Coin Metrics’ State of the Network: Issue 48 – Creating a Better Supply Measurement: Introducing Free Float Supply

Creating a Better Supply Measurement: Introducing Free Float Supply

An accurate measure of supply is crucial in determining a cryptoasset’s value. However, the definition and methodologies of calculating supply are not yet widely agreed upon or standardized, which leads to different supply counts across the industry.

One definition of supply is “current supply”, which represents the total amount of native units that are visible on a blockchain’s ledger. This number is publicly available and accessible to any individual that downloads a cryptoasset’s blockchain ledger, which acts as a source of truth. 

However, utilizing this metric can grossly misrepresent a cryptoasset’s underlying liquidity and capitalization, as most have a certain amount of supply that is effectively out of circulation. This poses an issue for creating investable financial products such as indexes that rely on supply data to determine a constituent’s weight in a multi-asset index. 

Understanding what portion of supply is unavailable to markets can help market participants to make smarter and more informed investment decisions. Currently, the market relies primarily on the ‘Reported Supply’, which is generally provided by the asset creator and is visible on the majority of data distributor dashboards. However, in the absence of regulatory requirements or strong incentives, reporting by these entities has historically proven to be infrequent and sometimes lacking in accuracy. 

Coin Metrics recently announced the CM Free Float Supply, a new metric that is being developed to more accurately represent the supply of an asset available to the market. By applying a standardized approach, this metric represents a cross blockchain unified portrayal of the crypto market’s liquidity. For greater detail on the rationale for the methodology decisions that make up CM Free Float Supply, please refer to our blog post, Cryptoasset Free Float: An Exploration of Supply Dynamics.

CM Free Float Supply overcomes the challenges of misrepresenting supply by restricting categories of token holders that do not provide liquidity to markets. Coin Metrics’ approach to Free Float Supply determination can be represented as follows:

1 – This approach has been taken for UTXO chains as opposed to the traditional mechanism of assessing individual UTXOs for inactivity. The purpose of this is to develop a more ubiquitous and consistent approach for applying logic across all blockchains.

This approach leverages some of the best practices for supply determination from traditional capital markets and applies them systematically to cryptoassets. Many of the values produced will not be familiar at first glance, but we hope that they help provide a new and more accurate view of cryptoasset liquidity in markets. At Coin Metrics we strive to set the standard for transparent and actionable cryptoasset data, and we believe this new supply metric will aid us in achieving that goal.

Case Studies on the Free Float Supply of Different Asset Groups

Given the varying architectures and token economic models of blockchains, it was essential to create a methodology that can be applied consistently across all cryptoassets, or risk introducing large amounts of subjectivity and expert judgment that could jeopardize the usefulness of the metric. It is therefore very important that our data suite is unbiased and can be interpreted across any asset class. 

In this section, we detail how the Coin Metrics Free Float Supply Methodology has been applied across different blockchain architectures and token economic models, including UTXO chains, forked assets, and Ethereum tokens.

Case Study 1: UTXO Blockchains

Bitcoin was the original UTXO blockchain, and many others have followed since. An unspent transaction output (UTXO) blockchain is a way of structuring a ledger whereby all coins (or bundles of coins) are only spent once. A UTXO is an output that a network user receives and has the right to spend at a time in the future.

The early UTXO chains are very simple to audit as most were launched with no pre-mine (i.e. private mining by select participants before the source code is released to the public to mine), no ICO, no foundation, and thus very few restrictions on their supply.

In the calculation of BTC’s free float supply in our free float methodology explanation blog, we considered send and receive transactions as the sign of an active address. However, after further investigation and internal deliberation, we have improved our approach and consistent treatment of the methodology across blockchains by only considering send transactions specifically as signs of activity. The rationale behind this is that for a small cost, supply values could be easily manipulated.

Note, this change has now been applied to all UTXO blockchain supply figures, and as such the below BCH and BSV values have been improved from previously reported values as well.

Both Bitcoin and Litecoin had no ICO, issued no founding team tokens and had no pre-mine. But both chains are over five years old and thus have UTXOs that have not been touched in over five years. As such, in both instances, five years into the chain’s existence, you will notice the divergence of Free Float Supply from Current Supply.

Bitcoin has 4.1M BTC unmoved tokens in over five years and Litecoin has 3.0M LTC unmoved in over five years that have been excluded from the CM Free Float Supply.

Note: Reported Supply is currently the most quoted industry value that is generally the value observable on a blockchain or as reported by a foundation where one exists.

Case Study 2: Hard Forked Blockchains

A hard fork can occur when a change in consensus rules results in non-backward compatible software. If any network participants decide to run their own version of the blockchain’s software that creates a different set of consensus rules, a new blockchain with its own native token will be created.

These new native tokens are not only credited to the holders of the parent chain, but they carry the full history of the parent chain as well. Despite this, given the pseudonymity of cryptoassets, it is difficult to understand who these token holders are and predict what their actions will be. For example, are they aware of their right to claim their tokens, and do they know how to claim? As such, assuming the full history and financial ownership of the parent chain may not accurately reflect the realities of the forked chain. 

Coin Metrics has taken a conservative approach and excluded all tokens from addresses on the newly forked blockchain that have never been activated since the time of the fork. Activation here is defined as an address that sends any amount of their assets (or UTXOs), thus proving that all tokens in that address are owned and monitored.

Ethereum Classic (ETC) was the first hard fork that met the criteria as outlined in the Coin Metrics Fork Legitimacy Policy. Interestingly, activation of ETC native units after the fork was extremely high. Currently, addresses that hold over 97% of tokens have been transacted post fork. 

Similar to Bitcoin, Bitcoin Cash (BCH) and Bitcoin SV (BSV) do not have any foundation wallets, founding team tokens, provably lost tokens (>0.1%), burned tokens (>0.1%), vesting tokens, or pre-mine. However, despite the publicity and media attention surrounding both forks, they did not receive the same amount of activation that the Ethereum Classic fork did. Bitcoin Cash addresses holding 6.4M BCH and Bitcoin SV addresses holding 8.5M BSV remain untouched since the time of their forks.

Case Study 3: Stellar and XRP

For both Stellar and XRP, foundations have, to date, held the majority of the current supply. As such, Coin Metrics has identified a significant amount of restricted supply. Both foundations and founding team member wallets have historically demonstrated relatively high levels of activity (i.e. token selling) which has continually resulted in increases in their free floats.

In the case of XRP, all foundation and team tokens identified and all escrowed tokens have been removed to determine the free float. There were no identified provably lost tokens (>0.1%) or burned tokens (>0.1%) still present on-chain. 

Stellar is similar. To Coin Metrics’ knowledge, there are multiple Stellar Development Foundation (SDF) wallets, tokens that are vesting (in the SDF) and tokens that are burned. These tokens have been removed to determine the XLM free float. 

Case Study 4: Ethereum Tokens

Most tokens launched on Ethereum, including Ether (ETH), were released through an Initial Coin Offering (ICO). The majority of these projects allocate significant portions of the current supply to the foundation and team members. Foundation tokens are often utilized to fund long-term growth initiatives and can remain within foundation wallets for extended durations. Team member tokens can be subject to vesting schedules early on, but in many cases, founding team members hold the majority of their tokens longer term.

Ethereum launched in 2015 with a relatively small foundation and team token allocation. Since then, many of the tokens have been moved and sold. The Ethereum foundation now only holds around 0.5% of the total ETH supply. A further 1.0% has been identified as founding team tokens and 0.5% was provably lost in the Parity wallet bug. 

Chainlink is a typical cryptoasset born from the modern day ICO, where fewer than half of all created tokens were distributed at genesis. Upon launching, the Chainlink foundation and the founding team members held 65% of the token’s current supply. Since launch, there has been some movement from founding team wallets that has increased the free float supply from 350M to 379M.

Applying Free Float

The CM Free Float Supply is designed to generate and maintain the most accurate representation of cryptoasset liquidity across the market. Coin Metrics is able to do so by hosting our own nodes that allow for detailed on-chain analysis and the real-time tracking of specific tagged addresses. Further, hosting nodes and managing data allows Coin Metrics to independently verify all information from cryptoasset foundations and teams through on-chain forensics and verification. 

