TTI index looks at P/S. The sales component of that equation is unlikely to change significantly month to month. However, the price might. As such, at the time of the rebalance the index effectively sells tokens that run far away from fundamentals, while adding exposure to token that underperform the fundamentals. This behaviour is very different from the way DPI rebalances, where as tokens run, their market cap gets bigger and so does their weight in DPI.

Essentially, having DPI and TTI would give investors two different way to play the space.

I do have two questions though:

  1. Can you please provide additional clarity on how the weights are determined. I’m particularly interested in the part in bold “Subsequently, the assets are ordered (lowest to highest) based on their P/S ratios, and given their relative weightings in the index.

  2. TTI will have some major rebalances as per below. All of these are over 10% weight changes, some are 20%. Let’s assume TTI reaches $50 million AUM, a 20% change in weights is $10 million of either buying or selling pressure on the token. Have you done the analysis on the price impact of these rebalances and if the liquidity is sufficient to move this much money, without significant slippage and price impact? If you have, can we please see it. Would also love to see this done at different levels of AUM, from say $20 million to $200 million.

Thanks!

1 Like

Thank you for your comments @akirillo and @verto0912. We compiled your questions and our answers into one post:

Locking, minting, burning
This requirement serves the purpose of ruling out token models, where passive tokenholders are diluted without a weighty financial reason. Liquidity mining is used for boosting the growth of the protocol, and could therefore be considered a weighty financial reason as it aims to increase both the usage and utility of the protocol.

Index dynamics (DPI -TTI)
Over time, the DPI and TTI will most likely gradually diverge, as the TTI aims to front-run price action, whereas the DPI is buying the price-wise top performing assets. This can already be seen when looking at the backtested composition from the past few months. As the DeFi space grows and new assets are listed to Token Terminal, this differentiation should become more clear.

Relative weightings
With “relative weightings” we refer to the same calculation that is provided in the proposal. This is how the weighting for each asset is calculated:

  • Calculate the P/S ratio for each asset
  • Calculate the inverse P/S ratio for each asset (1 divided by asset’s P/S ratio)
  • Calculate the uncapped weighting for each asset by dividing its inverse P/S ratio with the sum of all inverse P/S ratios
  • Sort the assets from the highest uncapped weighting to the lowest
  • Apply a 20% hard-cap (to assets whose weighting is over 20% of the index) by redistributing the extra percentages to the remaining projects pro-rata

Price impact of rebalances
In our view, the price impact of rebalances, and how to minimize it, should be discussed together with the Index Coop and Set Protocol team, who are responsible for the administration of the product. In the current TTI data, larger rebalances result either from (i) new assets being included to the index or (ii) larger changes occurring in a project’s P/S ratio. In the former case, the price impact could be reduced by making the purchases in several tranches over a relatively longer time period. In the latter case, we’re less likely to see as large month-over-month changes as the projects mature and more assets are added to the index. The latest daily trading volumes for the tokens with the larger rebalances give some indication of their current market depth: BNT ($47M), BAL ($90M), COMP ($97M), SUSHI ($665M), YFI ($1.1B), UNI ($750M).

Hopefully these answered your questions.

5 Likes

@TokenTerminal, thanks!

More on liquidity and market depth, though I appreciate that we’ll have to figure this out together.

Where are these numbers from?

My understanding is that currently, all new minting / redemptions / rebalances are done by routing trades via Uniswap. BNT, as an example, currently has $300k of liquidity on Uniswap. If we go from 20% allocation to 3% allocation, no matter over what time period, it will tank the price. We can go token by token. Most Balancer liquidity sits on Balancer, not Uniswap. Even then, +/- 2% depth is about $1.1 million. Even for SUSHI, Uniswap liquidity is limited.

One solution is to rout trades through an aggregator. I’m not a technical person, so I’m not sure how easy this is to do.

Another solution is to rebalance more frequently, perhaps every 2 weeks, so that the weight changes are less significant.

We could also consider capping the maximum month over month weight change per token.

And still another option would be to rebalance every two weeks AND cap weight changes at say +/- 5%.

1 Like

It is possible to export the daily returns of the TTI? I would like to generate some additional summary statistics on TTI portfolio as the return chart alone is an incomplete picture.

1 Like

Thanks @verto0912, these are all great points / suggestions. We’re confident that the rebalancing and liquidity hurdles can be overcome should there be enough interest for the proposed methodology itself.

@Dig, here’s a link to the excel with the daily index prices, hopefully this is what you were looking for: TTI_performance_2021_01_17.xlsx - Google Drive

2 Likes

@TokenTerminal Thank you for providing the updated dashboard showing the weights of assets in the TTI over time!

We can tinker with the details of this proposal, but overall, I am a supporter of the creation of a new index based on P/S ratio. I believe TTI offers meaningful methodological differentiation from DPI which should be attractive to investors and grow the overall market for decentralized indices. I also believe it would be great to have the Token Terminal team aligned with the success of the Coop.

I readily acknowledge the DPI cannibalization risk that @DarkForestCapital, @verto0912, and others have brought forth as a significant risk. However, I do not believe there is a way to easily predict the extent to which TTI will (1) grow the overall market for decentralized index funds offered by the Coop, and (2) cannibalize DPI.

