Introducing CWI: the Category Winners Index

Title: Category Winners Index (CWI)

Author: @MrMadila, @Lanks, @cacho

Created: 4 May 2021


We would like to propose that the Index Coop manage a product based upon an index of crypto’s category winners.

The Categories Winners Index (CWI) is an index that seeks long-term capital growth. It seeks to achieve this objective by identifying the key themes within the crypto arena and incorporating the dominant constituent of each category into the index.

Category identification is critical to ensure sufficient exposure to all aspects of crypto given the expanse and variety of applications and uses. CWI gives investors the ability to gain exposure to a wide spectrum of crypto, but with a focus on a subset of each of the category winners. A ‘category’ is identified by using Coingecko’s categorization & selecting the dominant token within it based on Coingecko’s market capitalization.

Using CWI reduces transaction costs for investors and gives diversified but targeted exposure via a single index aiming for long-term capital growth.


The Category Winners Index is designed to track protocols that have proven execution and have become dominant or disruptive in new emerging spaces. Crypto has progressed significantly from the early days of BTC & ETH and has evolved into an arena that now impacts a huge variety of applications. Crypto is now of significant size for us to be able to distinguish clear sub-categories. CWI aims to identify the market leader in each sub-category & provide holders explicit exposure to each category’s dominant token.

A scenario likely familiar to all IndexCoop community members: Friends/family/colleagues ask what to invest in within the crypto space, and how to transact. By the time you have explained that there is a wide variety of use cases within crypto & each token has different purposes/uses it’s highly likely the individual who was initially interested has decided they don’t have enough time to investigate further. CWI provides an easy answer to that common question by providing easy exposure to the top 10 crypto categories & the top asset within each category.

The target market for CWI however covers a broad investor spectrum, CWI can be used by beginners/intermediate investors who would like convenience & exposure to crypto with low fees & automated rebalancing. CWI can also be used by advanced and institutional investors such as UHNW/GFO who have limited time to track & rebalance their portfolios. There is also strong potential for DAO treasury diversification - DAOs who are looking to divest their treasury are looking for broad but targeted divestiture to spread risk across all crypto categories.

Currently, there is no product on the market that combines the convenience of a single index with a horizontal outlook across the crypto economy. This product enables a very straightforward passive hold strategy for ongoing crypto-asset exposure.


Before we explore the CWI for crypto, let’s walk through how this could work in the traditional investing world using the example of equity. Many say crypto is the next Internet; that it will be as disruptive and as value generative as the Internet has been over the last couple of decades. So what would a Category Winners Index look like for the Internet today? Here is our view:

  1. Mobile - Apple ($2.2T)
  2. Enterprise - Microsoft ($1.9T)
  3. eCommerce - Amazon ($1.7T)
  4. Search - Alphabet/Google ($1.6T)
  5. Social Media - Facebook ($900B)
  6. Gaming - Tencent ($800B)
  7. Payments - Paypal ($300B)
  8. Streaming - Netflix ($200B)
  9. Travel - Airbnb ($100B)
  10. Sharing - Uber ($100B)

Several points to note on this list. First, the names should sound familiar, and you’ve seen the various acronyms like FAANG and MAGA , etc. CWI is somewhat broader and meant to capture a wider spectrum of current activity on the Internet. Second, the list today would very much be different from what it was in 2010 or 2000, categories evolve over time, not constantly, but enough to make it matter.

For example, Sharing, Streaming, and Social are relatively new. In 2000 you might have had Portal (Yahoo), Access (AOL), and Auctions (eBay) instead. Similarly, the winners in each category can change over time, Google was not the first/largest search engine (remember Inktomi, Lycos, Altavista?), and neither was Facebook the first social media site (MySpace, Friendster, etc). And while several competitors exist in each of the categories, our premise is that the winner takes most of the value (network effects!), so we always want to have the current winner represented in the index.

Finally, here’s how we determine the winners. We list all companies worldwide by market cap. We then assign a category for each of the predominantly Internet-focused companies. The top market cap company for each category is then selected as the winner. This is kept up-to-date every month with the current list of Internet category winners. So imagine having this index in the early days of the Internet. All you had to do back then was to decide that the Internet was likely to turn into something big and then just buy and hodl the Internet CWI index for 20 years :-). Likewise, if you think Crypto is that big thing now, the Crypto CWI index is exactly for you!

