©Risk Concern. All Rights Reserved.

Commodities investing, which is often regarded by analysts as stock trading on steroids, is, of course, a risky endeavor, more so than other common asset classes traded publicly. Investors and market participants, therefore, must have a thorough understanding of the risks involved.

This report presents a quantitative analysis on the risks involved in investing in some of the most traded commodities: Gold, Silver, Copper, Platinum, Crude Oil WTI, Brent Oil, Natural gas, Wheat, Corn, & Soybeans. How much can the aforementioned fall in a market crash & which is the best option to hold as an asset (investment) amongst the commodities discussed here, i.e., providing the highest returns relative to risk, is expounded in this work.

If you trade (or want to) commodities, or want to manage the risk of exposure, it is imperative that you thoroughly understand the factors discussed in this report.

For an assessment of same parameters for Cryptocurrencies, see __our report on top Cryptos__.

The table below (Table 1) presents the critical risk measures (details of factors presented—numbered according to row headings):

1. The long-term mean monthly returns

2. Volatility (as measured by standard deviation): how volatile monthly returns can be (how much they can deviate [±] from the mean)

3. The highest monthly price fall recorded in time-series

4. The highest monthly price increase in time-series

5. Sharpe ratio (mean returns – the risk-free rate of return [3.5% p.a. used] ÷ volatility): This ratio is used to measure the returns per unit of risk, & the higher this ratio is, the more attractive the asset is; for example, an asset with 5% average returns and 20% volatility (risk)[assuming 0% risk-free rate] would have a value of 0.20 as per this ratio, this means that the asset generates 20 cents of returns for every dollar risked

6. Monthly Value at Risk VaR [98%] is the statistical examination of potential losses per a given confidence level [98% used here]. Thus, as per the data and the figures calculated, maximum losses, 98% of the time, should be within the figure calculated; Nonetheless, there remains a 2% probability of losses exceeding the VaR figure, & hence, the possibility of higher than estimated losses remains

7. Quarterly VAR—it is the same parameter as above but calculated for a quarter

8. Is the percentage of months in the time series analyzed when returns were negative

9. Is the percentage of months in the time series analyzed when returns were positive

10. Best commodity to hold short-term, as per the analysis (calculated as Sharpe ratio ÷ monthly VaR); this parameter is an assessment of the best returns offered, adjusted for two parameters of risk; the higher this figure is, the more attractive the asset is for holding as an investment

Table 1. Findings

*long term data calculated post 1971 [the abandonment of the Gold-standard]

## Key takeaways

In a market crash, Gold 98% of the time, as per the data, shouldn’t fall more than -14.20% monthly (M) & -28.39% quarterly (Q), the same figure for Silver stands at -26% (M) & -52% (Q); for Crude Oil the figure is -26.09% (M), & -52.19% (Q).

This means that if the price of Gold, Silver, & Crude (in that order), for example, is $2000, $25, $80, in a crash, it can decline to $1716, $21.75, $59.13, in a month. A bear market which can continue declines for longer periods, can cause significantly more losses to holders (quarterly VaR is more relevant for estimating losses in a persistent negative period).

For all other commodities analyzed in this report, see column 7 & ([6] in the rows) of the table above.

See also:

Overall, as per the volatility of returns & VaR, energy is the riskiest investment sector: Natural Gas is the riskiest commodity to hold or invest in, as prices can deviate significantly; the second place goes to Crude Oil WTI & Brent Oil; nonetheless, Natural Gas (1.44%), Crude (1.06%) and Brent (0.91%) also have the highest mean monthly returns (see next section for further clarity).

Natural Gas is considerably riskier than WTI, as its price volatility is 64% higher than that of WTI; this difference is also statistically significant; the risk profiles of brent & WTI, on the other hand, are comparable and statistically similar, i.e., aren't different enough to have statistical significance (results of F-tests attached at the end).

Platinum has the lowest mean returns commensurate to the risk, while US Wheat & Corn come in second & third place for the lowest returns compared to risk.

## Commodities that offer the highest risk-adjusted returns

As per the Sharpe ratio analysis, Gold has provided the best risk-adjusted returns (8.28% return per 1 unit of risk, or put another way, 8.23 cents per every dollar risked by investing), followed by Crude Oil (8.31%) & Natural Gas (7.59%). For comparison, this ratio (long-term) for S&P 500 and Nasdaq stands at 12.1% & 18.8%.

