Coin Metrics' State of the Network: Issue 25

Tuesday, November 12th, 2019

The Evolution of Ethereum Tokens

In 2015, Ethereum ushered in a new era for blockchains: the age of the token.

Broadly defined, a “token” represents a utility or asset and is typically issued on an existing blockchain. In contrast, a “coin” is a crypto asset that is native to its own blockchain and is primarily used as a currency (“coin” and “token” are sometimes used interchangeably, but we will keep the distinction throughout this piece). For example, BTC and ETH are coins, while MKR and BAT are tokens.

The concept of crypto tokens has existed in various forms since well before Ethereum. For example, Bitcoin “colored coins” can be used to tokenize items using BTC, without the need to issue a new asset. This can be done by “coloring,” i.e. marking, specific coins using OP_RETURN. Alternatively, this can also be done by agreeing that specific Satoshis, which is the smallest unit of BTC, represent a real-world asset.

But Ethereum introduced a new, more user-friendly way to create tokens. Using a simple smart contract, Ethereum made it easy for anyone to launch their own token for just about anything.

Before long, thousands of tokens were launched on Ethereum. The explosion of tokens made standardization increasingly important, to ensure that the tokens could be exchanged for each other. ERC-20 implemented a standard interface that made it trivial to exchange any ERC-20 token for another, and to integrate ERC-20s within crypto wallets and decentralized applications (dapps). In early 2018, ERC-721, which are tokens that are specifically used to represent unique, non-fungible digitally scarce tokens (for example, a CryptoKitty or a one-of-a-kind piece of crypto art) was also adopted as an official standard.

Since then, the pace of Ethereum token change has increasingly accelerated. In this piece we explore the evolution of Ethereum tokens and look at where they might be headed.

Network Value to Token Value

One way to evaluate tokens is to look at the market cap of the smart contract platform’s native coin (i.e. ETH) compared to the aggregate market caps of tokens launched on that platform. We refer to this ratio as the “Network Value to Token Value” (NVTV) ratio, as proposed by Chris Burniske

In this case, we calculated the ratio by dividing ETH’s market cap by the aggregate market cap of a selection of the biggest ERC-20s. Although there are thousands of other tokens that have been launched on Ethereum, the selected tokens represent a large majority of the total ERC-20 token market cap. A full list of the tokens we used can be found as a footnote under the below chart. 

Ethereum’s NVTV ratio has been steadily declining. On April 1st, 2019, Ethereum’s NVTV ratio hit an all-time low of 1.57. As of November 10th, the ratio is 1.90.

ERC-20 Assets: ant, bat, bnb, cennz, ctxc, cvc, dai, fun, gnt, gusd, ht, icn, knc, leo_eth, link, loom, gno, lrc, mana, mkr, omg, pax, pay, poly, powr, ppt, qash,rep, salt, srn, tusd, usdc, usdt_eth, wtc, zrx

Realized cap tells a similar story. Realized capitalization is a metric created by Coin Metrics that is calculated by valuing each unit of supply at the price it last moved. This is in contrast to traditional market cap which values each coin uniformly at the current market price. Realized cap can be thought of as a measure of the average cost basis (cost basis is basically the total amount originally invested).

The realized cap version of Ethereum NVTV has also been steadily decreasing and is currently at an all-time low of 2.57. The decreasing NVTV ratios signify that tokens have steadily been gaining ground on ETH in terms of valuation. 

Most of this growth since mid-2018 has been coming from a specific subset of ERC-20 tokens: stablecoins. 

The below chart shows the share of market cap for utility tokens, exchange tokens, and stablecoins. We used a simple, high-level taxonomy for categorizing tokens; however, other groupings or taxonomies are possible. 

A “utility token” is a subset of tokens that are “used to finance the network by providing its buyers with a guarantee of being able to consume some of the network’s products” (definition via BitcoinWiki). Utility tokens were typically issued during the ICO boom to serve as a way to raise money as well as a way to make payments within a project’s ecosystem, access a particular service or feature, or participate in a particular activity such as voting.

“Exchange tokens” are a subset of utility tokens created by cryptocurrency exchanges (e.g. Binance’s BNB token). Exchange tokens are typically used to raise funds for exchanges and offer discounts on things like exchange fees. 

“Stablecoins” are tokens that are designed to fix their value to another asset, often a fiat currency such as the USD. Tether is currently the biggest stablecoin by most measures, but other stablecoins built on top of Ethereum include DAI, USDC, PAX, and TUSD. 

Although exchange tokens were gaining ground in early 2019, BNB switched over from an ERC-20 token to a mainnet version of the BNB token (on their own blockchain) in April, which caused Ethereum exchange coin market cap to plummet. 

On July 1st, 2018, Ethereum utility tokens had an aggregate market cap of $7.52B, compared to $2.98B for exchange tokens, and $109M for stablecoins. As of November 10th, 2019, utility tokens have a market cap of $5.19B, exchange tokens have a cap of $2.55B, and stablecoins have a market cap of $3B, up by over $2.8B from just a year and a half earlier.

The below chart shows the percent share of market cap for each of the three token categories. A complete list of the assets we used for each category can be found in the footnote under the below chart.

Utility tokens: ant, bat, cennz, ctxc, cvc, fun, link, loom, gno, gnt, icn, lrc, mana, mkr, omg, pay, poly, powr, ppt, qash, rep, salt, sr, wtc, zrx

Stablecoins: dai, gusd, tusd, usdc, pax, usdt_eth

Exchange Tokens: bnb, ht, knc, leo_eth

Furthermore, most of the growth has been coming from one specific stablecoin: Tether (USDT). 

As we’ve covered in past issues of State of the Network, Tether exists on multiple different protocols, the two biggest of which are the Omni protocol (which itself is built on top of Bitcoin) and Ethereum. Over the last several months, usage has been shifting from the Omni-based version to the Ethereum-based version of Tether. 

The Transaction Flippening 

Historically, aggregate token transaction count has been lower than Ethereum’s non-token transaction count (i.e. total transaction count minus ERC-20 and ERC-721 transaction count which was largely comprised of simple transfers of ETH). But since May 2019, token transactions have been threatening to pass non-token transactions. As of November 10th, ERC-20’s had about  303,000 daily transactions vs about 290,000 for ETH.

The below chart shows transaction counts for ERC-20s (red line), ERC-721s (green line), and non-token transactions (blue line, ETH transactions minus ERC-20 and ERC-721 transactions), smoothed using a 7-day rolling average.

A lot of ERC-20’s rapid transaction count rise has also been due to USDT. The below chart shows the market share of the ten ERC-20 tokens with the highest daily transaction counts (averaged over the last 30 days) over the course of 2019. USDT started gaining ground in May and now has over 80% of the share of transaction counts of the top ten tokens.

ERC-721s Unchained

While ERC-20s have been the dominant type of token up to this point, we may be on the cusp of the rise of ERC-721s. 

As of late October, ERC-721 transfer count has shot past both ERC-20 and ETH transfer counts. Previously, ERC-721 transfer count peaked during the CryptoKitty craze of late 2017. November ERC-721 transfer counts have already rocketed past peak CryptoKitty transfer counts.