Whilst it may seem unfamiliar to consider the supply of Bitcoin as 14.3M or Bitcoin Cash as 12.0M doing so can provide some significant benefits when conducting market wide analysis and designing portfolios. Through better representing the supply and demand relationship of cryptoassets across the market, portfolio managers can reduce tracking error, unnecessary portfolio rebalance costs and slippage, and the management effort required. 

To that extent, CMBI market cap weighted indexes will leverage the CM Free Float Supply to create a suite of highly investable indexes that accurately track and represent the underlying cryptoasset market. These CMBI Indexes will provide markets and customers with industry leading solutions that aid in performance benchmarking and asset allocation.

Network Data Insights

Summary Metrics

Ethereum (ETH) had another strong week as it continues to rebound after the March 12th crash. ETH market cap increased by 9.0% week-over-week, and realized cap grew by 1.2%. Adjusted transfer value, transfer count, and daily fees all also showed solid growth, signaling that Ethereum on-chain activity is increasing along with the market cap.

Bitcoin (BTC) also had a relatively strong week, with market cap growing by 4.3%. BTC transaction fees grew by 34.2% week-over-week, which is BTC’s strongest week of fee growth over the last month.

Network Highlights

Tezos (XTZ) and Chainlink (LINK) market caps have both grown by over 70% over the last 30 days. Zcash (ZEC) and Ethereum (ETH) have also rebounded well, growing 51% and 48%, respectively.

More than one million Ethereum smart contracts have been created since March 12th. The number of smart contracts deployed on Ethereum grew from 13.36 million on March 12th to 14.41 million on April 26th. 

For context, the number of Ethereum smart contracts increased by about 470K between January 1st and March 12th.

Market Data Insights

All large capitalization cryptoassets saw gains this week with Tezos (XTZ), Stellar (XLM), and Cardano (ADA) managing gains in excess of 20%. One of the key portfolio level decisions for investors in the space is the relative allocation between major assets like Bitcoin and Ethereum versus the allocation to the longer tail of assets. Accurately determining which market regime we are in and how the various market capitalization segments will perform is important. Recent market activity has shown us some isolated examples of certain cryptoassets in the longer tail exhibiting outperformance with high beta to Bitcoin’s returns — a trend that deserves continued observation. 

Bitcoin volatility measured on a one month rolling window has almost completely reverted to normal levels. For the past few years, an annualized volatility of around 50% seems to be a critical level. Volatility rarely goes below this level and oftentimes bounces higher off it for brief periods of heightened volatility.

Usually lower volatility will cause investors to become more complacent and use more leverage, but the market crash on March 12 may have led to a longer lasting change in trader behavior. Implied volatility levels also have almost completely reverted and open interest on the major perpetual swap contracts are still down from recent highs. Volatility will likely remain muted until the memory of March 12 has faded somewhat and risk taking using leveraged financial instruments comes back. 

CM Bletchley Indexes (CMBI) Insights

All CMBI and Bletchley Indexes performed well this week as global markets continued to rally after the early March crash. The CMBI Bitcoin Index and the CMBI Ethereum Index returned 6.9% and 7.6% respectively, but it was the small-cap Bletchley 40 that was the strongest performer, returning 12.9% week-over-week. 

The strength of the cryptoasset performance was experienced across the market, evidenced by all even weighted indexes outperforming their market cap weighted counterparts. 

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics is hiring! 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 [email protected]

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 47 – Assessing the Long-Term Fallout from Crypto’s Black Thursday

Assessing the Long-Term Fallout from Crypto’s Black Thursday

by Antoine Le Calvez and the Coin Metrics Team

On March 12th 2020, now called “Black Thursday,” Bitcoin’s price dropped by nearly 50%, one of its largest drops since the Mt. Gox debacle.

Source: Coin Metrics Reference Rates

In our 43rd issue, we looked at the impact of that price drop on the crypto market structure – specifically, we explained how BitMEX’s liquidation engine could have amplified the drop, the lingering impact on liquidity left by the crash, and how stablecoins seemed to have gained from it.

In this issue, we’ll look at the impact of Black Thursday on on-chain data and exchange market share.

Transaction Fees

One of the immediate impacts of the Black Thursday drop was on transaction fees. As traders rushed to move coins in and out of exchanges in order to add margin to positions or to profit from arbitrage opportunities, the demand for block space heated up.

Most Bitcoin wallets use dynamic fee estimation, and txstats.com’s archives of these estimates allow us to see how they reacted to the increase in activity (txstats.com is presented in partnership with BitMEX Research). Fee estimates for rapid confirmation (10-20 minute wait time) went from 27 satoshis/byte of transaction data to 70 sats/byte in just a few hours.

Source: txstats.com

More surprisingly, median transaction fees measured in dollars for both Bitcoin and Ethereum shot up almost five-fold. Bitcoin’s fees remained elevated for two weeks while Ethereum’s fees recovered in two days.

Black Thursday highlighted in red. | Source: Coin Metrics Network Data Pro

Overlaying Bitcoin’s block space utilization (i.e. block weight used as a proportion of block weight available per block) shows that fees stayed elevated following Black Thursday despite block space utilization returning to normal levels.

Black Thursday highlighted in red. | Source: Coin Metrics Network Data Pro

What explains this discrepancy? One possible explanation is that the lingering effect on transaction fees could indicate the presence of a feedback loop between the various fee estimation APIs. Similar to how traffic jams form from a single clog in the system, a single and brief perturbation can have effects far outlasting it.

BitMEX

Black Thursday’s drop stopped when BitMEX, the leading Bitcoin futures platform prior to Black Thursday, experienced a denial-of-service attack. This led to many theories as to why the price recovered after BitMEX went offline, with the most prominent being explained in State of the Network Issue 43.

Following that fateful day, the number of Bitcoins held by BitMEX (on behalf of traders) first rapidly increased then dropped significantly over the following two weeks, stabilizing recently.

Black Thursday highlighted in red. | Source: Coin Metrics Network Data Pro

There could be many explanations as to why this happened. The early increase could have been caused by traders depositing coins to either trade the very high volatility or add margin to existing positions to avoid liquidation.

This is visible by looking at the distribution of deposits to BitMEX. On the 12th and 13th, the 90th percentile of deposits, by value, jumped several fold indicating that traders made larger deposits to BitMEX.

Black Thursday highlighted in red. | Source: Coin Metrics

As for the long drawdown, two main things can potentially explain it:

  • BitMEX lost some market share following Black Thursday
  • Crypto traders are deleveraging and withdrawing their remaining unused trading capital

One way to test the first hypothesis is to look at BitMEX’s market share in Bitcoin futures markets:

In both open interest (size of futures contracts held by traders) and in volume, BitMEX lost market share following Black Thursday. The biggest market share winner was Binance futures:

As for the deleveraging part, aggregate open interest fell by 50% in a single day and still hasn’t recovered (although it’s been slowly and steadily rebuilding since it bottomed).

Conclusion

It’s been more than a month since Black Thursday and while most of its immediate impact has now faded away with volatility reducing and spreads tightening, some of the longer term consequences are only starting to become visible.

Notably, since the crash there has been a reshuffling of the top futures marketplaces for crypto assets with BitMEX losing some of its market share to Binance. This may have an on-going impact across crypto markets, especially considering BitMEX’s outsized influence on price discovery. Only time will tell if BitMEX is able to recover the lost market share, or if the marketplace is undergoing a true changing of the guard.

Network Data Insights

Summary Metrics

Source: Coin Metrics Network Data Pro

Bitcoin (BTC) usage showed signs of growth this past week. BTC daily active addresses  grew 12.1% week-over-week, and topped 900K on both April 7th and April 15th. The last time BTC had more than 900K daily active addresses was in June 2019.

Bitcoin Cash (BCH) usage, on the other hand, declined over the past week. Active addresses dropped by over 46% week-over-week. Additionally, BCH estimated hash rate continues to decline after its April 9th halving.

