This article will dive into what “value at risk” is and how you can use it to manage crypto assets.
The cryptocurrency market is notorious for being extremely volatile. The price of any cryptocurrency has a tendency of constantly fluctuating within a short period of time. At any moment, a digital currency can either go up a little or drop tremendously. The unpredictability of it all makes it nerve-racking and intense, but there are some who are drawn to that. Whether you love these fluctuations or hate them, it’s obvious that a fair share of risks can come from this.
In a market that is full of so much uncertainty – and in fact being well-known for it – there comes the topic of risk management. In this field especially, managing risks is very crucial for any trader. It is only by analyzing the potential investment risks that traders can determine the extent of probable losses in their portfolios. Not only that, but they can also figure out the occurrence ratio of those losses.
So, how can one go about evaluating portfolio risk? What tools are there to use? Well, there are a variety of different tools in the market to choose from. One of the most popular is ‘Value at Risk’ (VaR). The primary purpose of this tool is to calculate what is essentially the worst-case scenario that could occur while trading.
What does it mean?
VaR is a statistic that measures and specifies the exact level of financial risk within a firm, portfolio, or position. Particularly, over a certain period of time. The more common users of this metric are investment and commercial banks. They typically use it for determining the occurrence ratio of possible losses that could happen in their institutional portfolios. They also utilize it to figure out the extent of these losses. Many refer to this statistic as the “new science of risk management.”
Risk managers will often use VaR as a way to measure and control the level of risk exposure. One can apply VaR calculations to specific positions or even hole portfolios. In addition, they can apply these calculations in order to gauge firm-wide risk exposure.
The core purpose of VaR modeling is figuring out the potential for loss within the entity that’s subject to assessment. Moreover, to determine the likelihood of occurrence for the defined loss. One is able to measure VaR by way of evaluating the amount of potential loss and the probability of occurrence for the amount of loss. It can do the same thing when it comes to determining the time frame.
Example of how to use it
Let’s use a hypothetical situation with a financial firm as an example. This firm may come to the conclusion that an asset has a 3% one-month VaR of 2%. This is indicative of a 3% chance of the asset’s value declining by 2% during a period of one month. The conversion of the 3% chance of occurrence to a daily ratio establishes the odds of a 2% loss at one day per month.
Employing the use of a firm-wide VaR assessment permits the determination of the increasing risks. Specifically, from positions in a collection that different trading desks and departments within the institution are holding. By using the data that VaR modeling provides, financial institutions are able to determine if they have satisfactory capital reserves in place. With these reserves, they could potentially cover losses. Alternatively, they use the modeling to determine whether higher-than-acceptable risks require them to cut down on concentrated holdings.
The main idea
Generally speaking, the most popular and traditional measure of risk is volatility. However, the primary issue concerning volatility is that it has very little interest in the direction of an investment’s movement. Stock is prone to being volatile because it can often jump higher suddenly and at random. Be that as it may, investors usually react with little distress.
For most investors, the risk mostly pertains to the odds of losing money. Moreover, VaR draws its foundation from that piece of common sense. With the assumption that investors care about the odds of a humongous loss, VaR effectively answers two question:
- “What is the worst-case scenario?”
- “How much could I lose during a bad month?”
At this point, we can start getting a little more specific with our exploration of VaR. The conventional Value at Risk statistic consists of three components: a time period, a confidence level, and a loss amount. In the case of the third, the alternate name is ‘loss percentage’. Be sure to remember these three parts as they are important when highlighting examples of the question that VaR answers:
- What is the most I can expect to lose in dollars throughout the course of the next month? And with a 95% or 99% level of confidence?
- What is the maximum percentage I can expect to lose over the following year? And with 95% or 99% confidence?
As you can see, the “VaR question” contains up to three elements. One is a rather high level of confidence, which is typically either 95% or 99%. Another is a time period, specifically using either a day, a month, or a year. The third is a rough estimate of investment loss that either dollar or percentage terms illustrates.
Methods of calculating value at risk (VaR)
Overall, there are three types of methods when it comes to calculating VaR. There is the historical method, the variance-covariance method, and finally, the Monte Carlo simulation.
1 – The Historical Method
What this method does is basically re-organize actual historical returns. In doing so, it puts them in order that ranks them from the worst to the best. Afterwards, by default, it assumes that history will repeat itself; from a risk perspective, anyway.
2 – The Variance-Covariance Method
This method is under the assumption that stock returns undergo normal distribution. Put simply, it requires that there is an estimation of only two factors: an average return and a standard deviation. This will allow you to plot a normal distribution curve.
3 – The Monte Carlo Simulation
This method incorporates the development of a model for future stock price returns. Moreover, it involves running an array of hypothetical trials through the model. Generally speaking, this simulation refers to any method that generates trials at random. However, by itself, it tells us nothing regarding the underlying methodology.
As is, there is no standard protocol for the statistics that help determine asset, portfolio, or firm-wide risk. For instance, statistics that arbitrarily come from a low volatility period will probably downplay the potential for risk event occurrence. Furthermore, it may understate the size of those events. Understating risk may increase with the use of normal distribution probabilities. Most of the time, these rarely consider extreme events or even black swan events.
Assessing a potential loss is indicative of the lowest amount of risk in a wide range of outcomes. Let’s use a VaR determination of 95% with 20% asset risk as an example. This evaluation represents an expectation of losing a minimum of 20% every 20 days on average. In the case of this particular calculation, a loss of 50% still provides validation for the risk assessment.
The financial crisis of 2008 would expose these problems, painting them as benign VaR calculations. As a result, they would understate the probable occurrence of risk events put forward by portfolios of subprime mortgages.
Value at risk calculations
The Daily Hodl did their own VaR calculation. They focus on the minute closing price of BTC/USDT between August 15–21, 2019. Just to clarify, this calculation assumes that log-returns undergo regular distribution.
Step 1: Calculate the minute log-returns
The calculation of minute log-returns use this formula:
In this case, we use the logarithm of returns rather than price returns. The advantages that come from using log-returns instead of prices is ultimately log-normality. If the prices are distributed log normally, then the log return distribution is conveniently normal. This is quite handy, considering much of classic statistics presumes normality.
Now, we divide the log-returns into 27 intervals: (-14%, -13%), (-12%, -11%), …, (12%, 13%), count the number of minute returns per interval.
Step 2: Calculate both the average and standard deviation of log returns
At this point, we can now calculate the average and standard deviation of log-returns. We can do so by drawing information from the formulas:
The average (µ) of 10,080-minute log-returns equals out to being 0.001083%. Furthermore, the standard deviation (σ) turns out to be 0.03170.
Step 3: Calculate value at risk (VaR) by drawing from confidence intervals belonging to normal distribution
If we operate under the assumption that the returns have normal distribution, we can see the worst 5% and 1% on the normal curve. They show the preferable confidence of traders and the standard deviation, as well as the average from this table:
In the end, there are two ways to get a better understanding of the VaR calculation results:
- We can expect that, with 95% and 99% confidence, the worst loss will not surpass 5.23% and 7.38% respectively.
- Investing $10,000 means you are 95% and 99% confident that your worst minute-loss will not exceed $523 (=$10,000 x -5.23%) and $738 (=$10,000 x -7.38%) respectively.
The bottom line here is that VaR is incredibly useful for calculating the maximum loss you expect from an investment. It determines this potential outcome over a specific time frame, as well as a specific degree of confidence. Traders are able to apply VaR to gauge the level of risk or possible losses of their trading portfolios easily. From this, they can take the measures necessary for keeping risks under control.