Transfer count paints a slightly fuller picture than transaction count of the real trading activity of individual ERC-721 assets. Since ERC-721’s each represent unique tokens, many tokens are often bundled together and transferred as part of a single transaction. The below chart shows transfer counts for ERC-20s (red line), ERC-721s (green line), and ETH transfers (blue line), smoothed using a 7 day rolling average. 

This large spike in ERC-721 tokens is due almost entirely to an Ethereum-based card game called “Gods Unchained.” 

Gods Unchained is a trading card game that is similar to the popular game Hearthstone. However, unlike Hearthstone, Gods Unchained is built on the Ethereum blockchain, and each one of its cards is represented by an ERC-721 token. This means that users truly own their cards and can trade them freely on the open market, similar to any other cryptocurrency.

Gods Unchained has been in the news recently due to an incident related to the Hong Kong protests. On October 7th, Blizzard, the maker of Hearthstone, announced that they were rescinding the prize money from a champion pro player and suspending him for a year because he had spoken out in support of the Hong Kong protests. 

The next day, in a tweet that has since been retweeted over ten thousand times, Gods Unchained stated that Hearthstone cared “about money more than freedom.” They also offered to pay for all of the banned Hearthstone player’s lost winnings and offered a free entry ticket into a large Gods Unchained tournament. Subsequently, Gods Unchained sold out their Genesis Card Pack for a total of 33,333 ETH, equivalent to about $6.2 million.

Although still early, Gods Unchained could be an example of a real use case for crypto tokens in gaming. Blockchain-based games put gamers in control of their in-game assets, which means they cannot be revoked or censored. Gods Unchained is only one example of many games that are now being developed on blockchains using non-fungible tokens (NFTs). NFTs are also being used in applications like the Ethereum Name Service and in virtual worlds like Decentraland, and will soon likely be used for many other types of applications as well. 

Although there are still only about 4,600 ERC-721 contracts compared to over 184,000 ERC-20 contracts (and over 12 million non-token contracts), ERC-721 contracts have been growing rapidly over the course of 2019. Since January 1st, the number of deployed ERC-721 contracts has grown by almost 350%, compared to about 39% and 36% for ERC-21 contracts and non-token contracts, respectively. 

Furthermore, overall Ethereum smart contract usage is growing. Ethereum contracts calls have been steadily climbing upwards, and recently hit an all-time high thanks in large part to Gods Unchained. As the Ethereum smart contract economy continues to grow and evolve, tokens will likely become an increasingly important part of the ecosystem.

The below chart shows Ethereum contract calls count smoothed using a 7 day rolling average.

Ethereum tokens have already evolved tremendously over their short life span, and will undoubtedly change just as rapidly moving forward. We will continue to monitor Ethereum’s NVTV, the rise of Ethereum based stablecoins, and the potential breakout of ERC-721s.

Network Data Insights

Summary Metrics

After XRP daily average transaction value temporarily surged passed ETH last week, both XRP and BCH adjusted transfer values dropped significantly this week. LTC’s adjusted transfer value, however, shot up over 89% after being down by over 20% the previous week. Despite transfer value being up, LTC’s transfer and transaction count were both down, signifying that a relatively small number of addresses were likely moving around large amounts of crypto.

BTC’s daily fees gained over 10% for the second straight week, after growing by 16% last week. BTC continues to climb ahead of ETH in terms of daily fees; over the past week, BTC averaged $346.9k of daily fees compared to $91.8k for ETH. XRP fees grew by over 100% this past week, but still averaged less than $1k per day. 

Network Highlights

LTC’s hash rate and difficulty have both been in free fall since July. Both are now on the verge of reaching lows not seen since early 2018. 

After BTC’s hash rate dropped last week, as we reported in SOTN Issue 24’s Network Data Summary Metrics section and on Twitter, BTC’s difficulty readjusted downward on November 7th.

Market Data Insights

Bitcoin’s price has remained largely unchanged over the past week at -2% while Ethereum (+4%), Litecoin (+9%), and EOS (+9%) have experienced moderate gains. 

Among large-capitalization assets, Stellar has seen the largest gains at +19% after the Stellar Development Foundation effectively burned 55 billion tokens by sending them to an account that cannot sign transactions. 

Among smaller capitalization assets, Cosmos (+24%) saw a large increase, although there does not appear to be a specific catalyst. Tezos (+41%) has seen the strongest gains among this set of 24 assets largely due to Coinbase’s announcement that it would offer staking rewards on its platform. Maker was up +30%, perhaps in part due to the upcoming launch of multi-collateral Dai scheduled on November 18. 

Revisiting the Bitcoin Safe Haven Thesis 

For the majority of this year, gold and other haven assets have seen large capital inflows due to a confluence of factors: 

  1. An environment of heightened geopolitical instability, particularly with respect to U.S.-China trade tensions but also in other localized areas.

  2. Softness in several key macroeconomic indicators in most developed world economies, particularly in manufacturing, a sector traditionally viewed as a bellwether of the overall economy.

  3. A sharp and unexpected pivot to more monetary policy easing, most notably from the Federal Reserve and from the European Central Bank, and a fear that more extreme monetary policy tools will be necessary.

These factors caused gold to rally above $1,500, peaking in late August. Market commentators also drew comparisons to Bitcoin because of its attractive store-of-value properties. Indeed, short-term measures of correlation between Bitcoin and gold returns earlier this year reached one of the highest levels in history (almost +0.50). 

Recent developments have made it clear that we are now witnessing another shift. Based on (1) the increase in long-term sovereign bond yields across most developed world economies, (2) a shift in forward guidance from the Fed, and (3) a sell-off in gold, market participants now believe we are past the point of peak monetary policy easing. Any further easing appears to be appropriately priced in. Recent firmness in macroeconomic indicators confirm that fears of a global recession are overblown and optimism for a U.S.-China trade deal is rising. 

Bitcoin received intense media attention as the need for safe-haven assets increased but it has largely been ignored as this need has abated. Recent price action and short-term measures of correlation between Bitcoin and gold returns complicate the simple narrative that Bitcoin benefits from safe-haven capital flows. 

Gold recently had one of the largest single-day sell-offs in years, but the 30-day correlation between Bitcoin and gold returns stands at -0.22. Not only does this cast doubt on the narrative established earlier this year, it suggests the reaction function of Bitcoin to macroeconomic and geopolitical developments is complex and inconsistent. 

CM Bletchley Indexes (CMBI) Insights

For the second week running the Bletchley Mid Cap and Small Cap indexes have outperformed the larger cap indices, returning 5% and 2% respectively. As evidenced above in the Market Data Insights, the outperformance of the Mid Cap Index is largely due to the performance of Tezos, which makes up 10% of the index and returned 41% for the month.

Since Bitcoin is a major component of both the Bletchley 10 (69%) and Bletchley Total (64%), their performance relies a lot on the returns of Bitcoin over the period. Bitcoin was one of the weaker performing large-cap assets of this week, and its impact on the Bletchley 10 and Bletchley Total Indexes is highlighted by the difference in returns between the indexes market-cap-weighted (-0.5%) and the even weighted (4%) versions. 