Proposed composition
(Data as of April 30th, 2021)

Ticker Name Category Market Cap. Weight
1 wBTC Bitcoin Cryptocurrency/Store of Value $1T 10%
2 ETH Ethereum Smart Contract Platform $314B 10%
3 sBNB* Binance Centralized exchange $86B 10%
4 UNI Uniswap Decentralized exchange $22B 10%
5 LINK Chainlink Oracle $15B 10%
6 wFIL* Filecoin Storage $10B 10%
7 AAVE Aave Lending/Borrowing $6B 10%
8 CHZ Chiliz Sports $2.8B 10%
9 SNX Synthetix Derivatives $2.6B 10%
10 ENJ Enjin Coin Gaming $2.5B 10%

*Pending sufficient liquidity, can reassess closer to launch and if still insufficient we will choose the next 2 categories/coins.



As above CWI seeks to identify the top 10 categories within crypto, and then select the clear market leaders within them. Here are the steps to this process:

  1. We start with the top 100 coins by market cap (per Coingecko).
  2. We then list Coingecko’s category definition for each coin. Some minor cleaning of the data is required to take into account multiple categories (we choose the most specific and dominant one), no category (we add one), inappropriate categories (eg stable coin), and conflicting categories. We also cross-reference with other sites like CoinMarketCap and Messari to ensure consistency and integrity of categorization.
  3. The next step is to indicate which of these coins meet our overall requirements. The major considerations here are that they exist on the Ethereum blockchain, have sufficient liquidity, and otherwise meet legal and regulatory requirements.
  4. Finally, the top market cap coin for each category (and which meets our requirements) is selected and added to the index. We continue until we get to 10 unique categories. For clarity, once a coin from a category is included, no other coins from that category will be selected. Also if a category is in the top 10 but it’s winner doesn’t meet requirements (for example not on Ethereum), we then don’t include that category.


CWI uses equal weight for each category so each component is fixed at 10%. The general goal of an equal-weighted index is to represent the performance of its constituents in equal proportion to one another. This way you get broad coverage and diversification across the spectrum of crypto. You also get a good balance between higher returns from relatively riskier/newer categories as well as some protection from relatively stabler/older categories.


Rebalancing is done on a monthly basis in three steps and over two phases. The three steps are as follows:

  1. Swapping winners. For a given category, if a new token becomes consistently larger than an incumbent, then the incumbent would be replaced by the new challenger. The average market cap of the challenger for the last 30 days, would have to be greater than 10% for this change to take effect. This is implemented to ensure that a new winner can definitely be crowned and to prevent excessive back and forth trading.
  2. Swapping categories. This would only happen if a new category became bigger and more important than the existing ones (per criteria above). In this case, the new category (and its winner) would replace the least important one. Similar to above, the new category winner must have a greater than 10% average market cap over the last 30 days, to replace the incumbent.
  3. Balancing to 10%. This would happen every month and is to keep the equal-weighted index intact and fixed to 10%.

Similar to other Indexes at Index Coop, we propose to implement a two-step process for rebalancing. The first phase is the determination phase and happens in the 3rd week of the month. This is when the changes for the next reconstitution are determined and communicated. The second phase is the reconstitution phase and happens in the first week of the month. This is when the changes are actually implemented, including the rebalancing to 10% and any category or winner swaps.

Core value propositions


  • Long term exposure to the best-in-class investments
  • Set and forget, 1 click exposure to market leaders
  • Low cost to hold with targeted diversification

Index Coop

  • First and only index with a horizontal cross-thematic approach
  • Broad investor spectrum; appeals to beginners, advanced and institutional investors + strong potential for DAOtreasury diversification
  • Very Large Total Addressable Market (TAM)

Size of opportunity

The current global crypto market consists of ~7,000 listed crypto assets with a total market cap of $~2.3T.
~60% of this total value is captured across just 10 of the assets under consideration for CWI. This percentage becomes higher still once stablecoins and “wrapped” coins are deducted from the initial total. With this skew of capital accruing to just these assets, a proxy such as CWI could become a huge magnet for users looking for convenient exposure.