Network Highlights

Stablecoins have gained another $1B in market cap since the start of April. Most of the growth continues to come from Tether issued on Ethereum (USDT_ETH) which went from a market cap of $4.43B on April 1st to $5.14B on April 19th.

The following chart was created using Coin Metrics’ community charting tool, which you can access for free here.

Source: Coin Metrics Community Data

Stablecoin transfers also continue to rise, with USDT_ETH once again leading the way. Although daily transfers for USDC, BUSD, and DAI all peaked on March 18th, daily transfers for USDT_ETH and PAX have continued to rise. USDT_ETH has grown from about 83K daily transfers in the beginning of November 2019 to about 141K daily transfers as of April 19th.

Source: Coin Metrics Network Data Pro

USDT_ETH transfer count has been growing faster than both BTC and ETH over the last 180 days. Although BTC and ETH both still have significantly more daily transfers, USDT_ETH transfer growth has had a noticeable uptick since mid-March. 

Source: Coin Metrics Network Data Pro

Market Data Insights

ETH leads all large cap assets this week with a 13% gain. As noted in the Network Highlights section, issuance of stablecoins designed using the ERC-20 standard and transfers of stablecoins on the Ethereum network continue to grow strongly. Given these developments, one of the more interesting debates right now is whether stablecoin growth is good or bad for ETH. 

Stablecoins are increasingly becoming critical tokens to the cryptocurrency industry and dozens of projects have been launched. The vast majority of projects have chosen to launch as ERC-20 tokens on ETH which should further solidify the ERC-20 standard and strengthen a network effect that is difficult for competing standards to overcome. When examining the first order impact, stablecoin growth should increase the demand for ETH because every stablecoin transaction requires ETH for transaction fees. 

On the other hand, as stablecoins become increasingly used, there is the potential for stablecoins to lower the monetary premium of ETH. ETH has a credible claim as money within the crypto space, but stablecoins challenge this view. Stablecoins have the potential, due to their lowered volatility, to become the store-of-value, medium of exchange, and unit of account for crypto transactions and smart contracts that need to store value. This is already beginning to happen at the margin, such as MakerDAO adding USDC as an option to serve as collateral in Dai loans. 

This week the narrative is that stablecoin growth is good for ETH, but it will be interesting to see the evolving impact of these competing forces. 

Source: Coin Metrics Reference Rates

CM Bletchley Indexes (CMBI) Insights

Across the board, cryptoasset markets performed strongly through the week as reflected by all the Bletchley multi-asset indexes. The CMBI Ethereum Index was the strongest performer through to April 19, returning 10.4%, whilst the CMBI Bitcoin Index finished the week flat. The small-cap assets yielded the highest returns for the week, with the Bletchley 40 returning 7.1%, followed by the Bletchley 20 (mid-cap assets) which returned 6.2%.

The highest performer of the even indexes was the Bletchley 40 Even, which returned 6.4%, indicating that the strong performance of small cap assets was spread across all constituents within the index and not just a select few.

Source: Coin Metrics CMBI

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics is hiring! 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 [email protected]

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 46 – Investigating Bitcoin’s Changing Correlations

Is Bitcoin an Uncorrelated Financial Asset, A Safe Haven Asset, or Both?

Investigating Bitcoin’s Changing Correlations

By Nate Maddrey and the Coin Metrics Team

The narratives about Bitcoin’s value have shifted over the years, and continue to be debated today. At different points, Bitcoin has been described as electronic cash, censorship resistant digital gold, and an anonymous darknet currency.

Source: Visions of Bitcoin

Among all of the competing narratives there are two that are particularly important when considering Bitcoin as an investment or store of value:

  1. Bitcoin is an uncorrelated financial asset
  2. Bitcoin is a safe haven asset

But data/analysis about these theories is often lacking. In this week’s State of the Network we analyze end-of-day and intraday correlations between Bitcoin and stocks and gold to test these popular narratives.

Methodology

To calculate the correlations in this report, we first took the natural log returns for each asset and then calculated rolling Pearson correlations over specified time periods. We analyzed end of day correlations using New York close pricing (4:00 PM EDT) over a 250 day rolling time window. We also looked at intraday correlations by looking at 5 minute returns over a 60 hour rolling time window.

Since cryptocurrency markets never close but equity markets close on nights and weekends, we cut out Bitcoin results for the days/hours when equity markets were closed and stitched together the results. This helps remove some noise, but has a side effect of sometimes creating exaggerated correlations around the opening/closing of each market day.  

Data sources include Coin Metrics, Finnhub, and FRED. You can check out Bitcoin (and other cryptoasset) correlation charts in our free community charting tool, although the numbers will be slightly different due to the different methodology used for this piece. 

Uncorrelated Financial Asset 

Bitcoin’s correlation with other financial assets is an important consideration when analyzing it as an investment option. If Bitcoin has low correlation with traditional financial assets like stocks, it can effectively be used as a portfolio diversifier. 

The stock market is large and complex, and it can be difficult to analyze Bitcoin’s true correlation with the stock market as a whole. One relatively simple way to examine correlations with stocks is to look at Bitcoin’s correlation with the S&P 500 index (specifically SPY), which can serve as a proxy for the stock market at large. 

Historically, Bitcoin has been relatively uncorrelated with the S&P 500. Since 2012, Bitcoin/S&P 500 correlation has generally stayed between about .15 and -.15, which signals little to no correlation.

But over the last month, Bitcoin/S&P 500 correlation has suddenly increased to new all-time highs. The following chart shows end-of-day correlation calculated on a 250 day rolling basis. 

On a more granular level, intraday data shows that correlation peaked on Black Thursday (i.e. March 12th), when the crypto markets and equity markets both experienced historic, sudden losses (see State of the Network Issue 42 for our analysis on how the crash was mostly driven by short term holders). Correlation then decreased back to relatively normal levels by the end of March. But since then it has started to climb again.

Does this signal that Bitcoin and the S&P 500 are now suddenly correlated?

Probably not. Although short-term correlation shot up, it was under very unique market circumstances. As news about the spreading COVID-19 pandemic began to grow more and more dire on March 12th, investors across the world suddenly began rushing to cash and selling off assets en masse. As a result, correlation shot up between most assets on March 12th. 

For example, the correlation between SPY and GLD suddenly shot up to its highest since 2013. This was likely due to the liquidity crunch caused by the pandemic, which led to sell-offs across the board.

Over the last year, Bitcoin and the S&P 500 have had a correlation close to zero. The following chart shows distribution of intraday correlation (5 min returns, 60 hour rolling correlation) over the last 365 days. It’s relatively centrally distributed around 0, with a mean of -.0075. This shows that under normal market conditions, Bitcoin and the S&P 500 are not significantly correlated.

Bitcoin fundamentals did not change over the last month. However, the outside world changed significantly. Over the long-term, Bitcoin and S&P 500 correlation are likely to revert to the mean and return to levels of near zero (unless there are fundamental changes in Bitcoin and/or the S&P 500). But over the short-term, or at least as long as the liquidity crisis lasts, there may continue to be a relatively high correlation between the two, since global conditions appear to be influencing both markets. 

Safe Haven Asset

Another popular narrative is that Bitcoin is a safe haven asset. Generally speaking, a safe haven is an “investment that is expected to retain or increase in value during times of market turbulence”. Gold is traditionally used as the primary example of a safe haven asset. In uncertain times, the price of gold often increases relative to other asset classes. 

Historically, Bitcoin and gold have not had a very strong correlation. But Bitcoin and gold correlation also suddenly increased in March, similar to Bitcoin and S&P 500 correlation. 

Although gold is typically considered a safe haven asset under normal market conditions, like most other asset classes gold had a relatively large sell-off on March 12th due to the liquidity crisis. Bitcoin and gold correlation increased as prices for both dropped on March 12th (although the price of both dropped, correlation increased positively since both assets were moving in the same direction). But interestingly, they have stayed relatively highly correlated since.