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics is hiring! We recently opened up 6 new roles, including Blockchain Data Engineer and Data Quality and Operations Lead. Please check out our Careers page to view the openings.

As always, if you have any feedback or requests, don’t hesitate to reach out at

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in depth research and analysis.

Coin Metrics' State of the Network: Issue 24

Tuesday, November 5th, 2019

Weekly Feature

Macroeconomic Indicators for the Crypto-Economy 

Just like a nation state’s economy has economic concepts (such as gross domestic product, inflation, and unemployment), a crypto asset can represent a miniature economy that has similar economic concepts. We present our network data using a simple analogy that many people are familiar with: macroeconomic indicators. 

Miniature Crypto-Economies

For certain ERC-20 tokens, viewing crypto-networks as small-scale economies is relatively straight forward. For example, Golem (GNT) represents a miniature economy in which idle computing power is the sole service offered and purchased. Similarly, Basic Attention Token (BAT) represents an economy in which the sole services provided are an individual user’s attention and advertising. 

When considering Bitcoin’s crypto-economy, things are a little more complex. Under a narrow interpretation, Bitcoin itself is the sole good that is manufactured and sold in this virtual economy. An alternative interpretation is viewing the Bitcoin network as a full-fledged economy in which Bitcoin is used as a medium of exchange for a wide range of goods and services. This represents Bitcoin’s use on the dark web, as one example. 

Although analyzing a nation state’s economy is difficult because of its almost infinite complexity, government statistical agencies report on the current state of the economy by creating a conceptual, idealized definition of what they are attempting to measure, and use sophisticated sampling and survey techniques to create an estimate. Despite their best efforts and steady improvements in data collection methods and calculation methodologies, macroeconomic data is reported with lengthy lags and is subject to revisions long after the initial release. Policymakers and researchers are forced to rely on imperfect information to drive their decisions. 

Crypto-economies are less susceptible to the lag and measurement error present in macroeconomic reporting in nation-state economies. The most fundamental action in any economy is a transaction between a willing buyer and a willing seller for a good or service, and a crypto-asset’s shared, immutable, and open ledger reveals each individual transaction (except for layer 2 transactions, and transactions taking place inside custodians), allowing analysis of the economy in real-time. In theory, if an actor knew the identity behind every address and the nature of each transaction, the state of a crypto-economy could be reported with no lag and with close to perfect precision. 

Bitcoin’s Gross Domestic Product (GDP) Growth

Among macroeconomic indicators, a country’s gross domestic product (GDP), which measures the economic value of final goods and services produced by an economy over a given time period, is often viewed as the single most important indicator. 

What is the network metric that most closely matches the conceptual definition of GDP? Coin Metrics’ adjusted transfer value, which measures the USD value of the native units transferred over a given period of time, is a strong candidate. According to the latest data, $2.33 billion of value is transferred in the Bitcoin economy each day. 

Although adjusted transfer value is likely directionally correlated with GDP, it likely overstates the true GDP figure perhaps by several orders of magnitude because of two reasons. One, transactions involving the purchase of goods and services are likely a small fraction of total activity. Two, of the transactions that involve the purchase of goods and services, this metric does not exclude the purchases of intermediate goods (i.e. a product used to create a final product), which are explicitly excluded from the GDP calculation. 

However, Coin Metrics does make several adjustments in order to make this metric more closely represent true economic activity. For UTXO-based chains, each output transaction must be comprised of prior inputs. All inputs are wholly consumed such that it is impossible to definitively know which outputs represent a legitimate, economic transfer of value and which outputs represent change being sent back to the sender (see here for more details on change). Coin Metrics employs several sophisticated heuristics to detect change outputs and other non-economic transfers. Such adjustments are the first step to creating an accurate GDP calculation. 

Although a true measure of Bitcoin’s GDP is currently unknown, it is likely significant. Under the assumption that only 1 percent of adjusted transfer value is for final goods and services, this would result in a nominal Bitcoin GDP of $8.4 billion, approximately equivalent to the world’s 140th biggest economy. 

For policymakers, the level of a country’s GDP is rarely of interest. Instead, they are interested in the growth of GDP and how this growth fluctuates against theoretical potential GDP growth. Below we show Bitcoin’s GDP growth over time, represented as annualized three-month continuous growth. Bitcoin’s “business cycles” can be seen here, sometimes growing rapidly and sometimes contracting rapidly. Current growth is slightly negative at -15% but up sharply from recent lows. 

Extreme growth numbers can be seen in Bitcoin’s history. For example, a continuous growth rate of 600% (last seen during the peak of the recent bubble), sustained over a year, leads to an increase of over 40,000% from its initial value. 

Note on smoothing: All level charts are smoothed with a 28-day centered moving average, followed by a 2-day centered moving average. Recent observations are extended forward using a 7-day non-centered moving average. Continuous growth charts are calculated after these transformations are applied.

Bitcoin’s Population Growth 

Demographics are often linked to macroeconomic reporting because working-age population growth is a primary factor in a country’s long-term potential growth. Assume that the Bitcoin network represents a virtual nation-state. What is the population of this nation and how is it growing? 

One estimate of the total population is the number of addresses with a balance greater than 0.001 Bitcoin, a number that is large enough to exclude dust balances and small enough to represent a meaningful economic amount for most people in the world. Current estimates indicate 14.42 million users of the Bitcoin network, although the true number remains unknown -- there can be a many-to-one or one-to-many mapping between individuals and addresses. Based on publicly released user counts for some exchanges, this figure likely understates the true number. 

User growth remains strong at 19% although down from historical averages. User growth has become negative only once (as measured over this interval) in Bitcoin’s history during early 2018, although this artifact is likely to have been caused by UTXO consolidation as high fees subsided. Growth figures are lower over the past two years, perhaps because of the increased use of exchanges and custodians as well as more efficient transaction batching and wallet software. 

Bitcoin’s Transaction Count Growth 

Although Bitcoin’s transaction count does not map to an existing macroeconomic indicator, it can serve as an alternative measure of growth. Similar to how retail sales, personal income,  durable goods orders, manufacturing and services PMI all serve to measure more narrow components of a country’s GDP, analyzing transaction count can provide a more focused view on one element of economic activity. 

Current transaction count stands at 302,000 transactions per day. Peak transactions have never exceeded 400,000 transactions per day, and as we approach this level, fees begin to rise and economic participants will be incentivized to move their transactions through second layer channels. Counting transactions on the blockchain layer and higher layers (in the future) is one important aggregate measure. An alternative related metric is transfer count which counts the total number of transfers within all transactions. 

Here transaction count growth illustrates a problem of Bitcoin’s deflationary nature for economic activity. Current transaction count growth is -30% and has been negative for several months. What we see here is that as Bitcoin’s price rises (denominated in U.S. dollars), the price of goods and services denominated in Bitcoin decreases. A deflationary economic environment reduces the incentive to spend because holding existing Bitcoin means that an individual can purchase more goods and services at a later time. This is a key reason why almost all the planet’s central banks have price stability as their central mandate and a small but positive inflation target of 2% as one of their policy objectives.

