What’s In Your Wallet: Cryptocurrency Coin Correlations

This post originally appeared on Cointracker’s blog and was written by Chandan Lodha.

At CoinTracker, we are building a portfolio & tax manager for cryptocurrency. CoinTracker syncs crypto transactions and balances from the top exchanges & wallets to deliver a unified dashboard of your cryptocurrency portfolio performance.

CoinTracker also helps users calculate their cryptocurrency taxes in the United States, Australia, Canada, and the United Kingdom. Since launching, we have grown to help thousands of users file their cryptocurrency taxes, with over 50,000 connected exchange accounts on CoinTracker, tracking over $300,000,000 of crypto-assets.

Individual user data is completely private on CoinTracker (privacy policy,security policy). We have however been able to aggregate anonymized data from our platform to deduce some interesting patterns across cryptocurrency users. For example, we were wondering what is the probability that a user holds coin X if that user also holds coin Y? Essentially, what is the correlation between coins in a given user’s portfolio.

We pulled the anonymized dataset, worked with the world-renowned firmStamen to visualize it, and now we would like to share it with the community.

Here is a visualization of the coin correlations between each of the top 100 cryptocurrency coins & tokens by the current market capitalization, segmented bi-annually for the past five years. Each dot represents the magnitude of the correlation (hover over the dot to see the approximate percentage of correlation).

Some caveats:

• The data set is across ~40,000 users
• The data is biased to users who track their cryptocurrency on CoinTracker
• If no users held an asset in a particular time period, that entire column/row is discarded
• In order to be considered an asset holder, a user must have held at least 5 USD worth of the asset on the date of the snapshot
• There are some user data errors (e.g. coins entered before they were created)

You’ll notice that the diagonal is all blank because it is redundant to indicate that 100% of bitcoin holders also hold bitcoin. You’ll see other gaps where (close to) zero users of one coin hold another.

Here are some patterns we have noticed:

1. Correlation by use case. Across many protocols competing for the same use case (e.g. money: BTC vs. BCH, Dapps: ETH vs. EOS, privacy: XMR vs. ZEC, etc.), often users choose to bet on multiple projects in the same space. If a user bets on the smaller project (by market cap), then they are much more likely to also have bet on the larger protocol as well. This intuitively makes sense as larger market cap projects typically have more visibility, lower perceived risk, and therefore more holders. For example:

25% of bitcoin (BTC) holders hold bitcoin cash (BCH):

86% of bitcoin cash (BCH) holders hold bitcoin (BTC):

29% of Monero (XMR) holders hold Zcash (ZEC):

55% of Zcash (ZEC) holders hold Monero (XMR):

17% of ether (ETH) holders hold EOS:

74% of EOS holders hold ether (ETH):

Also the top 10 coins most correlated coins with BTC are

1. ETH (68%)
2. LTC (43%)
3. XRP (33%)
4. XLM (26%)
5. BCH (25%)
6. ADA (22%)
7. NEO (21%)
8. BNB (20%)
9. TRX (19%)
10. IOTA (17%)

This is similar to, but not exactly overlapping with the top 10 coins by market cap (after BTC): ETH, XRP, BCH, EOS, XLM, LTC, USDT, ADA, XMR, IOTA

2. Interesting tax strategies. When we first looked at the data, we were confused because 100% of users holding bitcoin cash (BCH) in January 2014 were holding bitcoin (BTC). But the BCH fork did not occur until August 2017, so the BCH column should have been blank here — did we have a data error? Possibly. However our colleagues at Stamen pointed out another theory — some users were tracking BCH back to their original acquisition of bitcoin and treating bitcoin cash as the coin they were holding the entire time, and later treating the fork as a bitcoin fork (rather than a bitcoin cash fork). We saw similar behavior with other forks as well.

3. The market changes quickly. Most of the top 100 coins were born in the last 12–24 months so it’s hard to predict what will be around in the future. Bitcoin, however, has proved resilient thus far.

As we continue serving more users around the world, we look forward to uncovering more interesting insights about how cryptocurrencies are being used around the world.

Feel free to explore the data set (released under a CC-BY license). If you find any interesting insights in the data, tweet @cointracker_io and let us know!