I propose that the Coop do the following:

1. Include TTI as one of the Index Coops fund offerings

2. Extend the retention analysis spearheaded by @jdcook to evaluate market growth and cannibalization concerns. We can address questions like:

  • Is TTI growing the market for decentralized index funds? Is it helping the Coop acquire new holders that we would otherwise not have attracted? What is the overlap of TTI and DPI addresses? What is the overall growth in AUM driven by new unit sales of the TTI and DPI funds?

  • What fraction of existing addresses with DPI also invest in TTI? How does DPI unit retention change for these existing addresses? How does overall unit retention change? This will help the Coop address cannibalization concerns between the two products. (h/t @jdcook)

3 Likes

@TokenTerminal

I have a few minor detail questions that I wanted to separate from my forum post above.

  1. Why is there a cap of 20% for any given asset? I noticed that both Sushiswap and Uniswap are up against this cap.

  2. Is Token Terminal calculating revenues with onchain data or is it doing some manual calculations?

  3. Why did Uniswap go from a 0% to 20% weighting on December 1st?

Thanks!

Thanks for the comments @Thomas_Hepner. Answers below:

Why is there a cap of 20% for any given asset? I noticed that both SushiSwap and Uniswap are up against this cap.
The cap is meant to mitigate situations where the TTI would have too much exposure in only a few assets. The cap is likely to become less meaningful over time as the industry and the index grows.

Is Token Terminal calculating revenues with on-chain data or is it doing some manual calculations?
Yes, our metrics are based on on-chain data. Our metrics - Token Terminal Docs

Why did Uniswap go from a 0% to 20% weighting on December 1st?
Uniswap launched its token in September, so December was the first month it could be included in the index (exclusion applies to launch month and the two subsequent months).

Hopefully these answered your questions.

6 Likes

Hi all,

Due to the concerns around slippage in the original TTI proposal, we’ve made a few amendments that we think enable sound rebalances without affecting the nature and performance of the TTI.

The amendments and the underlying dataset for the index can be found under the following link: Updated index proposal - Token Terminal Smart Beta Index (TTI)

We’ve incorporated these changes into the original proposal and are ready to move to DG1 with the updated product.


Issue

The original index proposal included relatively large monthly rebalances which, in practice, would result in undesirably high slippage. Below, we’ve identified the core drivers behind the large rebalances in the initial index and also proposed a way to fix the issue.

Solution

We identified that the large rebalances were primarily driven by two factors: the original index included relatively few assets and, in some instances, allowed for the addition of new assets with a large (20%) weighting directly from the beginning.

To eliminate such large rebalances, we propose a rule where the portfolio weighting of an individual asset can only change a maximum of 5% during a rebalance or inclusion – with the excess weighting being distributed to other assets pro-rata. We also propose to eliminate the 20% max. weighting per asset to reduce the number of constraints that affect rebalances.

‌While restricting, these proposed changes do not fundamentally change the nature or performance of the TTI, i.e. it still favors the most widely used and fairly valued assets.

Next steps

Should this amended proposal of the TTI pass the DG1, our team would be:

  • Open-sourcing the index. We would build an open API with an accompanying dashboard through which the members of the Coop would be able to inspect the index and all its component parts in real-time. We encourage Coop’s members to contribute by opening issues and merging pull requests.

  • Adding new assets to Token Terminal. By adding new assets to Token Terminal, we continue to expand the eligible asset universe of the index and thereby also decreasing its potential overlap with the Coop’s other products, such as the DPI. Examples of assets that will become eligible for the TTI in the near future include 1inch, KeeperDAO, and Perpetual Protocol among others.

  • Preparing the product together with the Coop. This includes making the necessary adjustments that enable the on-chain implementation of the TTI.

8 Likes

A 5% cap on changes to weights each month makes sense. Good work TT

2 Likes

Great to hear that this is ready for the next phase.

Regarding the next steps you outlined, could you add some detail around the API dashboard and how we can support. Are you looking for feedback on the methodology? Thanks!

1 Like

Thanks for your comments @Lavi.

The purpose of the dashboard is to provide transparency to the methodology and calculations of the TTI. In practice, it would include historical views and tools for simulations. Token Terminal has sufficient developer resources to build the dashboard, but we would be more than happy to invite members of the Coop to participate in and share feedback during the process.

The amended proposal is a result of feedback and comments from the community and our users. We welcome all feedback, but consider this as the final proposal prior to DG1.

5 Likes

I may have missed this, but can you explain the decision (iiuc) to count all money paid to LPs and the protocol as sales/revenue (more like GMV imo) vs just the protocol’s cut of the fees. Would this materially impact the weightings?

Thank you for your question, @LemonadeAlpha.

The primary motivation to use total fees as revenue has to do with the fact that most of these protocols are still at an early-stage in their development, which means that not all of them have protocol revenue, yet.

Our intention was to create a product that can capture a broad range of highly used and fairly valued protocols. We believe that using total fees as revenue works well for this purpose, as it’s ultimately the figure from which (eventual) protocol revenue will also originate from.

5 Likes

3 months ago i requested a csv with historical index weights to analyze the index.

i have not see this data, was it provided on another channel?

without this informationnit is very hard to assess how badly actual performance will disappoint vs zero slippage assumptions in published chart.

Thanks for your comment @scott_lew_is.

Our prior proposal (which is now outdated) came with a dashboard that included data on all the historical weightings. Since it included relatively high slippage, we decided to make amendments to the methodology to reduce slippage.

The amended proposal has the complete underlying dataset attached to it as well. The “Summary” sheet includes the historical weightings:

Amended TTI - December to February 18th - Google Sheets

4 Likes