Although Bitcoin and gold may not act as safe havens during a global liquidity crisis, they may act as a safe haven during increases in monetary inflation and quantitative easing. Given the recent $2 trillion+ injection from the Federal Reserve and the unprecedented uncertainty of global health and economic conditions, it is possible that Bitcoin is acting as a safe haven in response to some events but not to others, and potentially even changing day to day. However it is still hard to draw strong conclusions given the huge confounding factor of the global pandemic.

Bitcoin and gold correlation also showed similar signs of growth earlier in 2020. As covered in State of the Network Issue 33, correlation increased in January as Bitcoin and gold prices both increased as US/Iran military tensions escalated. This suggests that Bitcoin may have reacted as a safe haven asset similar to gold, at least temporarily.  

These are small pieces of evidence that the correlation between Bitcoin and gold may be growing. However, Bitcoin’s overall correlation with gold is still relatively weak.

The following chart shows distribution of intraday correlation (5 min returns, 60 hour rolling correlation) between Bitcoin and gold over the last year. It’s also relatively centrally distributed around 0, but has a mean of 0.1194, which is slightly higher than the mean of the Bitcoin/S&P 500 correlation distribution.

Conclusion

Historically, Bitcoin has not been very highly correlated with stocks or gold. Although correlations recently reached all-time highs, it is unlikely that Bitcoin and S&P 500 correlations will remain elevated in the long-term without major changes in the fundamentals of one or both markets. But there is some evidence that correlation between Bitcoin and gold may be starting to increase, at least slightly.

Although the short-term is still uncertain amidst the global pandemic, this could potentially be a long-term inflection point for Bitcoin if federal banks around the world continue to inject money into the global economy at historic rates.  The following chart shows BTC correlation with the 5-Year Forward Inflation Expectation Rate (T5YIFR) from the St. Louis. Fed. Theoretically, if Bitcoin is used as a safe haven in times of monetary inflation, Bitcoin prices should go up as expected inflation increases (and vice versa). Interestingly, Bitcoin/T5YIFR correlation also shot up on March 12th, as the Fed sprung into action in response to the global pandemic

We will continue to follow the situation as it unfolds, and provide context around Bitcoin’s changing narratives. 

Network Data Insights

Summary Metrics

Source: Coin Metrics Network Data Pro

Ethereum (ETH) metrics surged this past week, with active addresses growing 15% week-over-week. ETH realized cap increased 1.6% from the previous week, its first significant increase since the March 12th crash. 

Bitcoin Cash (BCH) also had a big week, with hash rate and difficulty falling due to the scheduled block reward halving. The BCH halving also led to a significant spike in activity, with active addresses and transfers growing 39.3% and 60.6%, respectively.

Network Highlights

Ethereum-issued Tether (USDT-ETH) usage continues to climb. On April 7th, USDT-ETH reached 143.32K daily transactions, the highest daily total since September 2019. The April 7th transaction count also eclipsed the 145.71K transactions on March 13th in the direct aftermath of Black Thursday. Omni Tether (USDT) and Tron Tether (USDT-TRX) both still have significantly less daily transactions than USDT-ETH.

Source: Coin Metrics Network Data Pro

The amount of BTC held by Bitfinex has been dropping since April 2nd. Bitfinex BTC supply had declined by about 24% over the last 30 days, which almost matches BitMEX’s 26% decrease. 

Source: Coin Metrics Network Data Pro

Simultaneously, the amount of ETH held by Bitfinex has increased by 13% over the last 30 days, which is a larger percentage increase than any other exchange in our coverage. 

Source: Coin Metrics Network Data Pro

Market Data Insights

Markets are up solidly for the week with high correlation among cryptoassets but meaningful dispersion in returns. Most large capitalization cryptoassets outperformed Bitcoin this week. If this pattern sustains, it could foreshadow a change in the market regime where smaller cryptoassets experience a higher beta to Bitcoin. Chainlink (+53%) and Tezos (+21%) are outperformers for the week. 

Cryptoasset volatility is off its recent highs as global financial markets stabilize in response to the use of monetary policy tools designed to provide liquidity by virtually every major central bank around the world. 

Source: Coin Metrics Reference Rates

Stablecoins across the board have all seen dramatic increases in their issuance over the past month or so. The drivers of this increase have been a bit of a mystery, with some market participants pointing to an increased need of stablecoins to ride out market volatility, some pointing to continued growth in the lending market for stablecoins, and others pointing to increased retail activity and dip-buying behavior on exchanges. 

Here we examine a broader macroeconomic driver for stablecoin issuance: a global shortage of U.S. dollars. Over the past several decades, the dollars’ status as the global reserve currency has been unrivaled. Meaningful amounts of global commerce is invoiced in dollars, major commodities are priced and transacted in U.S. dollars, and large amounts of U.S. dollar denominated foreign debt has been issued by foreign companies and governments. Under normal circumstances, these foreign entities need a steady supply of dollars to engage in trade and to service their debts. 

Now, with disruptions on both the demand and supply side in global trade, the torrent of dollars from the U.S. to the world has declined to a small stream. The situation is especially acute now because dollars are needed as a store of value to ride out volatility and for margin calls as leveraged positions unwind. Dollar strength, as measured by the trade weighted U.S. dollar index, jumped nearly 8 percent in a few days starting on March 10. 

On the margin, stablecoins should benefit from the global shortage of dollars. USD Coin, the second largest stablecoin, saw its supply increase by over 50 percent in just a few days, also starting on March 10 as the dollar shortage became severe and the dollar strengthened. 

The dollar has since given up some of its gains as the Fed has instituted dollar swap lines with major foreign central banks to make sure foreign entities and institutions have access to the dollars they need. In response, USD Coin issuance is also showing signs of slower growth. 

Source: Coin Metrics Network Data Pro

CM Bletchley Indexes (CMBI) Insights

In designing the CMBI market cap weighted indexes, Coin Metrics have recently released commentary on many of our methodology decisions for determining the ‘free float’ (circulating) supply of cryptoassets. Information can be found in our blog article, Cryptoasset Free Float: An Exploration of Supply Dynamics, released last week.

All CMBI and Bletchley Indexes finished the week positive for a third week running, but still far from recovering from their mid March drop. The CMBI Ethereum Index was the strongest performer, returning 14.5% over the week. Mid cap cryptoassets were the best performing market size group, with the Bletchley 20 returning 12.4% for the week and being the only market weighted index to perform better than Bitcoin. 

Source: Coin Metrics CMBI

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • In designing market-cap-weighted indexes, the primary goal of a benchmark administrator is to provide the most accurate representation of the underlying market by creating a robust methodology and leveraging high-quality data. For an in-depth exploration of our index methodology check out our recent blog article: Cryptoasset Free Float: An Exploration of Supply Dynamics.
  • Coin Metrics is hiring! 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 [email protected]

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 45 – The Signal and the Nonce Redux: From S9s to S17s

The Signal and the Nonce Redux: From S9s to S17s

By Karim Helmy and the Coin Metrics Team

As Bitcoin enters its next halving, the network is experiencing several simultaneous transitions. In addition to the severe change in mining economics being caused by the reward adjustment, Bitmain’s Antminer S17 is in the process of replacing the longstanding S9 line of miners as the dominant mining hardware on the network. The first rig in the S9 line, the eponymous Antminer S9, was released by Bitmain in 2016 and quickly became the most popular SHA-256 miner; as a result, the network has not experienced such a shift in years. 

Due to a lack of publicly available data about the types of mining hardware used by individual miners, it can be difficult to measure the rate at which this transition is occurring. One unexpected source does shed some light on mining hardware trends, though: the network’s nonce distribution. The arrangement of these arbitrary numbers, included by miners as part of each block’s hash, hints at how mining hardware usage has shifted over the years.

In a previous issue of state of the network, we investigated how nonce distributions patterns can be used to spot the rise of ASICs. In this issue, we’ll dig further into the peculiarities of Bitcoin’s nonce distribution and the sources behind the distribution’s striations to investigate the recent shifts in mining hardware. We’ll then break down this data by mining pool, which allows for more granular insight into the hardware used by specific pools.

