2019 Q2 Crypto-Correlations Review

In the financial industry, correlations have been widely used by analysts and fund managers to diversify and efficiently allocate their assets across sectors and industries. Despite the nascency of the cryptoasset industry, these metrics have gradually become more scrutinized and analyzed by all crypto-stakeholders, institutional and retail.

This report will analyze correlations across all largest cryptoassets over the first six months of 2019.

1. 2019 Q2 overview market performance

Despite a pullback in the final days of the second quarter, the overall performance of the cryptoasset market was extremely positive in the first six months of 2019.

The total value of the cryptoasset industry rose by +139% in the past quarter, marking the third best quarter the industry has seen since 2014. In absolute dollar terms, the total USD value of cryptoassets increased by roughly $199 billion in the second quarter of 2019, the second-highest quarterly growth on record, ranked only behind the bull run of 2017 Q4.

Chart 1 - 2019 Quarterly cryptoasset market cap change (%)

Chart 2 - 2019 Quarterly cryptoasset market cap change (USD billion)

2. A look at correlations over the second quarter of 2019

“Correlation statistically measures the strength of a linear relationship between two relative movements of two variables and ranges from - 1 to + 1.”

In general, assets with a correlation above 0.5 or below -0.5 are considered to have strong positive/negative correlations. Similarly, a close-to-zero correlation indicates a lack of linear relationship between two variables, and for the purpose of this analysis, the returns of two assets.

If the returns of two assets do exhibit a positive correlation, it implies that the two assets tend to move in the same direction, and therefore share similar risks. On the other hand, a negative correlation between the returns of two assets indicates that the two assets move in opposite directions, and it is thus possible to use one asset as a hedge against the other.

In this report, 30 of the largest cryptoassets (excluding stablecoins) were selected based on their average market capitalization over the first 6 months of 2019.

In the following section, correlations in the second quarter of 2019 are discussed along with insights and observations from the dataset.

Chart 3 - Large-cap asset 2019 Q2 daily return (USD) correlation matrix

The overall correlation between these top digital assets remains quite high, with a majority exhibiting correlations above 0.5 with other cryptoassets.

Additionally, no single pair of cryptoassets exhibited a negative correlation, which indicates that the influence of idiosyncratic factors on the performance of individual coins is still relatively minor, and individual cryptoasset prices are still heavily influenced by overall market changes (as highlighted in our previous report¹).

Nevertheless, an analysis of several idiosyncratic factors and events suggests they may have had an impact on correlations of specific crypoassets in Q2 2019:

• Chainlink (LINK): in June, it was announced² that Google would integrate Chainlink for its smart contracts. Matic Network, a project that successfully completed a token sale on Binance Launchpad, also announced a partnership with Chainlink³.

• Bitcoin SV (BSV): there were many exchange rumors, ⁴" along with major delistings from large exchanges (e.g., Binance and Kraken delisting BCHSV/BSV⁵), which caused unusual price fluctuations independent from other projects in the overall market.

• BNB: BNB moved from the largest ERC20 on Ethereum to become the Binance Chain Native Token and exhibited a continuous price rally over the first two quarters of 2019. Other news and events related to Binance also brought added interest to BNB, such as the release of margin trading along with other new features on the CEX (e.g., One-Cancel-Other orders) and the development of new fiat exchanges worldwide (e.g., Binance Singapore).

In addition to project-specific catalysts, consensus mechanism design and exchange listings were previously discussed in our March 2019 report as potential key factors influencing correlations among specific assets. In revisiting these catalysts, we observed similar effects during the second quarter of 2019:

• “Binance Effect”: Tezos (XTZ), Doge (DOGE) and Bitcoin SV (BSV) all exhibited lower correlations with other cryptoassets than the remainder of the set. Interestingly, Bitcoin SV being delisted from Binance led to its decorrelation with other cryptoassets between Q1 and Q2⁶.
• Consensus Mechanism: relative to findings from our previous Q1 report, consensus mechanism design still appears to impact correlations among assets. Specifically, Proof-of-Work assets were, on average, more correlated with each other than with non-PoW assets over the second quarter of 2019.

Interestingly, additional idiosyncratic factors⁷ impacting correlations were observed from the dataset (see chart 3):

• Privacy factors: DASH (DASH) and Monero (XMR) were highly correlated (0.85), which seems intuitive given their similar focus on privacy. If Litecoin (LTC) were to implement MimbleWimble in the future⁸, it would be interesting to monitor closely Litecoin’s correlation with both of these cryptoassets going forward.
• Similar functions: XRP and Stellar (XLM) displayed an extremely high correlation (0.87). Despite Stellar being initially built on the Ripple protocol, its code was quickly forked and revamped. Yet, these two digital assets still share several similarities as both aim to “reshape the global remittance industry”.

In the next subsection, these results are compared with results from the first quarter of 2019.

3. Quarter-on-quarter comparison

Chart 4 - Large-cap asset 2019 Q1 daily return (USD) correlation matrix

Most of the correlations are similar to the results from our March 2019’s report about crypto-correlations⁹. However, it is worth noting that the average correlation between Bitcoin and all other major cryptoassets fell from 0.73 to 0.61 in the second quarter.

