As a crypto asset holder, I have been wondering for a while as to what the best way would be to manage a portfolio in such a volatile market. BTC has clearly earned a position as digital gold. It has worldwide recognition, huge daily trading volume, and so many supporting crypto exchanges. But is it reasonable to just hold it and wait until its time has come? How can I minimize risk and maximize profit? Unfortunately, none of us really know what the future may hold. Therefore, yearning for a better portfolio management approach is a natural consequence of being a crypto asset holder.
I would like to invite us to think together about the following scenario: Let’s suppose I invest in both coins A and B. If the price movements of coin A and coin B shows an inverse relationship, then I can make a strategic move by switching coin B to A when the price of coin B drops, or vice versa. To check the feasibility of this plan, I explored the correlation among 3 well-known coins: BTC, ETH, and XRP. If there is no strong positive correlation among those coins, I can construct a diversified portfolio to minimize risk and maximize profit.
For the analysis, I downloaded 1-hour OHCLV data of the appropriate coins for the year 2019 from Binance. “Close” data is used for the analysis. Note I normalized the prices on these 3 coins for the purpose of chart comparison as they have different data scales. I applied the min-max scaling technique for normalization. So all price values have the same range from 0 to 1.
Let’s take a look at the raw value of closing prices. As you can see, we can sense the trend of each coin’s price but a comparison is difficult due to different scales.
Here is a scaled version of the price chart. Now we can compare each coin’s price.
Note: The red line is the normalized BTC price.
At first glance it looks like a correlation is present but I need concrete proof.
While there are several ways to conduct correlation analysis, I am going to use the Pearson correlation coefficient as it is a simple and well-known method. The coefficient tells whether two time series are correlated or not. The value ranges from -1 to 1. A value close to 1 means that two data sets have a positive relationship, whereas a value close to -1, implies a negative relationship.
Since the price data has a trend, first-order differences are used to calculate the coefficient.
The coefficient is shown below:
According to the coefficient, BTC and ETH have a strong positive relationship, 0.76. As a rule of thumb, we say that two variables have a strong positive relationship when their coefficient is 0.6 or greater. The strong positive relationship indicates that when the price of BTC goes up, so will the price of ETH with a great likelihood. The coefficient between BTC and XRP is 0.55-smaller than 0.76 yet still relatively high value. We can conclude that all three coins have a strong positive relationship one to another.
The previous correlation was done for whole 2019 data. But if we can determine the correlations on shorter time scales and track that correlation it will offer more insight into the correlation.
Let’s compute the correlation again with a rolling window. A rolling window means that sub-datasets of the full data set are used.
In time series analysis, too often recent lags have higher predictive power than old ones. Therefore, a model that is applied to the whole series (or on a ‘training’ subset) might hide several characteristics of the whole series.
The rolling window, therefore, consists of the selection of given window size (by analyzing the Root Mean Square Error). For example, if the best window size is 24, we will be applying the model on the sample that is comprised of the first 24 observations; then on the second window, comprised of observations 2–25; then observations 3–26; and so forth.
The rolling window helps you to assess whether the parameters estimated by linear methods (intercept and slopes) are time-invariant (an important assumption of the model).
Let’s dig further through histogram
The below chart is generated from the BTC-ETH Pearson coefficient through a 24 hours rolling window.
As you can see, the average Pearson coefficient is “0.82” which is a really strong positive relationship.
The chart below is BTC-XRP:
The average Pearson coefficient is “0.68”. It is a high number. So we can say that BTC and XRP have a strong positive relationship. This is the same result compared to the previous one.
Here is a small conclusion. Both ETH and XRP have a strong positive relationship in a 24-hour rolling window. That means that if the BTC price goes up, ETH and XRP’s price is likely to go up in a day. Before moving forward, let me make one thing very clear: correlation does not tell us about the order. In other words, it does not tell us which one would move first.
Knowing correlation is good, but that’s not enough. Think about when you are driving with your GPS. If your GPS tells you only the list of turns (left or right) without when to take those turns, it would be of no use. It is the same with the correlation data. To make the data practical and relevant for investment and portfolio management, we need to add the chronological dimension.
We can supplement that piece of information with differenced time-series data. Differenced time-series data could help us extrapolate chronological information from the data set by telling us the exact relationship between the time series data. In layman’s terms, this can tell us more information on which one moves first. Let’s look into the differenced time-series data of the three coins at hand (BTC, ETH, and XRP). For this study, I created two sets of differenced data set: BTC-ETH and BTC-XRP.
Those 2 datasets have correlation information from other perspectives.
Let’s take a look at the chart below.
We can interpret the value in following.
0 = no difference between 2 prices
>0 or <0, the difference between 2 prices
Big value means that the difference is big.
The BTC-ETH differenced chart shows that from 2019-01 to 2019-05, BTC and ETH move similarly because the value is close to 0. But since 2019-07 it has changed.
Suppose the differenced data show a pattern, then we can derive a mathematical function from the pattern and utilize that as a tool to manage the portfolio. Let’s check it.
We can model a function to understand the relationship between two coins. If the model explains the movement of differenced data, we can use it to decide when to sell or buy, and what to sell or buy.
The red line is a fitted model function. It seems to be a good fit. To know how much the fitted function can explain the residual data set, I calculated the r2 score to be “0.61”.
R-squared (R2 ) score is a statistical measure that tells you how well the model fits the data. The score ranges from 0 to 1, and 1 means that the model explains all the variability of the data. (r2 score: https://blog.minitab.com/blog/adventures-in-statistics-2/regression-analysis-how-do-i-interpret-r-squared-and-assess-the-goodness-of-fit)
With an R2 score of 0.61, we can conclude that the model explains the BTC-ETH residual quite well.
The BTC-XRP has an even bigger r2 score of “0.78”. So when I looked at the r2 score. The BTC-XRP pair seems to be a better choice than the BTC-ETH pair because their fitted model is more reliable than that of BTC-ETH pair by having scored a higher r2 value.
The test result shows that ETH and XRP are correlated with BTC. The result implies that the three will likely show similar price movements. Therefore, we can conclude that it is not the best idea to hold both ETH and XRP as an attempt to offset the loss from BTC. We did not expect XRP to show a positive correlation with BTC as XRP had displayed a price movement different from that of BTC for a brief period in last year.
The following chart displays green triangles on points where BTC and XRP least correlate (coefficient<0.2). “2019-05” was the month with the least correlation with BTC. However, most of the time, it correlates with BTC.
In portfolio management, diversification is heavily emphasized as a key principle to minimize risks. In this volatile and unpredictable market, minimizing potential risks is highly recommended. As the crypto market is highly correlated and dominated by BTC, a very careful design and study are required to construct a stable portfolio that can weather a high level of risk.
If you would like to check the performance vs. risk of your crypto portfolio for free, visit https://moa.finance and try our free analyzer.
Copyright © 2020, moa.network / Powered by moa.network