Bitcoin’s Block Size Capacity Utilization 

A nation state’s capacity utilization is also a commonly examined growth indicator. Capacity utilization measures the degree to which a nation’s productive capacity is being utilized. Most of the economic activity in the developed world economies have now shifted to services, and capacity utilization is normally a measure of a nation’s manufacturing industry. Nonetheless, capacity utilization is important because the manufacturing sector is often seen as a bellwether for the broader economy. High rates of capacity utilization also signal wage and price pressures which are important considerations for a central bank’s policy response. 

Here we show Bitcoin’s block size capacity utilization -- the mean size of a Bitcoin block as a percent of maximum theoretical size. Although the analogy is imperfect, analyzing block size as an important indicator of growth draws several similarities. As block size reaches close to its theoretical maximum, it places upward pressure on Bitcoin’s fee market and generally leads to a slow down in growth of other macro indicators. 

Maximum block size has only been reached twice in Bitcoin’s history and capacity utilization currently stands at 87%. 

Bitcoin’s Industrial Production 

Similar to capacity utilization, industrial production is a measure of a nation state’s manufacturing sector -- including mining. Although the analogy is again imperfect, Bitcoin’s hash rate is conceptually similar. Hash rate has increased from less than 5 million hashes per second during its first day of existence to a current hash rate of 90.91 million trillion hashes per second. 

Similar to a nation state’s manufacturing sector, miner behavior is important for the health of a crypto network, not only as a measure of security, but also because miners are one of the only natural sellers in the market. Assuming that Bitcoin miners breakeven in the long-term, miners must sell over $6 billion of Bitcoin over the course of a year based on today’s prices. 

Hash rate growth lags price growth by several months. The current hash rate is in line with recent historical averages but down from recent highs, mirroring price declines. Growth, measured over this interval, has only turned negative three times in Bitcoin’s history. Declines in hash rate are critical events and have strong implications for the amount of miner-led selling pressure. 

Bitcoin’s Money Growth 

Although Bitcoin’s annual issuance rate is often compared to a nation state’s inflation rate (usually measured as CPI growth), the comparison is incorrect. Currently, Bitcoin does not serve widely as a unit of account and few goods and services are denominated in Bitcoin terms. Therefore, a basket of goods and services denominated in Bitcoin does not exist and an inflation rate cannot be calculated. The more appropriate comparison is growth in base money or perhaps a measure that represents credit creation in addition to base money, like the Federal Reserve’s balance sheet. 

Although critically important for Bitcoin’s appeal as a store-of-value, Bitcoin’s issuance rate is uninteresting precisely because of its predictability. Current issuance is 3.6% and will drop to 1.8% when the block reward halves around May 2020. However, the implications for miner-led selling flow are significant and market participants’ belief that the halving is impactful on prices may become self-fulfilling. 

Although early in Bitcoin’s development, a rudimentary credit market is being developed for Bitcoin. Already we are seeing the rise of short-term, collateralized lending from providers such as Genesis Capital. Discussion of Bitcoin-denominated bonds are active and in the far future, a bank that takes Bitcoin deposits and originates Bitcoin loans is possible. The effective upper limit on Bitcoin in circulation could therefore exceed the current supply if trusted parties begin issuing credit-like instruments, not too dissimilar to the differences between Money Stock (M1 and M2) and the Monetary Base in traditional money markets. How this will impact the Bitcoin economy remains to be seen and deserves greater study. 

Further Study 

As the field of network analysis advances and as crypto assets continue to develop, data on the state of these economies will improve in quality. For example, a method that can characterize the entire address space will allow the reporting of headline metrics and its subcomponents. The rise of a robust lending market lends itself to yield curve analysis and credit creation metrics. Many existing metrics can also be reported on a per capita basis using active addresses or other metrics representing user count. 


A small country (like Iceland, containing a population of 360,000) has a government statistical agency publishing a complete set of macroeconomic indicators that report on the state of the economy. Bitcoin and other crypto-assets have user counts and levels of economic activity exceeding small nation-states, and these virtual crypto-economies deserve first-class macroeconomic reporting. Coin Metrics is committed to using the principles and best practices established by government agencies and private organizations to provide transparency on the state of these networks. 

Network Data Insights

Summary Metrics

BTC and BCH market cap grew for the second straight week (measured on a week-over-week basis), while ETH, XRP, and LTC bounced back from down weeks. However, although market caps were up across the board, BTC and ETH usage were both down: BTC and ETH active addresses were down over 2% week-to-week, and transaction and transfer counts are both down by over 4%. 

After surging to new all-time highs at the end of October, BTC hash rate is down over the last week. There are reports that the drop in hash rate may be related to the end of China’s rainy season, which causes some hydropower stations to decrease their capacity and forces many miners offline in search of cheaper electricity. However, it may also be due to a lag between price and hash rate decline, following the price drop from $10k to $8k. Alternatively, it could simply be due to variance. 

Network Highlights

BTC realized cap continues to hit new all-time highs; on November 3rd BTC realized cap hit a new high of $102,936,158,856. “Realized capitalization” is calculated by valuing each unit of supply at the price it last moved and can be thought of as the average cost basis for crypto asset holders. Read more about how we calculate realized cap in State of the Network Issue 14.

BTC’s realized cap has grown by about 30% since the beginning of 2019, which is more than any other large crypto asset over the same period. For comparison, XTZ’s realized cap grew by 12%, LTC’s grew by 3%, and ETH’s declined by 9%.

After ETH total daily transaction fees threatened to pass BTC over the course of September, BTC has now pulled back out into the lead. BTC fees surged to over $439,000 on October 26th, and have remained at $300,000 or more for most of the days since. ETH had a little over $100,000 daily fees on October 26th, and has not topped more than $106,000 since.

Both BTC and ETH continue to dominate all other blockchains when it comes to overall fees. The following chart shows the percent of mining revenue (which we define as total transaction fees plus newly issued tokens, i.e., block reward) composed of transaction fees, averaged over the last three months.

In the long run, most blockchains’ block rewards will gradually decrease towards zero due to regularly scheduled block reward halvings. As block rewards decrease, fees begin to become a larger percentage of overall mining revenue and therefore become a more and more critical part of a chain’s long term sustainability and health.

Interestingly, after the October surge in ETH fees, about 4.5% of ETH’s mining revenue is composed of fees, compared to 1.8% for BTC. However, no other chain comes close to either ETH or BTC. Only 0.23% of DASH mining revenue is composed of fees, and only 0.03% of BCH revenue currently comes from fees.

Market Data Insights

Bitcoin’s two biggest forks have outperformed most of the rest of the major crypto assets over the past month. Surprisingly, Bitcoin Cash SV (BSV) has provided top monthly returns, rising 56% over the last 30 days. Bitcoin Cash (BCH) is not far behind, with a 31% gain, compared to a 13% gain for Bitcoin itself. The following charts show trailing month returns, from October 3rd to November 3rd.

China-based crypto assets, like Tron and NEO, have also exhibited strong gains over the last month. The direct cause of Tron and NEO’s outperformance remains unclear, but remarks from Chinese President Xi Jinping urging greater development in blockchain technologies may have been one of the catalysts, as we wrote about in last week’s State of the Network.