The real value add from CWI however is in the construction. It is optimized for upside reward by singling out only the most successful market themes and subsequent dominant investments. Filters out the laggards, all whilst maintaining balanced diversification.

In short, CWI is diversified yet targeted. This approach is beneficial to a wide range of investors as mentioned above.

One comparison we can draw from traditional finance would be the FAANG+ (Facebook, Apple, Amazon, Netflix & Google) ETF’s. Just these 5 stocks worth ~$6T makeup around 15-20% of the entire SP500 index and have significantly outperformed it over the last 5 years. Over certain time periods, they have been responsible for almost 40% of the index’s gains.

According to Investopedia’s “Best FANG ETFs for Q2 2021,” just the top 3 FAANG+ focused ETF’s have combined AUM’s of $290b. Not all of this AUM is attributed just to these 5, however, they are heavily concentrated in them. $290b equals 4.8% of the $6T combined market caps of the FAANG stocks under the management of just these 3 ETFs.

CWI capturing just 0.01% of the value of the constituents would equal an AUM of $~130-150M

Or $1.3-1.5B at 0.1%

A streaming fee of 95bps would net $1.2-1.4M or $12.4-14.3M pa respectively.


CWI is different from any other product within the crypto ecosystem. CWI is both broad i.e. covers everything from blue chips (BTC, ETH) to DEFI (UNI, AAVE), and NFT platforms (CHZ, ENJ) to infrastructure (LINK, FIL), whilst also being targeted to the dominant tokens. CWI is not simply a ‘top 10’ of crypto assets index, it covers all major market segments. It is different from existing IndexCoop products, and IndexCoop competitor products such as CC10, BCP, TCAP, etc. It is also different from any of ETF contructions in the traditional finance space too.

Next, we consider the overlap between existing Index Coop products. In perspective, we believe they are not significant.

DPI constituents: 14 + CWI constituents: 10 = Total 24. Overlap: 3 UNI, SNX + AAVE = 13%
MVI constituents: 15 + CWI constituents: 10 = Total 25. Overlap: 1 ENJ = 4%

BED constituents: 16 + CWI constituents: 10 = Total 26. Overlap: 5 BTC, ETH, UNI, SNX & AAVE = 19%

FLI constituents: 1 + CWI constituents: 10 = Total 11. Overlap: 1 ETH = 9%

** does not factor in weightings

On-chain liquidity analysis

CWI by definition contains the category winners i.e. large cap assets, all of which have deep liquidity on major DEX’s. The components are highly likely to be in the top 100, if not the top 50 list of largest coins by market cap. However, we may face liquidity issues with wrapped or synthetic versions of tokens so will have to consider that carefully. (see notes under proposed composition above)


We propose a 0.95% Streaming fee with an 80/20 split to Index Coop (0.76bps to Index Coop, 0.19bps to meths).

Intrinsic Productivity

There are currently no predetermined plans for intrinsic productivity. We are aware of the potential for instance to LP the WBTC: ETH part of the portfolio and/or stake ETH, AAVE, etc in order to generate returns and re-invest them back across the portfolio.

We would like to invite the community to provide feedback and expert guidance around this topic.

Author Background and commitment

Index Coop contributors:

@MrMadila Is a “Gold Owl”. He is an active member within IndexCoop leading the Org working call, is involved with the work of the TWG & contributing across the community more generally with Twitter accounts such as Documenting DPI, FLI, MVI, etc. He has 10 years of previous experience in traditional asset management operations.

@Lanks has been involved in crypto for 2 years & has 15 years of experience in traditional finance covering investment banking, wealth management, and more recently is leading new business activities at a fintech startup. He has recently been contributing to Index Coop’s BD team with input on custodial solutions and institutional sales.

@cacho has been a full-time investor since 2012 mostly in crypto, with 20 years past experience in tech and startups both in development and product management roles.

Lastly our thanks to @DarkForestCapital, @Matthew_Graham and @Lavi for the help and input at short notice!


