Weekly Market Report - 15th February 2019
This weekly report aims to provide an overview of the crypto markets focusing on secondary trading. Though nothing here is investment advice, we hope this provides some useful and targeted information.
Weekly Market Report - 15th February 2019
Market Overview
We are focusing our market overview on the top 100 tokens from CoinMarketCap for now and the sector classification is roughly in line with what MyToken uses with some minor modifications. To be sure, we will continuously be updating the sectors and its constituents as we develop a deeper understanding of the crypto ecosystem.
This week’s new participants in the top 100 coins:
QNT, MOAC, VERI, ETP, ELF, GXC
Coins that dropped out from the top 100 coins compared with last week:
BCZERO, POLY, GXS, DENT, WAN
Largest Price Gainer (Previous 30 days): THETA +158%
Holo’s price has increased approximately 78% since our last analysis end date (Feb 1, 2019), and with about 109% price increase (in 30 days) from its lowest to the highest point.
In this report we choose the “Largest Price Gainer” by considering not just the price gain, but also based on the exchange it is listed on as well as the overall daily trading volume.
Returns vs Volatility
This is a look at mean and total daily returns vs volatility for the 17 sectors as well as the overall crypto and equity market. Some sectors only contain one or two coins/tokens while others have more than a dozen, so some numbers may be quite extreme –
Mean Daily Return vs Volatility from January 14, 2019 to February 14, 2019
Mean Return vs Volatility from January 14, 2019 to February 14, 2019
We have abbreviated the names of several sectors to make it easier to view:
M = Market
DC = Digital Cash
CP/MP = Computing Power/Mining Pool
A/M = Advertising/Media
G/E = Gaming/Entertainment
C = Classics
D/GT = Dividend/Governance Token
E/T = Exchange Token
OC/I = Off Chain/Interoperability
Rolling Returns of the Top 100 Tokens by Sector
Returns of the top 100 Tokens by sector from January 14, 2019 to February 14, 2019
Correlation Between Daily Returns of Each Sector
Correlation between daily returns of each sector from November 14, 2018 to February 14, 2019. The correlation ranges between -1 and 1. Correlation close to 1 or -1 means a very positive or negative relationship between the two subjects, respectively. Correlation close to 0 means no linear relationship between the two subjects.
The above figure shows the correlation between the daily returns of each sector. Correlations are very high between all sectors in crypto with the exception of stable coins and education (since there is only 1 token here). Stablecoins having a 0.32 correlation to BTC and 0.3 to the market is interesting to note since they are not supposed to move at all and should have near zero volatility.
Focus Spotlight — Chinese New Year and the Crypto Market
January effect is a seasonal anomaly in the stock market where prices tend to rally at the start of the year. One possible explanation suggests that investors often close their positions prior to yearend and start new investments in January. Does this also hold true for the cryptocurrency market?
Surprisingly the cryptocurrency market seems to exhibit the exact opposite behavior. If we look at Bitcoin prices from 2015 to 2018, the market has experienced major sell-offs during the first few weeks of January. Crypto investors named this phenomenon the “Chinese New Year Effect”.
Price of Bitcoin 60 days before and after Chinese New Year from 2014 to 2018.
The red line in the center marks the first day of the lunar year.
As shown above, although prices fell in January, these drops occurred prior to Chinese New Year. From 2015 to 2018, the market actually recovered around the holiday season.
Below we apply the decomposed additive model to Bitcoin prices from 2013 to 2019. Decomposition is a procedure used to identify the trend and seasonal factors in a time series data. In short, the model suggests that the actual price consists of three parts: Trend + Seasonal + Residual.
1. Trend is the long term moving averages of Bitcoin prices.
2. Seasonal is the pattern that repeats with a fixed period of time. Our time period is 365 days since we are analyzing patterns that occur every year.
3. Residual is the actual price subtracted by trend and seasonal, and is considered noise of price fluctuation.
The aim here is to identify the seasonal component and see if any patterns occur around January every year.
The procedures for creating this model are as follows:
1. The first step is to estimate the trend. Here we use the 365-day moving average.
2. Next we “de-trend” the price data by subtracting the trend estimates from the actual price data. Now we try to estimate the seasonal factors. For daily price data, this entails estimating an effect for each day of the year. A simple method for this estimation is to the average the de-trended values of each day.
For example, the seasonal factor for Jan 1st is calculated by first taking the actual prices on Jan 1st of every year subtracted by their trend estimates. Then we sum these numbers up and divide them by the number of years used. If the seasonal factor for Jan 1st is $1000, this means that on average, prices on Jan 1st are $1000 greater than their trend estimates. We repeat this procedure 365 times so that we have a seasonal factor estimate for each day of the year.
3. Now that we have the trend and seasonal estimates, we can calculate the residuals by subtracting these two estimates from the actual (observed) price. The residuals identify the price variation that is not captured by the model. Therefore, it is expected that the residuals should be random.
Decomposed Additive Time Series of Bitcoin from 2013 to 2019
The graphs above are the decomposed additive time series for Bitcoin historical price data since 2013 using 365-days moving averages. (The observed price is the sum of the trend, seasonal and residual values. Note that the seasonal effect repeats every year by nature due to our definition of seasonal effect)
The decomposition shows seasonality in the months of December and January. However, a pattern seems to form in the residuals from 2014 to 2017. It appears that the seasonal component is largely affected by the extreme fluctuation during late 2017 and early 2018. We decided to take another look at the decomposition using price data up to only 2017.
Once again, we observe a seasonality effect at the start of each year, where price plunged in January but recovered later in February or March. This time we see less of a pattern in the residuals.
The price drop can be attributed to the fact that Chinese New Year’s traditions include giving red envelopes and expensive gifts. People also tend to settle their debts prior to year-end to display model behavior and avoid bad luck. As a result, people would likely close their investments and convert them to cash beforehand.
Binance is one of the largest cryptocurrency exchanges that has a huge Chinese user base. Below is the comparison between the trading volume and trade counts on Binance and Bitfinex around 2018 Chinese New Year.
Although there had been no significant changes in terms of trading volume, from the trade counts we see an increasing level of activity on Binance, which continued to last even after the holiday. Due to the hype around crypto during that time, many investors continued to trade during the holiday instead of spending time with family and friends.
This year, however, seems to have no specific patterns in trading volumes on either exchange. Trades counts also remained stable for the weeks leading up to holiday. One possible reason for such difference is that the volatility in 2018 during Chinese New Year was at its peak. The price of Bitcoin was undergoing sharp reversal and high volatility spurred trading activity. On the other hand, the 2019 Chinese New Year sees a relatively calm period when the heavy sell-off culminated in mid-December 2018.
Another possible reason for such differences is the disposition effect, which is the tendency for investors to only sell assets that have made gains and keep the ones that have dropped in value. Therefore, due to the bear market, people may have been less incentivized to close their positions.
Data Source
We included data from sources such as CoinMarketCap for analyzing largest price gainer, volatility, mean daily return, total return and correlation between each sector; MyToken for sector breakdown; Binance and Bitfinex data for BTC/USD (USDT) trading volume and trade counts analysis.
Stay Tuned Here
KRONOS is a leading quantitative research firm based in Taipei, Shanghai and Beijing. We’re bringing new asset management strategies to the crypto world by leveraging our combined decades of experience trading in global traditional markets.
Website: https://kronostoken.com/
Telegram: https://t.me/Kronos_E
Twitter: https://twitter.com/KronosToken
Linkedin: https://www.linkedin.com/company/kronostoken/