Meanwhile, after relatively strong market cap growth over most of the past year, Tezos’ price has started to decline, dropping 6% over the last 30 days.

CM Bletchley Indexes (CMBI) Insights

Despite a historic Friday that saw Bitcoin pump over 40%, the Bletchley Indexes returned mixed results this week. The Bletchley 20 (Mid Cap) and Bletchley 40 (Small Cap) performing the best, returning 0.5% and 1.5% respectively, while the Bletchley 10 (Large Cap) had a disappointing week, falling 2.5%.

October broke a 3 month downwards streak for crypto asset prices, with all indexes returning above 8%. The Bletchley 20 (Mid Cap) performed the strongest, appreciating 15% against the USD and 4% against Bitcoin. 

There was no asset turnover between indexes during the November monthly rebalance.

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics is hiring! We recently opened up 6 new roles, including Blockchain Data Engineer and Data Quality and Operations Lead. Please check out our Careers page to view the openings.

As always, if you have any feedback or requests, don’t hesitate to reach out at

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in-depth research and analysis.

Coin Metrics' State of the Network: Issue 23

Tuesday, October 29th, 2019

The Signal and the Nonce: Hunting for ASICs in Nonce Distributions

The Great ASIC Debate

Since the early days of Bitcoin, the community has argued about whether to fight against increasingly powerful mining hardware. 

In Bitcoin’s early days, mining was performed by CPUs, which are the standard processing units found in most computers. This gave practically anyone the chance to mine BTC since no specialized computer hardware was needed.

As time went on, some miners started using more and more powerful hardware in order to gain an edge over the competition. Miners eventually started using GPUs, the more powerful “graphics processing unit,” which are typically used for gaming and 3D rendering but can be used for many general purposes. GPUs are more expensive than CPUs but are still generally affordable for the average individual.

Then came ASICs. ASICs, which stands for “application-specific integrated circuits,” are pieces of specialized mining hardware that are optimized to mine specific hashing algorithms. ASICs are designed to mine crypto, and only to mine crypto. They are significantly more powerful than GPUs in terms of raw hash power.

With ASICs came a shift in mining economics. Simply put, the companies that manufacture new ASICs have a large advantage over the rest of the mining community since they inherently have a hash power edge (at least temporarily, until other ASIC manufacturers catch up), and also control the supply of new ASICs being released to the market. The large capital upfront investment necessary to manufacture a new line of ASIC hardware also reduces the number of people able to participate in this business. There are large economies of scale for large miners, which makes it harder for the average miner to compete.

Because of this, many projects have tried to protect against ASICs. Notably, after Bitmain and others announced they were developing a Monero-specific ASIC in early 2018, Monero hard-forked to remain “ASIC-resistant,” to keep the mining community as decentralized as possible. Monero has had several hard forks since then to try to stay ahead of ASICs and disincentivize their further development. 

Similarly, Ethereum has been “ASIC-resistant” for most of its history (since version 1.0, Ethereum’s hashing algorithm Ethash has been designed to be ASIC-resistant) but ASIC manufacturers are now starting to catch up. As a result, many within the Ethereum community are now arguing for the implementation of ProgPow, a revision of Ethash to make Ethereum once again ASIC-resistant. 

Although these projects continue to fight to remain ASIC-resistant, it is hard to consistently stay ahead. Large miners are incentivized to develop ASICs since mining specialized hardware on ASIC-resistant blockchains can yield a large advantage over other smaller miners. This means that there is a constant game of cat and mouse between ASIC miners and blockchain developers.

Into the Nonces

Fascinatingly, examining nonce distributions gives a potential look into the rise of ASIC mining on certain chains and the subsequent attempts to keep them at bay. 

Proof of Work mining consists of hashing a block’s header over and over again until its hash is less than a protocol-defined target value. This is done by taking the block header as an input, and then running it through a cryptographic hashing algorithm, which for Bitcoin is Secure Hash Algorithm 256 (SHA-256, applied twice in a row).  

In order to get a different hash for each attempt without having to fully rebuild a new block header, a special field is provided to miners as part of the header: the nonce field. This is an arbitrary number that miners can change in order to modify the header and produce a hash that is less than the target hash value. The nonce is a number that can vary from 0 to whatever the upper limit set by each protocol. 

Given that, in theory, any nonce can result in a winning hash, it’s not unreasonable to expect that nonces are chosen randomly and therefore distributed uniformly. However, analysis of many blockchains shows that only few follow that expectation. Explanations for this are varied, but changes in nonce-picking strategies can often be linked with the introduction of new mining hardware suggesting that different mining hardware have different nonce-picking strategies.

This pattern was seemingly first noticed for Bitcoin by Twitter user @100TrillionUSD in early January 2019. Further analysis has shown some strange patterns in other chains like Monero, Ethereum, and Litecoin.


The most famous nonce distribution is Bitcoin’s. At the start of its history, it presents a common pattern: a lot of nonces are close to 0. This can be explained by a simple strategy that consists of incrementing the nonce for each hash. As hashrate was very low in Bitcoin’s early history, mining was performed using CPUs only and a winning hash was commonly found before going over all the possible nonce values. Sergio Lerner exploited this fact to give the most serious attempt at identifying Satoshi’s coins.

After GPUs were introduced, the nonce space became seemingly random. But around height 400,000, a new pattern, yet unexplained, emerged leaving thin stripes of nonce values that are picked less frequently by miners.

BitMEX research wrote extensively about this pattern but found no clear explanation for it.

Looking at a histogram of Bitcoin block’s nonces (over the course of its entire history) clearly highlights that pattern, as well as the prominence of low nonces.

The red line is the expected value assuming a uniform distribution

For even more granular analysis, one can look at the distribution of bits. The nonce field in Bitcoin is comprised of 4 bytes or 32 bits. An analysis of the average value of each of the 32 bits of the nonce shows some interesting patterns:

The darker a cell, the more often the nonce bit is set to 1 instead of 0; X-axis is time in blocks

At the beginning of Bitcoin’s history, the higher bits were often set to 0 as miners’ nonce-picking strategy was to simply increment it. The lower byte (the last 8 bits at the bottom of the chart) seem to have always been used with some sort of patterns but only recently have changes in patterns been observed in the lower bits. 


One of the most interesting assets to apply nonce analysis to is Monero. It’s also one of the most analyzed, with various articles and tools looking into it.

Monero upgrades by hard-fork every 6 months and some of the past few upgrades have been accompanied by tweaks to the proof of work algorithm in order to circumvent dedicated mining hardware, but not general-purpose hardware. The first of those hard forks was somewhat contentious as it resulted in several forked projects/assets.

We can therefore study the impact of these changes on the nonce distribution. 

At first glance, we can notice several interesting patterns, but things get more interesting when we overlay network difficulty and the scheduled hard forks (that tweaked the PoW algorithm) on top of this chart.