Golden Marbles – Understanding The Mining Process 

Mining is a critical part of Bitcoin’s security model and arguably the most important improvement on previous attempts at creating digital money. Although the implications of mining are quite complex, the concept behind it is relatively simple to understand.

From the perspective of a miner, mining a block resembles repeatedly selecting marbles from a bag without replacement. In this analogy, the number of marbles is very large, with a large proportion of blue marbles and a small proportion of gold marbles. Miners receive a payout on pulling a gold marble from the bag.

Explained in more technical terms, Bitcoin miners compete to find a golden nonce that, when appended to the header of a proposed block, hashes to below a certain value determined by the network’s difficulty parameter. Miners search for this nonce, or arbitrary number, that can only be used once, by guessing values and checking if the resulting hash is below some threshold. The first miner to find such a value for a valid block and broadcast it to the network receives the right to choose and order the transactions within the block, a necessary step for these transactions to be eventually considered valid. 

In return, the miner also receives a payout from the block reward and fees from any transactions included in the block, both of which are received in a special coinbase transaction. Provided that the purported properties of the SHA-256 hash function hold, the distribution of golden nonces for any given block is random and a golden nonce cannot be found except through brute-force computation.

Because a reference to the coinbase transaction is included in the block header, each mining entity is sampling from a different distribution. In other words, each entity is pulling marbles from a different bag, where bags contain the same number of marbles and, in expectation, the same ratio of blue and gold marbles.

The proportion of golden marbles is determined by the network’s difficulty parameter, and is fixed for the relevant period. The difficulty parameter is automatically adjusted by the network. Today, due to a high block difficulty and random variance, there are often no golden nonces for a specific block header. In other words, there are some bags that contain no golden marbles. 

Miners who exhaust the nonce space of a proposed block typically increment the block’s timestamp to generate a new set of nonces. That is to say, when a miner runs out of marbles, they grab a new, full bag. If the timestamp has reached the point where further adjustment would render it invalid, the miner must adjust the set of transactions included in the block. Analogously, if a miner runs out of bags in their room, they need to grab some more from another room, which is time-intensive.

To increase their probability of finding a golden nonce in a fixed period of time, miners can parallelize their computations, which is analogous to pulling marbles by the handful rather than one at a time. Nonce-finding can be parallelized by using hardware suited to the task, in particular graphical processing units (GPUs) and specialized chips known as ASICs. ASICs are much more efficient at parallelization than any alternative, and today account for virtually all of the network’s computational power, or hashpower.

In another form of parallelization, several miners coordinate their nonce-finding and agree to split any payouts. This strategy reduces both the size and variance of a miner’s payouts, and does not change expected long-run revenue. Groups of miners acting in this way are known as pools. The operator of a pool typically charges a fee, which individual miners accept in exchange for a reduction in income volatility.

Bitcoin’s Nonce Distribution

Every two weeks, Bitcoin’s difficulty parameter is adjusted so that a new block would be produced every ten minutes on average if the amount of computation performed on the network were to remain constant. This feature ensures the network will continue to operate in spite of potentially large changes in hashpower. In a sufficiently competitive mining market dominated by miners who are computing values in parallel, then, we would expect the plot of golden nonces over time to look like evenly distributed static. Surprisingly, it doesn’t.

The non-random distribution near the left-hand side of the plot can be attributed to mining by iteratively testing values starting at 0. If a miner is mining without parallelization on a CPU and as an individual, and therefore has no possibility of running collisions with other members of their pool, this strategy is as valid as any other, since the nonce distribution for each new block is independent. The disappearance of this pattern coincides with the introduction of GPU miners, which parallelize computation.

Near the right-hand side of the plot, there is a striated pattern of regions with very few nonces. To our knowledge, this anomaly was first noticed by Twitter user @100TrillionUSD in January of 2019. The same plot, with the striated region labeled, is shown below.

The bizarre pattern was the subject of a BitMEX research piece shortly after, which speculated that the anomaly was the result of a quirk in an implementation of AsicBoost, a controversial mining optimization technique.

There are two variants of AsicBoost: covert AsicBoost, which cannot be definitively observed on-chain, and overt AsicBoost, which can be. The BitMEX research team discusses both variants, but is particularly interested in the effect of covert AsicBoost, the use of which was made practically impossible for non-empty blocks with the activation of SegWit in August of 2017. The researchers could not confirm their hypothesis. 

The streaked pattern indicates that miners are systematically undersampling certain ranges of possible nonces. By excluding certain ranges from sampling, miners are effectively partitioning their marbles into a small number of different bags and refusing to pull from certain bags. In expectation, the color ratio of each bag is equal, so miners do not change the probability of selecting a golden marble on the first try by doing this. Because the number of marbles in each bag is very large, the reduction in effectiveness caused by adopting this strategy when sampling repeatedly is small. This strategy does, however, increase the frequency with which miners must grab a new bag of marbles, which can be expensive. Since each mining entity is sampling from their own distribution, other pools cannot use knowledge of an entity’s adoption of this technique to their advantage by strategically sampling.

In October of 2019, State of the Network Issue 23 looked at Bitcoin’s nonce distribution in depth and noted the streaked pattern. Since then, the striated pattern has faded, and the nonces of recently-mined blocks appear to be more randomly distributed.

The anomalies in the nonce distribution do not appear to be directly related to AsicBoost. Covert AsicBoost became unusable in 2017, and the first firmware update enabling overt AsicBoost was publicly released in October of 2018, but the striations are clearly visible between these two dates. Additionally, while overt AsicBoost usage remains high, the patterns are no longer visible in blocks mined either with or without overt AsicBoost.

Instead, the patterns in the nonce distribution may be caused by the manner in which Bitmain’s Antminer S7 and S9 families of miners sample nonces. This artifact is likely an unintended side-effect of optimization, and is ultimately harmless to both the miner and the network.

The S7 and S9 lines contain several related models using the BM1385 and BM1387 chips, respectively. The period in which each line of miners was dominant on the network corresponds to a distinct phase in the patterning of Bitcoin’s nonce distribution.

When observing all nonce values on the network, the streaked pattern first becomes clear in late 2015, coinciding with the release of the S7 in late August of that year and the fulfillment of orders in late September.

The Antminer S9 was released in late May of 2016, with the purchasers of the first batch receiving their orders in mid-June of that year. Shortly afterward, the streaks become more narrow in concurrence with supersession of the S7 by the S9 as the dominant miner on the network in late 2016.

The pattern’s recent breakdown coincides with the transition from the S9 to the Antminer S17 as the dominant miner on the network. While the S17 was released in April of 2019, the use of S9s on the network has until recently remained common as they have continued to be economical to operate.

Pooled Analysis

Stratifying the dataset by each block’s miner allows for a more fine-grained view of the nonce distribution. Identifying the miner of a block is relatively straightforward but carries several caveats.

The miner of a block is typically identified through a tag left in the block’s coinbase data field. These identifiers are voluntary and falsifiable: miners are not required to leave a message, and may choose to leave another pool’s tag in place of their own. In certain situations, these misleading behaviors may even be incentivized, so the shortcomings of this approach should be recognized. This technique is the industry standard, however, and while many miners choose not to leave an identifier, we have no reason to believe that falsification is occurring on a significant scale.

The miner of a block is labeled according to their most recent identifiable block mined. This provides robustness against anomalies like hashpower voting, in which the coinbase data is used to signal support for a fork rather than to identify the miner.

Mining entities are also identified through reuse of a known payout address. This approach requires pools to reuse addresses, and is sensitive to the initial seed set of known addresses. For our purposes, this approach is used to supplement tagging based on the coinbase data in order to provide coverage for miners who do not leave a consistent tag.

Once we have categorized blocks by miner, we can incorporate this information into our plot of Bitcoin’s nonce distribution.

We can also take a look at the nonce distribution of individual pools. Even at this level, the anomalous patterns remain visible. Consider the plot below, which shows the blocks mined by Antpool and BTC.com, both of which are owned by BitMain, as well as ViaBTC, in which BitMain is an investor.

The streaked patterns are significantly clearer in the nonce distributions of Bitmain-affiliated pools than in that of the network as a whole. This indicates a higher proportion of S7s and S9s in these pools during the relevant period, which is to be expected given the pools’ association with the manufacturer of these miners.