On the other hand, the average correlation between two cryptoassets remained stable around 0.57-0.58 between Q1 and Q2. Nevertheless, the average correlation coefficient for each of these cryptoassets changed, as illustrated in the table below.

Table 1 - Average correlation changes from Q1 to Q2

Some interesting insights can be drawn from the table above:

• Bitcoin (BTC) became less correlated with other cryptoassets in Q2 2019 relative to the first three months of 2019. Correlations declined between Bitcoin and altcoins, with a decrease in average correlation of -0.11. As indicated in the previous section, the overall market capitalization rose by 139%, whereas altcoin aggregated market capitalization(including stablecoins) increased by “just” 71% over the same period. This can likely be attributed to a "flight-to-quality" behavior by crypto investors in an early bull-market state, a behavior that may be further discussed in upcoming reports.
• Other Proof-Of-Work cryptoassets (e.g., BCH, DOGE, BTC, ETC, BTG) exhibited lower average correlations with other cryptoassets than in the first quarter of 2019. Instead of exhibiting larger correlations with each other, PoW coins all have displayed larger decreases in correlations with other assets.
• Tezos (XTZ), Ontology (ONT) and Tron (TRX) accounted for the largest increase in average correlation with other coins.
• On the other hand, Bitcoin SV’s (BSV) average correlation fell the most, which can likely be attributed to it being delisted from Kraken¹⁰ and Binance¹¹. Interestingly, Dogecoin (DOGE) also became further decorrelated against other cryptoassets, but maintained a significant correlation (0.55) with Litecoin (LTC), possibly due to the merge mining of the two coins. Despite some structural changes from the first three months of 2019 to the second quarter of 2019, correlations remained fairly high between most cryptoassets, with the average correlation being 0.57 in Q2.

As a result, it is still difficult to diversify within the crypto-industry through long-only positions in large market cap cryptoassets. However, alternative strategies such as investing in lower-market cap cryptoassets or a combination of long/short positions may help investors building market-neutral portfolios. With increased access to DeFi¹² and margin products, investors may be able to access more liquid, sophisticated instruments to allow them to construct more unique strategies in the future.

4. A look at Launchpad

In the first three months of 2019, Binance Launchpad completed three successful IEOs (Initial Exchange Offerings):

• BitTorrent (BTT) on February 3rd 2019.
• Fetch.AI (FET) on March 2nd 2019.
• Celer Network (CELR) on March 24th 2019.

This section analyzes the strength of correlations among these Q1 IEOs during Q2 between March 31st and June 30th, 2019. Though the projects are unrelated in their target niches (BitTorrent focuses on decentralized file sharing, Fetch focuses on advanced data marketplaces, and Celer on layer-2 scalability), they were all listed under similar market conditions, reaching the open market shortly after completion of their public sale on Binance.

The top 7 market cap coins, based on June 30th market cap, were selected along with the 5 projects that offered tokens on Binance prior to the beginning of Q2 2019. Correlations across March 31st to June 30th were then calculated with each project’s returns.

Chart 5 - 2019 Q2 daily return (USD) correlation matrix of Launchpad coins

Most clearly, all three Q1 2019 IEO projects were highly uncorrelated with the rest of the market, exhibiting the lowest correlations observed against the top market cap coins; this effect was even stronger for recent Launchpad projects, as opposed to Binance’s first two Launchpad projects, Bread (BRD) and Gifto (GTO), which first offered their tokens in late 2017. On average, the three IEO projects exhibited an average correlation of 0.32 against all other coins in the cohort, compared to an average of 0.58 amongst the other 9 projects.

Notably, the three recent IEO projects also exhibited fairly low correlations with BNB, with an average of 0.22, despite their initial offering being (at least partially) priced in BNB. Most interestingly, BitTorrent (BTT) exhibited the highest correlation (at just 0.27) with BNB amongst the three, despite being the only one that had a sale that accepted two tokens for the raise: BNB and Tron (TRX).

Amongst the three IEO projects, correlations were relatively higher compared to correlations between recent IEO projects and non-IEO projects, with the lowest correlation being BTT and FET (0.48). One outlier was the correlation between Fetch and Celer, which recorded a shockingly high correlation of 0.87 during Q2 2019. One potential explanation for this phenomenon was that both were launched in March under similar market conditions, whereas BitTorrent launched a full month earlier.

It remains to be seen how much sector or focus of projects also contributes to general correlations amongst various cryptoassets. With more assets going through IEO, it remains to see whether some of the results will hold in the future and whether some of the findings are applicable to a broader set of cryptoassets¹³.

5. Conclusion

Compared to the first quarter of 2019, the average correlation between Bitcoin and most other large cryptoassets declined; however, general correlations amongst other cryptoassets remained quite high in the second quarter of 2019.

In line with our previous reports, idiosyncratic factors are seen as the key factors impacting the strength of correlations among cryptocurrencies and digital assets.

The noticeable decrease in Bitcoin’s correlation with other assets might be explained by a potential “flight-to-quality”, paralleling the same Bitcoin dominance increase at the beginning of the 2017 bull run.

This report focuses on recent changes in correlations in the industry and further analysis could help in assessing whether today’s crypto-market environment behaves similarly to historical early stages of bullish environments in traditional financial markets.