Red lines indicate the timing of hard forks whereas black lines represent network difficulty

We can see that all 3 PoW upgrades lead to some drop in difficulty and that 2 of them stopped pre-existing nonce patterns. Interestingly, the introduction of those same nonce patterns was also associated with a sharp rise in network difficulty.

Simply by observing these nonce distribution patterns against difficulty and PoW-adjusted forks, we can potentially see the effectiveness of the first fork in stopping the first generation of dedicated hardware. Additionally, we can see the rapidity at which some miners came back with a second generation of hardware after the second fork, which was again thwarted with a third fork. 


At first glance, Ethereum’s nonce space shows barely any nonce distribution patterns or irregularities.

Looking closer, we can see some darker horizontal lines at the bottom of the space after block number 7,000,000. And if you zoom in, you can spot slanted lines starting from the bottom of the space between numbers 2,000,000 and 4,000,000. Those are likely the signature of a simple nonce-picking strategy: starting at 0 and incrementing the nonce at each try.

A histogram of the block’s nonce shows that there’s a slight preponderance of lower value nonces over time:

However, a very interesting pattern is visible if we look at the average value of each bit of the nonce over time (note that Ethereum’s nonce is comprised of 64 bits, not 32 as in Bitcoin):

The darker a cell, the more often the nonce bit is set to 1 instead of 0; X-axis is time in blocks

Starting at around block 1,380,000 the middle bits of the nonce started to get set to 0 much more often than the other bits. Over time, other bits started having non-random uses too. What makes this interesting is that a cursory glance at the overall nonce distribution or histogram doesn’t reveal this pattern because tweaking the middle bits doesn’t visibly affect the nonce’s histogram.

Interestingly, Ethereum Classic’s nonce bit distribution shows the exact same pattern:

The darker a cell, the more often the nonce bit is set to 1 instead of 0; X-axis is time in blocks

The white vertical stripes at the top indicate some miners were incrementing the nonce from 0.

Bitmain announced the first publicly known Ethash ASIC miner in April 2018 and said first deliveries were expected in mid-July of that same year. Annotating the previous chart with both dates shows something very interesting:

Dotted red line: E3 announcement, solid red line: E3 first known deliveries

From a first glance, the average value of bit 41 for Ethereum and Ethereum classic blocks dropped around the time of the Antminer E3 announcement. Focusing on this specific bit on both Ethereum and Ethereum classic, the pattern is even more striking:

Dotted red line: E3 announcement, solid red line: E3 first known deliveries

Prior to mid-March 2018, the average value of bit 41 hovered around 0.5 (which is the expected value assuming uniform distribution as represented by the grey horizontal dashed line). However, it started to be set increasingly to 0 from that point onwards on both chains and at the same rate. Its average value then dropped drastically right at the time the first deliveries were scheduled to happen (July 16th, 2018, red solid horizontal line) but only for Ethereum. On both Ethereum and Ethereum classic, bit 41’s average value settled in mid-June 2018, one month before the first announced deliveries of E3 miners, but its value further dropped only for Ethereum.

Dotted red line: E3 announcement, solid red line: E3 first known deliveries

Looking at Ethereum’s top 5 mining pool at the time, we can see that the average value of bit 41 started dropping across all pools but that from July 16th onwards, the date of the first announced deliveries of E3 miners, its value further dropped for only 2 major pools: Sparkpool and F2Pool, both administered in China.

The Battle Continues

As ProgPow looms and Monero continues to hard-fork to try to stay ahead of ASICs, the battle against specialized hardware is likely not going away anytime soon. As the war wages on we’ll be watching closely, and continue the hunt for the signal in the nonce.

Network Data Insights

Summary Metrics

BTC rallied over the past week, with a 3.5% rise in market cap. Comparatively, ETH had a down week, with market cap dropping 2% from the previous week. 

BTC daily transaction fees are beginning to pull significantly back ahead of ETH, after ETH threatened to flip BTC fees over the last few months. BTC’s average fees grew by 42.2% week over week, while ETH’s grew by 18.4%. BTC had a daily average of over $268K fees over the last seven days, while ETH only averaged $93.5K. 

Interestingly, although BTC led ETH in most other metrics, ETH had a 4.6% increase in daily transaction count, while BTC’s daily average decreased by 0.1%. ETH’s transfers (i.e. transactions that include a monetary exchange), however, decreased by 2.9%, while BTC transfers increased by 2.2%. This suggests that Ethereum Dapp (i.e. decentralized application) usage may be increasing, since Dapps often produce non-monetary transactions.

Network Highlights

USDT ETH adjusted transfer value rose significantly over the past week, and has now taken a commanding lead over USDT Omni. As we’ve covered in past issues, Tether exists on multiple different protocols, the two biggest of which are Bitcoin based USDT OMNI and Ethereum based USDT ETH. Over the last several months, usage has been shifting from the Omni to the Ethereum based version of Tether. Notably, Tron based Tether is also reportedly on the rise

On October 25th, BTC had one of its biggest upward price movements in 2019. However, the movement was relatively quick, with the price rocketing up and then falling back down over a period of 12 hours. This led to the highest 12 hour BTC rolling return since 2017. But in terms of end of day, 24-hour change, it was the second-highest of 2019. The below chart shows the daily growth (from the previous day) of BTC price over the course of 2019, with October 25th highlighted with an orange dot.  

Although the 25th was a big day in terms of price, many other on-chain metrics did not follow suit. Adjusted transfer value only grew by 14.6% on October 25th, which is far below 2019 highs. 

Similarly, BTC active addresses (which is the unique number of addresses either sending or receiving a transaction over a 24 hour period) only grew by 9% on the 25th, which is a little below the average of 10% daily growth over the course of 2019.

Market Data Insights

Despite experiencing one of the largest 24-hour returns in the history of Bitcoin, price is up a more modest 16 percent when measured on a weekly basis. Prices continue to experience high directional correlation but wide dispersion in returns with some major assets nearly flat for the week. 

Among major assets, Bitcoin Cash SV led the market with a 48 percent return. Bitcoin Cash SV remains one of the most volatile large-capitalization assets with three-month rolling volatility running at over 100 percent annualized. We previously wrote that fork uptake (the number of native units that have moved after a fork event) is particularly low and only roughly 8 million units of Bitcoin Cash SV have been claimed post-fork. With the low claimed supply and being delisted from major trading venues, the amount of Bitcoin Cash SV available for price discovery remains low. 

The direct catalyst for the large movement in prices remains unclear, although some market participants believe that remarks from Chinese President Xi Jinping urging greater development in blockchain technologies as the cause. In support of this thesis, major China-based blockchain projects TRON, NEO, Ontology, Qtum, and others have rallied much sharper than the rest of the market. 

Although certain previous movements have shown limitations in the market microstructure of crypto markets or instances of suspected price manipulation, orderly markets were observed over the past week. Despite extremely rapid price movement, spreads between major exchanges remained small indicating that market participants have the ability to quickly transport liquidity across markets. A similar analysis performed on BitMEX’s perpetual futures contract and major markets quoted in Tether also show orderly trading. 