The proportion of blocks mined by unknown entities shows a large drop-off in 2015. This is a result of the block size wars, during which many previously anonymous miners began to identify themselves on-chain in order to signal support or opposition to a block size increase. Today, the vast majority of miners by hashpower are identifiable. The striated patterns are faintly visible in the nonces of blocks mined by unknown miners, as is their gradual dissolution.

Conclusion

The Antminer S9 has until recently been the most-used miner on the Bitcoin network since its release in 2016. Despite the release of the S17 last year, the S9 remained economical to operate for a period, but the miner is being phased out in light of increasing hash rate and changing market conditions. The shift in dominance from the S9 to the S17 that is currently taking place in anticipation of the halving has not been properly considered in many analysts’ assessments of the network.

In parallel with miners’ transition from predominantly using the S9 to the S17, the streaked patterns that were formerly the defining feature of Bitcoin’s nonce distribution have dissolved. The source of these mysterious streaks, which appear in a space that should look random, has been the subject of significant speculation. The timing of the streaks’ visibility lends credence to the theory that these lines were an artifact of the hardware used to mine on the network, in particular the S9 and the S7 that preceded it.

Nonce data allows us to gauge the scale and pace of this shift, using only public information, in a manner that would otherwise be impossible. By taking advantage of the artifacts left by the S9 in the sampling of nonces, we may be able to estimate the proportion of these miners on the network. The segregation of this data by pool gives unique information on the efficiency of miners’ operations, to be covered in a future report. This issue paves the way for a formal assessment of this type.

Network Data Insights

Summary Metrics

Source: Coin Metrics Network Data Pro

Bitcoin (BTC) mining showed healthy signs of recovery this past week with estimated hash rate increasing by 7.3%. Inefficient miners have likely already started to capitulate and are being replaced by more efficient miners, which is positive for the long-term health of the network. For more on miner economics and the implications of the recent difficulty decrease, see State of The Network Issue 44 – Understanding Miner Economics From First Principles

BTC active addresses also showed positive signs of recovery this week, growing by 6.3%. But Ethereum (ETH) active addresses went in the opposite direction, dropping by 13.4% week-over-week. ETH daily active addresses peaked at 537K on March 21st, the highest daily total since May, 2018. However, since then they have been declining – ETH had 310K daily active addresses on April 5th.

Network Highlights

The number of addresses holding relatively small amounts of BTC has been increasing since the March 12th crash. 

The number of addresses holding between one billionth (1/1B) and one hundred millionth (1/100M) of the total BTC supply (i.e. between 0.000000001% and 0.00000001% of total supply) has increased about 6% over the last 90 days. Similarly, the number of addresses holding between one hundred millionth (1/100M) and one ten millionth (1/10M) of total supply increased about 4%. 

Both had a noticeable increase in growth rate beginning around March 12th. This could signal that adoption is growing, as new users start acquiring relatively small amounts of BTC.  

Source: Coin Metrics Network Data Pro

The amount of ETH held by the exchanges in our coverage (listed below) has grown over the last 30 days, while the amount of BTC held by exchanges has decreased. 

ETH has increased by about 5%, while the amount of BTC held on exchanges has decreased by about 3%. The drop in BTC is largely due to the rapid decrease in supply held by BitMEX, as covered in the Network Highlights section of State of the Network Issue 44.

The below chart shows the 30 day growth for the total supply held on the following exchanges: Bitfinex, Binance, Bitstamp, Bittrex, Gemini, Huobi, Kraken, BitMEX, and Poloniex.

Source: Coin Metrics Network Data Pro

Bitfinex has had the largest increase in ETH supplies out of all of the exchanges in our coverage. The amount of ETH held by Bitfinex has increased by about 17% over the last 30 days, while no other exchange has had more than a 10% increase. 

Source: Coin Metrics Network Data Pro

Market Data Insights

Markets are up sharply over the past week and are beginning to recover losses sustained on March 12. Our previous research finds that this sell-off was driven by short-term holders and reinforced by the BitMEX liquidation spiral. A broad consensus is forming that this sell-off, mirroring the sell-off seen in traditional markets, was at least in part due to technical dislocations in the market. Sentiment among retail investors appears to be unaffected as evidenced by increased user activity from Coinbase

Source: Coin Metrics Reference Rates

The “risk-on asset” versus “safe haven asset” debate continues over Bitcoin. Some market participants have observed that Bitcoin’s safe haven narrative has been damaged over the past month which negatively impacts institutional investors’ appetite for entering the space. 

A closer examination of Bitcoin and gold provides some evidence that the narrative is intact and could be stronger than ever. While Bitcoin did sell off aggressively in concert with equity markets, gold did too due to forced liquidations that happened in nearly every financial market. Since then, both Bitcoin and gold have recovered some of their losses. 

Source: Coin Metrics Reference Rates

Bitcoin’s correlation with gold, measured over the past 30 days, is now at all time highs. Unprecedented monetary policy and fiscal stimulus from nearly every country in the world is forming the base for a credible narrative of increased risk of severe financial imbalances and the potential for long-term increases in inflation. Ultimately, these fears never materialized during the 2008 financial crisis, but conditions are ripe for these fears to resurface again. 

Source: Coin Metrics Reference Rates

CM Bletchley Indexes (CMBI) Insights

CMBI and Bletchley Indexes all had a great week with ~15% increases across the board. The Bletchley 40, small-cap assets, performed the best of the market cap weighted indexes through the week, returning 16.9%. Interestingly the lower weighted assets within the Bletchley 40 were the top performers, demonstrated by the large returns of the Bletchley 40 Even and the Bletchley Total Even Indexes. For reference, the even indexes weigh all constituents equally, giving more weight to the smallest cap assets within the index.

The relatively low returns of all CMBI and market cap weighted indexes against BTC further demonstrates the uniformity of the week’s performance across the asset class. 

Despite the strong performance of cryptoassets this week, monthly returns were still down significantly through March, with all CMBI single asset and Bletchley market cap weighted indexes down over 20%. The CMBI Ethereum index was the worst performer through March, falling 40.5% from $224.25 to close the month at $133.39. 

Source: Coin Metrics CMBI

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • Coin Metrics is hiring! 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 [email protected]

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 44 – Understanding Miner Economics From First Principles

Understanding Miner Economics From First Principles 

By Kevin Lu and the Coin Metrics Team |Tuesday, March 31st, 2020

The next halvening is approaching, yet debate about its impact on asset prices remains mired in controversy. 

Two camps have formed. One believes the halvening is already priced in by market participants, citing the efficient market hypothesis. The other camp believes it isn’t yet fully priced in and expects the halvening to plant the seed for further increases in prices due to the increase in perceived scarcity and the change in supply-side dynamics.  

Source: Coin Metrics Network Data Pro

We believe the intensity of the debate partially stems from the limited empirical record. Bitcoin has only experienced two halvings in its history and only a handful of other proof-of-work coins have gone through the same events. Discussions are stalled because of the lack of shared terminology, small sample size, and inaccessible data regarding critical questions. 

In this week’s State of the Network, we present a framework for understanding miner economics and how to best navigate the upcoming block reward halving. The intention is to present a framework reasoned from first principles rather than relying purely on empirical data. We also apply this framework in examining the upcoming halving for Bitcoin, Bitcoin Cash, and Bitcoin SV. 

Three Miner Axioms and Inferences 

We present a set of three miner axioms which we believe to be largely true as a starting point for further reasoning. We say “largely true” because there are always edge cases — they may not hold for certain assets or for certain miners. But for the most part, these axioms hold water. 

We also present three miner inferences that are built upon our three axioms. These inferences form the foundation of our framework for understanding miner economics. 

Axiom 1 

Miners operate as profit-maximizing commercial enterprises with large economies of scale

Mining has gotten so difficult and resource-intensive that is largely uneconomical for an individual or hobbyist to participate. Instead, there are large economies-of-scale to mining. Large miners locate themselves in areas of the world where electricity is cheap. They are able to negotiate lower rates with electricity utility companies, purchase large quantities of the most efficient mining equipment, and rent large facilities to operate the equipment. 

