Decred Price: A Seasonality study.
A look into the daily and monthly DCR returns patterns
This article is a simplified version of a paper on Decred’s seasonality, available here. I would like to thank Richard Red for his feedback when writing this article. Find me @itsoneanalyst.
TL;DR
- Seasonality effects surged in the 1980s for traditional financial asset returns (e.g., Stocks, Currencies, Commodities, FOREX).
- Recent research explored these effects in Bitcoin as the most popular cryptocurrency. However, new cryptocurrencies offering better performance by aligning PoS and PoW features and newer hash functions such as Decred have not been a target of financial literature.
- I study the day-of-the-week effect and month-of-the-year effect on DCR’s daily returns, meaning if any/several days of the week/month of the year show an abnormal superior/inferior return.
- I find a statistically significant result at the 10% level for Saturday’s mean return for the full sample, and other statistically significant days of the week in yearly periods.
- I don’t find any month-of-the year effect in the full sample.
- I present 4 strategies based on past returns in a theoretical scenario of no transaction costs and liquidity risks.
1. Introduction
Initial Focus on Bitcoin
Cryptocurrency is a novel research theme since Bitcoin, a decentralized currency protocol, was launched by Nakamoto (2008). Bitcoin offers a peer-to-peer decentralized currency in contrast to traditional FIAT currencies (e.g., USD) that operate under a central bank. The ability to secure, fast and private transactions have since been explored by many new cryptocurrencies.
Decred: A Better Solution
One of the cryptocurrencies offering advancements from these features is Decred (ticker symbol: DCR). Decred offers a hybrid solution between PoW and Proof-of-Stake (PoS) protocols, unlike Bitcoin.
By not being solely based on one of the protocols and by adapting a higher performance hash (Blake256), Decred can take advantage of this flexibility to offer a more complete solution. As such, Decred has been increasingly adopted worldwide, being the #38 biggest cryptocurrency in the market at publishing time.
Seasonality Analysis
The Month-of-the-Year, Day-of-the-Week effect, and Weekend effects have been studied in traditional financial markets (e.g., Stocks, Commodities, Forex). In crypto, most of the recent financial research is focused on Bitcoin. This article aims to increase the gap in financial research about Decred and seasonality effects.
Financial asset return’s anomalies occurring on a particular day of the week, a month of the year, or after a weekend (reflected on Monday) raise questions regarding the Efficient Market hypothesis (Fama, 1970). The fact that the cryptocurrency market is always open and its distinct features may cause different effects from those found on traditional markets.
Hence, this article aims to answer the following questions:
- Do Decred’s (DCR) returns have any day of the week with an abnormal value?
- Do Decred’s (DCR) returns show a month of the year effect?
- if those effects are present, could investors draw strategies based on these opportunities?
2. Background Research
Are Markets Efficient?
The Market Efficiency hypothesis (Fama, 1970), in which security prices fully incorporate all available information to investors, has been tested throughout the years in traditional markets. One of the market anomalies that raise questions about this theory is the weekend effect, presented in a comprehensive set of research about stocks, currencies, or Forex.
Initially, French (1980) found a negative return on Monday’s stocks return as much as 3 times more than in other days. That negative Monday return was later found to reduce, with these anomalies thought to disappear as markets mature.
Seasonality Across Cryptocurrencies
Kurihara and Fukushima (2017) find that Thursdays show the highest mean return (statistically significant) and present the idea that the market becomes more efficient as time passes.
More recent studies as Decourt, Chohan, and Perugini (2019) show Bitcoin’s returns anomaly in other days: Tuesday (highest mean return), followed by Wednesday and Saturday.
Baur et al (2019) document a higher positive mean return on Mondays. Ma and Tanizaki (2019) report the same higher Bitcoin’s mean return and volatility on Mondays, while the lowest happens on Wednesdays.
Caporale and Plastun (2019) investigate the day of the week effect across 4 cryptocurrencies (Bitcoin, Litecoin, Dash, and Ripple), but only find an anomaly in the case of Bitcoin. The authors also report a higher Bitcoin’s mean return on Mondays in comparison to other days. These results are reconfirmed by the findings of Aharon and Qadan (2019), who also expose a significantly higher return and volatility for Bitcoin on Mondays.
No effect after all?
In contrast, Mbanga (2019) argues that Bitcoin’s price clustering is not exclusive to one day of the week.
Yaya and Ogbonna (2019) expand the study on the week-of-the-day effect across 13 popular cryptocurrencies. The author’s results suggest market inefficiency for 6 cryptocurrencies: Doge, Ethereum, Maidsafecoin, Ripple, Stellar and Verge.
3. Data
The daily prices on Decred (ticker: DCR) from Coinmarketcap were retrieved. The available range period was from 11/02/2016 until 31/12/2019, accounting for 1420 unique observations. The daily logarithmic returns were computed.
Figure 1 shows the time series data for DCR’s price. During the sample, the daily mean returns for DCR were 0,2% with a 7,5% standard deviation.
4. Methods
A regression model was used to establish a relationship between the day of the week and daily returns to test the day of the week effect. The same methodology was applied between the month of the year and its returns.
Full methodology explained here.
5. Day of the Week Analysis
Full Sample
Considering the entire period, Saturday’s return is the highest from all days of the week (1.4%), and it is statistically significant at the 10% level. Hence, it can be assumed that there’s a day of the week effect on Saturday, even though it is only significant at the 10% level.