The large movement has meaningfully changed the distribution of Bitcoin by price at the time of last on-chain movement, a proxy for estimating each unit of Bitcoin’s cost basis. The current snapshot shows very little Bitcoin with a cost basis above $13,000. Large amounts of Bitcoin are distributed in the $2,000 to $13,000 range and nearly 4 million Bitcoin or 22 percent of total supply has a cost basis below $500. 

Here we show the change in the distribution between October 24 (immediately before the large price movement) and October 26 (immediately after). Extremely little activity was detected from holders with a cost basis above $13,000 or below $7,000 suggesting that the increase in price was not yet sufficient to incentivize these holders to sell. Instead, it is most likely Bitcoin between the $7,000 and $8,500 range was sold, although these amounts may be slightly overstated due to normal exchange hot wallet activity. A small amount of Bitcoin in the $9,000 to $12,000 range did show some on-chain movement, however. 

CM Bletchley Indexes (CMBI) Insights

After the market activity in the second half of last week, all Bletchley Indexes performed strongly, returning between 9% and 13%. Not many assets performed as well as Bitcoin over the weekend, which lead to the Bletchley 10 and Bletchley Total performing best week-on-week. This is further highlighted by the underperformance of all indexes in comparison to Bitcoin. 

Interestingly, the end of week performance figures do not tell the full picture of what occurred across markets during the week. As demonstrated below, all indexes were down close to 10% mid week, with Bitcoin down more than any index. 

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics is hiring! We recently opened up 7 new roles, including Blockchain Data Engineer and Data Quality and Operations Lead. Please check out our Careers page to view the openings.

As always, if you have any feedback or requests, don’t hesitate to reach out at

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in-depth research and analysis.

Coin Metrics' State of the Network: Issue 22

Tuesday, October 22nd, 2019

Weekly Feature

Bitfinex Market Share is Declining

Bitfinex, the controversial Hong Kong based exchange at the heart of ongoing investigations against Tether, has been one of the biggest cryptocurrency exchanges in the world since its launch in 2012. But recently there are signs that its market share may finally be slipping. 

Since the beginning of September, Bitfinex’s BTC-USD market share has been decreasing, while Luxembourg based exchange Bitstamp’s market share has grown. The below chart shows the market share of total daily BTC-USD trading volume (smoothed out using a 7 day moving average) for the following exchanges: Bitfinex, Bitstamp, Bittrex, CEX.IO, Coinbase, Gemini, itBit, Kraken, and Liquid. 

Although it may be tempting to think that Bitfinex’s volume decline is related to the recent Hong Kong protests, the most likely explanation is a combination of increased regulatory scrutiny and decreased fees. 

In April, the New York Office of the Attorney General’s (NYAG) announced it was beginning an investigation for alleged fraud against Bitfinex’s parent company, iFinex, which owns both Bitfinex and the Tether stablecoin. iFinex reportedly lost $850 million in a deal gone bad and then borrowed “at least $700 million” from the reserves that back Tether to cover the losses. After Bitfinex challenged the investigation, on August 19th the New York Supreme Court ruled that the NYAG has jurisdiction over Bitfinex, allowing the NYAG to move forward with the investigation.

Furthermore, in early October, several plaintiffs opened a new class action lawsuit against Bitfinex for price manipulation related to Tether, alleging over $1 trillion in damages.

Concurrently, at the end of July, Bitstamp announced they would be implementing lower trading fees. Initially, they planned to put the new fee schedule into effect on August 1st, but ended up delaying the launch until August 20th. Bitstamp’s new fee schedule lowered fees for high-volume customers, making their platform more attractive for institutional grade investors:

Notably, Bitstamp’s new fees are significantly less than Bitfinex’s taker fees for customers who have at least $1 million worth of trades or more over the previous 30 days. Additionally, on October 7th Coinbase started implementing higher fees for its pro users, although we have not yet observed a significant decrease in Coinbase’s BTC-USD volume market share. 

The following are Bitfinex’s current fees:

Although the Hong Kong protests began in March, Bitfinex had substantially more trading volume than Bitstamp from April through July. Bitstamp started catching up in late August, when the fee change went into effect, and the NY Supreme Court ruled favorably for the NYAG’s investigation against iFinex. By September, Bitstamp had passed Bitfinex in terms of total trading volume per month.  

Compounding iFinex’s problems, usage of the LEO token (which was created by iFinex and supported by Bitfinex) has mostly stalled since its launch in June. The following chart shows the number of unique addresses that hold any balance of LEO (the following stats are all for usage of Ethereum based LEO, as the EOS version has gotten very little usage). As of October 17th, there were only 1,948 addresses that held any LEO.

The below chart shows the number of LEO addresses that hold  balance of at least X amount of USD, where X ranges from $1 to $10,000,000. At time of publication, a little over 500 addresses hold at least $1,000 worth of LEO, and 1,182 addresses hold at least $100.

Furthermore, LEO’s active addresses have been steadily declining. We define “active addresses” as the count of unique addresses that either send or receive a transaction over the course of a day. On October 17th, LEO had 17 active addresses.

Network Data Insights

Summary Metrics

By most measures, the major crypto networks declined slightly over the past week. BTC and ETH market caps were down 3% and 3.9%, respectively, week over week. XRP, however, had a relatively good week, with a 6.4% increase in market cap.

XRP and LTC both had relatively large gains in adjusted transfer value compared to both BTC and ETH. ETH had a particularly rough week, dropping by 4.4% in adjusted transfer value and 7% in fees. This is the third week in a row that ETH fees have decreased, after showing strong increases through much of September. 

Network Highlights

BTC continues to dominate in terms of daily transfer value. The below chart shows the market share for adjusted transfer value between BTC, ETH, LTC, BCH, and XRP, smoothed using a 7 day moving average. As of October 20th, BTC has a little over 79% of transfer value market share, compared to about 11%, 5%, 4% and 1% for ETH, XRP, BCH, and LTC, respectively.

On October 19th, the Bitcoin network mined its 600,000th block. Later that day, the BTC supply reached 18,000,000 (technically the supply did not reach 18,000,000 at exactly the 600,000th block because historically, some miners did not claim their block rewards). As of end of day October 20th, BTC had a supply of 18,003,467.

Market Data Insights

Markets over the past week saw moderate dispersion in returns with many assets experiencing small gains and a slightly larger number of assets experiencing small losses. Among the large capitalization assets, XRP (+6%), Monero (+6%), and Bitcoin Cash SV (+10%) are notable outperformers. 

Despite a sharp decline in Bitcoin prices roughly one month ago, volatility measured on a rolling three month basis is again approaching three year lows. Prices staying range bound previously around $10,000 and now at $8,000 have caused volatility to decline to 59%. Previous recent lows in volatility occurred in November 2018 and March 2019 when prices were range bound around the $6,500 level and $3,500 level, respectively. In both those instances, volatility declined below 50% and a sharp change in price followed. 

Lowered volatility will incentivize traders to build up positions with increased leverage around the current price. The longer volatility remains muted, the more likely a violent reaction in price will occur. Should prices remain range bound at around $8,000 for another month or so, volatility will fall to levels where previous sharp changes in price occurred. 

Meanwhile, volatility for some other assets are approaching all-time lows and are now matching Bitcoin’s volatility levels, indicating that the irrational exuberance for these assets has faded.