The ability to mine at scale lowers the cost of mining a single coin. And since mining is a competition, miners organize as a profit-maximizing commercial enterprise. Miners do not operate for ideological or altruistic purposes and cannot continue operating in the long-term if they are not profitable. 

Axiom 2

Mining is a competition with a fixed total reward that is split among all participants with a regular cadence

Issuance is mandated by the protocol and controlled via a difficulty adjustment. In the case of Bitcoin, the protocol generates a block reward of 12.5 coins per block (currently) and regularly adjusts mining difficulty such that blocks are generated on average every 10 minutes. All miners compete for this reward, along with transaction fees. The total block reward revenue for all miners over a given period of time is predetermined and cannot be changed except for marginal amounts and for brief periods of time between difficulty adjustments. 

Axiom 3

Miner revenue is denominated in crypto while miner costs are denominated in fiat 

Miner revenue consists of the block reward and transaction fees, both of which are denominated in crypto. 

Mining costs include mining hardware, electricity to operate the miners, cooling fees, facility rental fees, server maintenance, internet connectivity, salaries, insurance, legal services, taxes, and so on. These costs are denominated in fiat since most traditional companies do not currently accept crypto as payment (electricity utility companies will not take Bitcoin as payment, for example). Even if certain expenses end up being paid using crypto, such as instances where miners pay for equipment or employee salaries using crypto, the price of the good or service is still quoted in fiat currency. 

Inference 1

Mining is a (Nearly) Perfectly Competitive Industry 

Based on the first two axioms above, the first inference we introduce is that the mining industry operates under an equilibrium of nearly perfect competition where the market price faced by each miner equals the miner’s marginal cost. This is achieved by two mechanisms. 

First, profit-maximizing miners enter the industry or invest in more equipment when mining is profitable and exit the industry or turn off miners when it is unprofitable. Second, the change in hash rate triggers a difficulty adjustment which constantly seeks to bring the cost of mining a single coin equal to the current market price. 

Mining is zero-sum (in the long-term) in that each miner is in competition with other miners over the same block reward. This also means that miners operate in an equilibrium state of zero economic surplus — that is, in the long-run, miners earn only normal profits and are compensated only for the opportunity cost of their time plus an allowance for risk. Due to the competitiveness of the miner economy, it seeks a long-term equilibrium where miners profit margins are small and close to zero. 

Miner profit margins, however, can meaningfully fluctuate around this equilibrium due to delays inherent in the system which has important implications on miner-led selling pressure. We discuss this more in the following two inferences. 

Furthermore, the industries that reside upstream to mining such as miner hardware and semiconductor manufacturers show elements of oligopoly market structure. Based on this supply chain, certain miners (such as the Bitmain affiliated mining pools) can leverage information advantages or access mining hardware earlier than their competitors which reduces the degree of perfect competition in the mining industry.

Inference 2

Miners Are a Continuous and Significant Source of Selling Pressure

Combined with the third axiom, we present an important inference: miners represent the single largest cohort of natural, consistent sellers. Their selling pressure is significant because miners must sell the crypto that they earn to cover their fiat-denominated costs. And since their profit margins tend to gravitate towards zero, miners must sell nearly all of the crypto that they earn. 

Here we use Bitcoin to contextualize the magnitude of miner-led selling pressure. Miner revenue in 2019 was nearly $5.5 billion dollars. Some researchers compare this number to the annual trading volume of Bitcoin, which is several magnitudes bigger, and conclude that miner-led selling pressure has a negligible impact on the market. However, selling from miners represents net capital outflows from the space and the fiat obtained by miners is unlikely to ever return to the market, which is not necessarily the case for other trading volume. Therefore, miner selling has an outsized influence on the rest of the market

Put differently, the best estimates indicate Coinbase has roughly 1 million Bitcoin in customer deposits. At current prices, this is equivalent to $6.8 billion dollars, an amount similar to annual miner revenue in 2019. Under the assumption that miners sell the majority of the crypto they mine, miner-led selling pressure is nearly equivalent to all customers on Coinbase liquidating their Bitcoin over the course of a year and permanently exiting the market. 

Source: Coin Metrics Network Data Pro

We extrapolate miner revenue for the entire year of 2020, assuming prices remain at current levels and accounting for the halving of the block reward. Under these assumptions, we should only see half-of-a-Coinbase worth of selling pressure this year — a significant reduction. 

Inference 3

Miners Have a Pro-Cyclical Impact on Asset Prices 

While the mining industry is constantly seeking a long-term equilibrium where miner profit margins are small but close to zero, the reality is that profit margins experience large fluctuations around this steady state. 

Elements influencing the cost side of the equation are slow moving and react with a meaningful lag. Decisions regarding entering or exiting the industry, purchasing additional equipment, and scaling up operations all take time. And difficulty adjustments by their nature have an approximately two week lag. 

On the opposite side, revenue is fast moving because one of the main determinants is the price of the coin which is subject to extreme levels of volatility. Bitcoin regularly experiences annualized volatility of over 50 percent. 

Source: Coin Metrics Reference Rates

Varying profit margins due to these factors mean that the amount of selling pressure by miners to cover their fixed, fiat-denominated costs varies as well. When prices are particularly volatile or trend in one direction over a sustained period of time, miner profit margins can be consistently positive or negative for meaningful amounts of time. These deviations are more likely to occur when prices are rising, since delays regarding more capital investment in mining hardware are more prominent compared to delays when prices are declining. The decision to shut off miners when prices fall below electricity costs can be made quickly. 

Inventory management of miners is not a well-studied topic since access to this information is not available, but it stands to reason that each miner makes their own decision on how much of their block rewards to sell for fiat and when to sell it. Since miner variable costs are slow moving and fairly constant in fiat terms, miners are required to sell less of their block rewards to cover their expenses during periods of rising crypto prices. 

On the other hand, when crypto prices are falling, they are required to sell more. Under this theory, miners have a pro-cyclical effect on the market, in that they further exacerbate price increases. There are limitations to this dynamic, however. Sustained increases in prices can compel miners to sell more of their block rewards to fund additional capital investment in new mining hardware suggesting a counter-cyclical impact on prices under certain market conditions. 

During periods of capitulation where profit margins for many miners are negative, miner-led selling flow is likely to be high. Miners may attempt to endure periods of short-term pain, and perhaps may temporarily operate at a loss until less cost-efficient miners exit the industry. Miners may be willing to sell block rewards earned in prior periods that they kept on their balance sheet to be used in an attempt to outlast other miners. 

All of these behaviors reinforce the direction in which crypto prices are moving and are a key determinant in why crypto prices regularly experience bubbles and crashes. 

While we believe this framework characterizes the pro-cyclical behavior of most miners, the rise of a robust lending market has the potential to change this dynamic. This allows miners to speculate on the future price of Bitcoin and engage in market timing where they pay for their fiat expenses using borrowed funds that are collateralized with the Bitcoin on their balance sheet. Miners that engage in this behavior believe that Bitcoin’s price will rise in the future and delay selling. The rise of a derivatives market which allows miners to hedge against future price movements can play a similar role. 

While the overall impact on the amount of miner-led selling flow depends on how accurate miners are in their market timing efforts, we believe that miners will tend to borrow fiat under certain market conditions. Assuming that miners intrinsically have a long-bias for Bitcoin, miners will tend to borrow fiat when they believe prices are well below the long-term fundamental value and when they believe we are firmly in a bull market. This should moderate the pro-cyclical impact when prices are declining but accentuate the impact when prices are increasing. 

The Upcoming Halving 

Bitcoin will soon experience its third halvening where the block reward will be reduced from 12.5 coins per block to 6.25 coins per block, equivalent to annualized issuance being reduced from 3.6 percent to 1.8 percent. This is anticipated to occur in approximately 45 days or May 14, 2020. 