Awesome work! I really like the concept of picking a theme (Crypto) and then tracking the blue chips by subsector (Category Winners). I think there’s a place for this in our suite of products. Two technical aspects that I have questions about or I believe require more thinking:

  • This methodology seems overly reliant on Coingecko’s category definitions. It’s a worthy starting point, but I’m curious if you guys have thought through what happens if Coingecko stops categorizing altogether, switches categories too quickly, or does a major overhaul in how they categorize overnight.

  • My impression of equal-weight indices is that it’s a nice-sounding concept that produces weird results in practice and increases rebalancing overhead. For example, Uber had a fairly rough March 2020 to October 2020 while Amazon was roaring. If your CWI - Internet Index was real in 2020 there would have been a lot of reallocation to Uber and away from Amazon/Netflix in that period. In some ways, precisely when you’d want exposure to Streaming and eCommerce. Now maybe you’re ok with that because “Sharing” bounced back and you wanted flat “Internet” exposure. Maybe my example is off, but I’d love to hear a little more about why you think equal-weight is the right approach. And maybe some backtesting will help to clarify.

Awesome to see the work being done to create products for the average investor. I see a massive upside of these type of investors when we start to get our tokens listed on centralized exchanges (coinbase, kraken, etc…) and mainstream investing apps (paypal, cashapp, etc…).

I do like this concept but I wanted to ask why you decided on the fixed allocation instead of a mkt cap approach.

I’ve been working with some fellow community member on what we call the Cryptocurrency Boomer Index that has some similarities with this proposal. This is still a very early draft, I’m working on a more complete proposal.

EDIT: Added a little bit more context.

Thanks a lot for the feedback and the questions, much appreciated!

On methodology, we’re looking at Coingecko, CoinMarketCap, and Messari as sources. We use Coingecko as the primary. To our surprise all 3 are fairly consistent, but there are a few nuances which require some massaging of data. So if one goes away, we have others to fall back on, and if one switches too much/overhauls, we might just be more conservative and wait for industry acceptance. The biggest issue so far has been slowness to recognize new categories so personally that’s what i’m most worried about. We have all this categorization on a working spreadsheet, so if anyone is interested happy to share.

On equal weight indices, we did do a backtest for crypto CWI (not Internet to be fair) using the past 90 days and trying out 4 different methodologies:

  • market cap
  • sqrt of market cap
  • start at equal weight 10%, let winners ride
  • start at equal weight 10%, and rebalance

Quick summary. The first two were just so lopsided with BTC and ETH overwhelming the index with hardly any exposure to the smaller coins (think smallest had like 1%). Also in a way b/c CWI chooses both categories and winners based entirely on market cap, feels like we’re already heavily relying on market cap as it is. The third one (let winners ride) was actually our preferred one going in, but it led to all kind of anomalies. For example Filecoin went from 10% to 30% in one month, and after 90 days both BTC and ETH were at 2-3%. This was obviously during a bull market, and would expect exactly the opposite to happen during a bear market, so again not ideal. The fixed at 10% actually best fit the mold, good balance, broad exposure and continue to get some risk as well as some safety. We want this to be a long term holding and relevant throughout the bull/bear market cycle. There is a spreadsheet here as well, if anyone is interested. I should also say we are more than open to other weighting suggestions if they do a better job!

And finally a proper backtest for longer periods of time for both CWI and Internet has come up a few times, can certainly do that later, we were just keen to at least get this initial proposal out for comments.

1 Like

Thanks for the thorough response! All of that detail helps me truly appreciate the depth of thought behind those choices and bolsters my support.

Product folks would know better but I think there’s certainly an opportunity for further backtesting between DG1 and DG2, but the detail on the 90 day backtesting goes a long way in understanding the proposal.

If you’re ok sharing then I wouldn’t mind looking at that categorization spreadsheet. I’m fallow8 on Discord but @MrMadila and @Lanks know where to find me as well.

Thanks Nate, have sent you one of the spreadsheets. We are just in the process of finalising another with some back testing data.