CM Bletchley Indexes (CMBI) Insights

CMBI Design Considerations

When considering the pricing methodologies of indexes two of the primary design considerations include:

  1. Timeliness of data - Proponents of timeliness often argue that ‘markets are markets’ and the most recently available data should be reflected. The indexes’ level should reflect the price that can be obtained by trading each one of its constituents in the market at any given point in time. 

  2. Manipulation Resistance - The other side of the argument is that taking the most recently available trade data can lead to easily manipulatable index levels, particularly in thinly traded markets. Thus, ingesting more data can add robustness to a reference price, be it more breadth (increased number of exchanges upon which price is ingested) or more depth (taking more observations than just the most recent trades e.g. within a ‘x minute’ time window). Proponents of reliability often argue that a more robust price should be used to prevent manipulation.

This is one of the key design considerations of the CMBI index suite that is currently being constructed behind the scenes at Coin Metrics. It is very easy to believe that in the short term a derivatives crypto asset product could track an index, either single asset or multi asset. But trade-offs between timeliness and reliability get particularly hairy when the derivative product’s volume is greater than the underlying asset. 

To address much of this, CMBI design has leveraged both Coin Metrics’ Reference Rates and Real Time Reference Rates. CM Reference Rates are designed to be robust and not easily susceptible to manipulation, achieved through the use of a time-weighted volume-weighted median price, whilst CM Real-Time Reference Rates are designed to be timely, achieved through the use of the most recent trade data available from selected markets. When implemented hand in hand for index construction can overcome many of the pitfalls of each design. 

Bletchley Weekly Performance

Most of the Bletchley Indexes started the week with declines before recovering much of the losses over the weekend. Large cap assets performed best with the Bletchley 10 finishing the week flat, despite both Ethereum and Bitcoin finishing the week with losses. 

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics is hiring! We recently opened up 7 new roles, including Blockchain Data Engineer and Data Quality and Operations Lead. Please check out our Careers page to view the openings.

As always, if you have any feedback or requests, don’t hesitate to reach out at

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in depth research and analysis.

Coin Metrics' State of the Network: Issue 21

Tuesday, October 15, 2019

Weekly Feature

BTC Is Close to $1B of All-time Fees, But Fee Growth is Slowing

BTC is close to eclipsing $1 billion of cumulative, all-time transaction fees. As of October 13th, there have been $996,458,718 worth of fees paid on the BTC network.

Coincidentally, BTC is also close to reaching another milestone: $15 billion worth of all-time miner revenue (also known as “thermocap”). “Miner revenue” includes block rewards (i.e. newly issued tokens) and transaction fees, both of which are typically paid to miners. 

However, since early 2018, BTC fees have been accounting for less and less of cumulative miner revenue. On January 31st, 2018, BTC fees accounted for about 12% of the total miner revenue (block rewards account for the remaining 88%). As of October 13th, 2019, BTC’s cumulative fees were only 6.6% of total miner revenue. This means that since February 2018, BTC cumulative fees have not been growing as fast as cumulative block rewards. 

BTC median fees shot up to nearly $34 in late 2017, compared to a peak of only about $3 for ETH. BTC median fees have since fallen and have remained at less than $4 since February, 2018. This extreme difference in median fees is one of the main reasons for the decline in cumulative fees’ share of total miner revenue. 

Comparatively, as of October 13th, cumulative fees make up about 3.6% of ETH’s cumulative miner revenue. But, compared to BTC, the ratio has stayed relatively flat since the start of 2018. 

This is mostly due to the fact that ETH’s fees have grown much faster than BTC’s over the period. BTC’s cumulative fees have grown about 74% since January 1st 2018, while ETH’s have grown close to 400%.

BTC and ETH’s market cap to thermocap ratios (“thermocap” is another term for cumulative miner revenue) also flipped in early 2018. The market cap to thermocap ratio can potentially serve as a (rough) proxy of a crypto asset’s valuation to revenue, where the market cap represents an approximate valuation and thermocap represents the total revenue generated by the network. As of October 13th, BTC’s market cap to thermocap ratio is 10.3 compared to 3.2 for ETH. Interestingly, the S&P price to sales ratio is currently about 2.18.

Network Data Insights

Summary Metrics

Market cap was up for all five major crypto assets over the past week. For the fifth week in a row, ETH’s market cap outperformed BTC’s. ETH’s market cap was up 4.6% for the week (and was up 1% last week), while BTC’s market cap grew by 2% (compared to a 3.1% loss last week). 

Hash rate, however, grew more for BTC than ETH. BTC’s approximate hash rate was up 3.5% over the last week while ETH’s dropped by 0.5%. This is the second week in a row that BTC’s hash rate has outgrown ETH’s hash rate. 

Network Highlights

The combined total adjusted transfer value of the seven stablecoins we track (DAI, GUSD, PAX, USDC, TUSD, USDT, and USDT_ETH) approached all-time highs at the end of September, hitting $1,546,234,810 on September 24th. However, the total has since dropped, averaging about $618,000,000 a day over the first 13 days of October. The below chart shows the seven day moving average of the combined total.

A majority of the total stablecoin adjusted transfer value comes from Tether, which is still the biggest stablecoin by most measures. The activity on Tether continues to shift from USDT (which is the original, OMNI based version) to USDT_ETH (which is the newer, Ethereum based version of Tether). 

Market Data Insights

Over the past month, most major assets experienced sharp declines. Bitcoin is down 20 percent over this time period driven primarily by a concentrated movement from $10,000 to $8,000 on September 24. Notice the stair step-like pattern for Bitcoin prices reflecting long periods of muted price volatility interspersed with short periods of large and concentrated price movement. 

Ethereum is flat over this time period but significantly outperformed Bitcoin, perhaps because of an emerging recognition that demand for Ethereum's block space represented by fees is growing (and momentarily eclipsed Bitcoin's daily fees). Both XRP and Stellar have experienced modest single-digit gains for this month. Most other major assets are down, but less than Bitcoin. 

Notably, there was no significant impact on prices despite major developments in crypto markets over the past week -- a class action lawsuit was filed against Tether and Bitfinex, the SEC formally rejected Bitwise's ETF and filed an emergency action against Telegram's token offering, and multiple members of Libra Association dropped out due to regulatory pressure. 

The fact that prices for smaller assets remained resilient (relative to Bitcoin) in the face of these developments suggests that increased regulatory scrutiny has already been priced into the market. 

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics is hiring! We recently opened up 7 new roles, including Blockchain Data Engineer and Data Quality and Operations Lead. Please check out our Careers page to view the openings.

  • The CM Bletchley Index (CMBI) Insights section will return next week.

As always, if you have any feedback or requests, don’t hesitate to reach out at

Subscribe and Past Issues

Coin Metrics’ State of the Network, is an unbiased, weekly view of the crypto market informed by our own network (on-chain) and market data.

If you'd like to get State of the Network in your inbox, please subscribe here. You can see previous issues of State of the Network here.

Check out the Coin Metrics Blog for more in depth research and analysis.

Loading more posts…