The lowest mean return day is suggested to be Tuesday; however, it is not statistically significant.
Monday’s mean return is not statistically significant, but it is found to be negative, as we’ve seen before in traditional markets. Thus, I cannot conclude that there’s a weekend effect for DCR returns.
Yearly Period Effects
When separating samples into the years Decred has been in the market, I find other significant days at a stronger significance level. During 2016, the days with the highest increase in price were on Mondays (3.2%) and Wednesdays (2.7%), both statistically significant at the 5% level. On the negative side, a statistically significant result at the 10% level, is found on Sundays (-1.6%).
During 2017, along with many other cryptocurrencies, Decred experienced a surge in price that may have impacted daily patterns, hence I don’t find any day of the week’s effect during this year.
In 2018, I find a statistically significant (5%) daily return at 2.7% on Fridays. Decred also experiences a great gain in price during this year, which changed the daily patterns seen so far.
In contrast, last year, Thursdays show a statistically significant negative return at -1.8%.
There is a shift in daily mean returns across years that challenges any conclusive pattern regarding daily returns. However, there is a significant result when considering a larger period (full sample) and one observable pattern on Saturdays— positive mean return across all periods.
6. Month of the Year Analysis
In this case, I find a statistically significant monthly mean return at the 10% level in January (1.7%). This finding goes hand on hand with the widely known January effect found in the stock market — superior price gains in January compared to other months. Even though this result is only significant at the 10% level, it shows some indication of this effect also present in the cryptocurrency market, in this case, regarding Decred.
7. Investment Strategies Simulation
By knowing the mean returns for each day of the week, an investor can draw simple strategies based on negative/positive day’s returns. In this article, we are simulating 4 strategies in a theoretical scenario of no transaction costs and liquidity issues, based on the coefficients found in Table 1 (full sample):
- “Holding”: Buying DCR on the first day of the sample used (11/02/2016) and sell it on the last day of the sample (31/12/2019).
- “Weekend”: based on the weekend effect. A strategy based on buying the currency at the closing price on Friday and selling it at the closing price on Sunday.
- “Saturday”: If Saturdays show the only significant positive return, I suggest a strategy based on buying the closing price on Friday and selling it at the closing price on Saturday.
- “Positive”: Based on holding the currency only during the consecutive positive return’s days. DCR returns are positive on Friday and Saturday, hence an investor can buy the closing price on Thursday and hold it until the closing price on Saturday, taking advantage of the consecutive positive days.
All strategies are based on the assumption that the investors dispose of a constant amount to invest each week during the sample period.
Figure 2 shows the cumulative returns for each strategy over the full sample period.
If an investor chooses to hold the currency from the first day until the last day of the sample, he/she has a period where the strategy is negative, which is not seen in any other strategy. However, the “Holding” strategy recovers from that period to end up as the second-best strategy.
In the end, the “Positive” strategy offers the best cumulative return. However, these cumulative returns do not account for the transaction costs related to weekly buying/selling of DCR and also its liquidity risks. Hence, the cumulative returns would be significantly less (35%-50% less of the total cumulative returns presented) if those were accounted for.
The Sharpe ratio, provided in Table 7, is a risk-adjusted measure that provides an investor with a better assessment of which strategy is adequate to its own risk aversion level (Sharpe, 1966).
The “Positive” strategy offers the best Sharpe ratio (4.39) across all strategies, statistically significant at the 1% level as well as a statistically significant annualized mean (759%) at the 5% level.
As we would expect, the “Saturday” strategy also offers a great option for investors with a 3.46 shape ratio, statistically significant at the 1% level, as well as, an annualized mean of 488%, significant at the 1% level.
As in the cumulative returns scenario, if transaction costs are accounted for, the Sharpe ratios would be significantly less — can be reduced by between 35%-50%.
This exercise is based on past returns, thus those do not give certainty that the same patterns or strategy’s returns will be achieved moving forward in 2020 due to the features of the currency, market conditions, and other outside factors.
8. Takeaway
Looking at daily patterns across the years, I find a considerable change in returns between days of the week when analyzing different periods. However, some statistically significant daily returns are found, at 5% and 10% levels, across all samples except for 2017. However, in the full sample — with the highest number of daily observations — a statistically significant result at the 10% level is found on Saturdays. Moreover, across all time samples the daily mean return on Saturdays is positive, even when not statistically significant, which leads to the good performance of the investment simulations (e.g., Positive and Saturday strategies).
However, the shift in daily patterns suggests that there is not a defined and consistent daily average that we can take into consideration as a reliable indicator for a future strategy, even with the historical positive returns from the test strategies. Those strategies show the possibility of employing seasonality analysis, but since the real impact of transaction costs and liquidity risks were not taken into consideration, they should be viewed mostly as a theoretical exercise.
Concerning the month of the year effects, I find some indication of a January effect in Decred’s returns following the same behavior as in stock markets.
It is worthwhile mentioning that a change in the time range used for all the seasonality analysis can impact the mean returns as well as their significance, hence the results are subject to changes as more data is available.
In the same direction, one of the reasons for the lack of significance in some analysis (e.g., month of the year effect) may be the low number of observations since Decred has only been in the market for 4 years. A follow-up study would be beneficial in the medium-term.
References
Full reference list here.
Disclosure
This article does not represent any kind of investment advice. The aim of it is to further educate the audience about Decred price behaviour.
Your own research should be develop if you are interested in investing in any cryptocurrencies or seek professional advice.