Source: Coin Metrics Network Data Pro

Prices have sharply declined over the past several weeks in concert with risk assets in traditional markets. The pro-cyclical behavior of miners implies that miner-led selling pressure should be increasing as well. Prices have almost certainly declined below the breakeven price for the set of miners who are least efficient and have the lowest profit margins. These miners have likely either temporarily or permanently shut off their machines. This can be seen in the most recent difficulty adjustment where difficulty declined by 16 percent, one of the largest drops in history.

Source: Coin Metrics Network Data Pro

Such a large difficulty adjustment indicates that inefficient miners are already reaching a point of capitulation where they are forced to sell all the coins they mine to cover their costs. 

Miner-led selling pressure for Bitcoin is likely to continue to increase because both Bitcoin Cash and Bitcoin SV will be experiencing their halving on April 8 and April 9, respectively. All three assets share the same SHA-256 mining algorithm and miners can seamlessly redirect their hash power to the asset that provides the highest return on investment. 

When Bitcoin Cash and Bitcoin SV halve their block rewards, this should force miners to direct even more hash power to Bitcoin as it will still have a 12.5 native unit block reward (instead of 6.25) for about a month longer. Therefore, we should expect difficulty increases for Bitcoin that should further squeeze profit margins for all miners. 

Source: Coin Metrics Network Data Pro

It is concerning that miners are in a state of capitulation even before the halving. Once the block reward halves, miner revenue will be cut in half while miner costs will remain constant, so we expect even more miners to capitulate in the months ahead. 

Miner capitulation increases selling pressure until inefficient miners are forced off the network, but in the long run these events are supportive for prices. Culling inefficient miners allows only the most efficient miners with the lowest cost of production to remain. Once inefficient miners exit the network, profit margins will improve for the remaining miners, which reduces selling pressure, increases prices, and should repeat in a virtuous cycle. Eventually, if prices bottom and recover, the pro-cyclical behavior of remaining miners should support further price increases. 

Conclusion 

Using a set of axioms, we provide a framework reasoned from first principles that illustrates how miners are a continuous and significant source of selling pressure that has a pro-cyclical impact on prices. Miner-led selling pressure for Bitcoin, Bitcoin Cash, and Bitcoin SV is currently high and is likely to increase further in the coming months as all three coins undergo their halvings.

We expect miners to follow a cycle of decreased profit margins, increased selling, capitulation, and a culling of the least efficient miners from the network. Once this cycle is complete, the miner industry should return to a healthier state that is supportive of future price increases. 

Network Data Insights

Summary Metrics

Source: Coin Metrics Network Data Pro

The major cryptoassets remained relatively stable this past week. Although market cap increased, realized cap stayed stable or dipped for all five assets in our sample. The ratio of realized cap to market cap (MVRV) can be used to help gauge market tops and bottoms, and also gives insight into the behavior or holders vs speculators, as explored in State of the Network Issue 41.

Bitcoin’s estimated hash rate rose week-over-week after plummeting in the aftermath of Black Thursday, which led to Bitcoin’s second largest difficulty percentage drop in history. Although fees have been down this past week, mining revenue increased due to Bitcoin’s price gains. 

Network Highlights

Binance USD (BUSD) market cap has almost tripled since March 1st, increasing from about $68M to over $181M. Huobi USD (HUSD) market cap has almost doubled, growing from $78M to $136M.

The following chart was created using our free community charting tool. We recently added BUSD and HUSD to fill out our stablecoin data set.  

Source: Coin Metrics Community Charts

Tether also continues its rapid growth. There has been $1.4B worth of Tether issued on Ethereum (USDT-ETH) since March 1st. Tron-issued Tether’s (USDT-TRX) market cap has increased by about $165M since March 1st, while Omni-issued Tether (USDT) has stayed relatively stable. Tether’s total market cap is now over $6.4B

Source: Coin Metrics Network Data Pro

The amount of Bitcoin held by most exchanges has decreased over the last 30 days. Out of the exchanges in our coverage, only Kraken and Bitfinex have increased their BTC supply over the last month (1% growth each). 

Source: Coin Metrics Network Data Pro

The amount of Bitcoin held by BitMEX has been in freefall over the past two weeks after experiencing mass liquidations on March 13th (see State of the Network Issue 43 for our coverage on the BitMEX liquidation spiral). As of March 29th, BitMEX held 244k BTC, down from a peak of 315k on March 13th. 

Source: Coin Metrics Network Data Pro

Market Data Insights

Most cryptoassets remained largely unchanged over the past week but with significant volatility. Bitcoin and other coins continue to show intermittent periods of high correlation with risk assets, particularly during pre-market trading. Volatility remains at high levels with Bitcoin’s rolling one month volatility at 148%, nearly the same level as during the late 2017 bubble. 

Source: Coin Metrics Reference Rates

Stablecoins are seeing large increases in issuance but their pegs to the U.S. dollar remain relatively stable outside of an unstable period on March 12. The reason for the increase in issuance across the board is still unknown, but stablecoins do not appear to be selling at a meaningful premium. While most stablecoins remain close to the $1 U.S. dollar peg (with the exception of Dai), there appears to be a small but marked regime shift between and after March 12. Stablecoin prices and bid-ask spreads appear to have entered a more volatile regime. 

Source: Coin Metrics Reference Rates

CM Bletchley Indexes (CMBI) Insights

Cryptoasset markets had a turbulent week with most CMBI and Bletchley Indexes experiencing intraweek highs of over 10% above last week’s close. However after a strong start to the week, all indexes fell significantly through the weekend, but still managed to finish up between 0-5%. 

The Bletchley 40 (small-cap) finished the week strongest, up close to to 5% against the U.S. dollar and 3.5% against Bitcoin. All Bletchley multi-asset indexes outperformed the CMBI Bitcoin and CMBI Ethereum single-asset indexes this week. 

Coin Metrics Updates

This week’s updates from the Coin Metrics team:

  • We are happy to announce that we completed a $6 million round of funding led by Highland Capital Partners with participation from FMR (Fidelity Management & Research), LLC, Castle Island Ventures, Communitas Capital, Collaborative Fund, Avon Ventures, Raptor Group, Coinbase Ventures, and Digital Currency Group.
  • Coin Metrics is hiring! 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 [email protected]

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.

CRYPTOASSET FREE FLOAT: AN EXPLORATION OF SUPPLY DYNAMICS

© Coin Metrics

In designing market cap weighted indexes, the primary goal of a benchmark administrator is to provide the most accurate representation of the underlying market through creating a robust methodology and leveraging high quality data. The two primary inputs into any market cap weighted index calculation are:

  1. Price data: Reliable and relevant pricing sources coupled with a robust pricing methodology is essential in order to represent the fair market value of the index.
  2. Supply data: Data that is reflective of market trading opportunities in order for index weights to accurately represent the supply / demand function of the market.

With the increase in quality global trading venues, trade volumes, order book depth and semi-regulated USD on-ramps, cryptoasset price discovery and aggregated price feeds are significantly better understood than they have been previously. The presence of institutional quality data feeds such as CM Market Data Feed and cross venue price aggregation methodologies such as CM Reference Rates demonstrate confidence that index price inputs are well on their way to meeting traditional capital market standards.
Supply, however, still remains a hotly debated topic within the cryptoasset ecosystem. While decentralized cryptoassets are, by definition, open networks with readily available on-chain data, reporting of token holdings is far from transparent (often very intentionally). Further, in these early and nascent stages of the ecosystem, multiple token economic models have been developed (fair launch, ICOs, IEOs, premines, hard forks, etc.), adding complexity to determination of assets that are considered ‘available’ for trade.
One of CMBI’s core principles is to design indexes that accurately reflect their underlying market. In the absence of transparent, independent and reliable supply metrics, Coin Metrics has designed the CM Free Float Supply to set a standard for the objective determination of a cryptoasset’s supply that is ‘available’ to the market. This newly designed metric will act as the basis for all CMBI Market Cap Weighted Indexes. Our supply methodology takes into consideration many of the best practices from equity markets and applies them to cryptoassets.

To learn more about Coin Metrics’ Free Float Supply methodology, read our latest piece:

Cryptoasset Free Float: An Exploration of Supply Dynamics