If anyone else is also interested in this additional data please DM me

Curious how you deal with the scenario where AUM grows quite large, and one of the small market cap holdings or categories needs to be ‘swapped’ to another winner. Would a sudden move like that, especially in a small caps coin, artificially drive down price / value of the holding we are exiting simultaneous driving up the price of the one we are entering?

Maybe an alternative would be: equal category weight, with market cap weightings for top 3 tokens in that category? Doesn’t completely solve the problem but would mean only the 3rd smallest market cap in the category would drop off at a time.

I like this idea as it coincides with a thought I had the other day regarding the trend following strategy often used in TradFi. Commodity trading advisors use varying trend following strategies to put on and take off market exposure (or go short). Giving the weightings some additional thought, it may make some sense to incorporate something like an Average True Range (ATR). Average true range is more of a volatility measurement where:

True Range (TR) = Max(High - Low), abs(High - Close), abs(Low - Close)
ATR = (1/n) * Sum(TR)

A lot of traders use it to gauge if their stock/index/etc is behaving “normally” over the last set number of days/weeks, or if the price movement that it’s experiencing that day/week is 2 or 3x normal ATR signifying something different.

I will DM you on discord @MrMadila as I am interested in taking a look at the data.


I agree with this concern, and would add that not only driving down / up the value of the tokens, if the methodology is clear and the timing of the re-balance trade is transparent, it would be easy to front-run the reconstitution.

However, I think what I’m reading is they’ll take a page out of DPI playbook to manage that risk? This is from the Tokensets page on DPI reconstitution: “As assets tracked by the index grow, the reconstitution window will expand to more than one day to lower the reconstitution’s market impact.”

Either way maybe someone can comment further on this risk. If it’s a billion+ dollar product, 10% on a smaller cap category is going to have an impact that isn’t an issue in the FAANG+ context of megacaps, and I think even more pronounced than in DPI because it’s all or none for the category winner.


Thanks @prairiefi @frnkmxwll & @liquiditat for the feedback and questions

@frnkmxwll Agreed, a sudden constituent swap could cause unnecessary price pressure. The reconstitution window for CWI would be expanded to more than one day to lower the reconstitution’s market impact - (similar to DPI as mentioned by @prairiefi). Pls also see related points below re front running.

Hi @liquiditat - It’s not clear to me if ATR could be used for rebalancing of the index, or for ongoing weighting calculation of the index. My understanding of ATR is it’s generally used with short term trading strategies. CWI proposes the average market cap of the challenger for the last 30 days, would have to be greater by 10% than the incumbent for a constituent change to take effect. This is implemented to ensure that a new winner can definitely be crowned and to prevent excessive back and forth trading.

Hi @prairiefi agree with this concern - front running or ‘index gaming’ as it is sometimes referred to is likely, and is in-play for any crypto or tradfi index where the index methodology is transparent. We did some research on this & looked at a traditional finance example: MSCI who (according to their website) have 14.5tn under management linked to MSCI indexes. The TLDR is an index cannot avoid being front run if it publishes its methodology/criteria publicly & transparently. But it can mitigate the impact of being front run by:

  • Spreading re balances/constituent changes over multiple days/weeks.
  • Sensibly reflecting changes in the asset universe - i.e. ICO/IDO may not be immediately included - there could be a minimum time period requirement before a new token could be included into the index.
  • Establishing a period of time between a change announcement and the implementation date (i.e. to give the market time to digest & price activity to settle/normalize).

The above 3 points are used in the MSCI example to mitigate the impact of front running & we should look at incorporating it into CWI too.

Hi All,

It’s an interesting idea and offers a diverse ownership of liquid tokens.

I think the biggest challenge I see is how big is the market? What problem is it solving for the customer?

To me, BED is a great product for people looking for no brain crypto exposure. DPI is great as it gives broad DeFI exposure to people who I would assume already have some BTC and ETH in their portfolio. I struggle to see how this product sits as a logical step on a customers journey between BED and owning something wider ranged (i.e. multiple tokens).

How do we market this as being something better than BED? How do we convince customers to swap BED for this (rather than adding some Binance, LINK, and FILE to their BED tokens)?

While the simple structure makes the Dev work straight forward, launching new products still takes coop resources and I’m not yet convinced of the upside vs downside balance for this suggestion.