However, when returns are assessed commensurate to two risk parameters (volatility & VaR), as calculated under RCD-VAR ratio [column 11 (10) row, in Table 1], while gold & Crude Oil still emerge as the best assets/commodities to hold (RCD-VAR = 0.63 & 0.32), the third-place goes to Copper (0.25), & fourth to Silver (0.24). It is worth noting that while the risk profiles of WTI & Silver are statistically similar, the risk profiles of Silver & Copper aren't statistically similar; Silver is a riskier asset to hold than Copper.

Another key point worth noting here is that Gold's returns over the last five decades (after the abandonment of the Gold-standard) have had a negative risk to returns ratio for 2 out of the last five decades; it has also had a negative Sharpe Ratio from the 1970s to late 2000s.

This fact means that it hasn't been a viable 'investment' from the 1970s to the 2000s; after the 2008 crisis period & decline in the risk-free rate, Gold has been, at least, a viable investment; nonetheless, its past record cannot be considered attractive when compared against other asset classes such as stocks & real estate, etc.

However, it is worth noting that if concerns regarding stocks being overvalued & bond yields being synthetically lowered persist, metals will have some level of attention. Still, their performance has been mediocre (to put it mildly) compared to other asset classes (particularly index-based broad ETFs).

Another key issue worth noting is that if the value of fiat currencies declines further, prices of metals, energy, & grains may become more volatile, as each currency unit equates to a lower value of an individual commodity. Simplistically, this means that if inflation becomes a critical issue in the future, we may see more price volatility in the commodity markets.

**Significance of work**

Those hedging against exposure to commodities, holding commodities, or interested in investing in (or trading) commodities can utilize the quantitative analysis provided in this work to gauge their own risks relative to exposure.

VaR figures provide details regarding expected losses those exposed to commodities should be prepared for, while other measures, such as the RCD-VAR ratio, provide clarity regarding risk-adjusted returns.

### Example usage of values provided in this work:

(1) Jessica may want to trade, say, futures contracts of Silver. She may have an assumption that Silver's price will continue to increase in the coming weeks, and thus, she may acquire a XAGUSD buy contract from her broker. However, she may also be concerned that the price may drastically fall, which may cause her losses on the buy contract(s). To determine how much leverage she should take on or to estimate how much the price can drop in a crash, she may utilize the VaR analysis of Silver (provided in this report).

As this value stands at -25.97% (monthly) & -51.94% (quarterly), at 98% confidence level, she should be able to estimate that for every $1000 invested in a buy contract, if the market declines, losses, 98% of the time should be -$259.7 or lower in a month, or -$519.4 in a quarter; alternatively, 2% of times, losses would at least be more than -$259.7 in a month, or -$519.4 in a quarter.

(2) Jason, concerned about inflation, may want to invest in commodities to shield his wealth from any deterioration of the value of currency he holds in the bank. He may examine the RCD-VAR ratio & the Sharpe ratios provided in Table 1 to identify the best commodity to acquire that offers the highest returns relative to risk. Data-driven insights are always superior to anecdotes & opinions; Jason should understand the importance of this point and make data-driven decisions.

Nonetheless, one should always understand that fluctuations in the period one invests in could differ from the mean figures presented in this work; thus, while long-term the actual values should be similar to the figures presented in this report, period-specific values, due to factors such as market microstructures and other developments, may differ.

See also:

For a further consultation on this topic, contact us!

### Statistical tests

__F test for variances, using F distribution (dfnum=378,dfdenom=428) (two-tailed)__** (validation)**

** 1. H0 hypothesis**
Since p-value < α, H0 is rejected.
The sample standard deviation (S) of

**NGas'**population is considered to be

**not equal to**the sample standard deviation (S) of

**WTI's**population. In other words, the difference between the sample standard deviation (S) of the two populations is big enough to be statistically significant.

** 2. P-value**
The p-value equals

**0**, ( p(x≤F) = 1 ). It means that the chance of type I error (rejecting a correct H0) is small: 0 (0%). The smaller the p-value the more it supports H1.

**The test statistic F equals**

__3. The statistics__**2.6779**, which is not in the 95% region of acceptance: [0.8216 : 1.2157]. S1/S2=1.64, is not in the 95% region of acceptance: [0.9064 : 1.1026]. The 95% confidence interval of σ12/σ22 is: [2.2028 , 3.2594].

__F test for variances, using F distribution (dfnum=585,dfdenom=428) (two-tailed)__** (validation)**

** 1. H0 hypothesis**
Since p-value > α, H0 is accepted.
The sample standard deviation (S) of

**Silver's**population is considered to be

**equal to**the sample standard deviation (S) of

**WTI's**population. In other words, the difference between the sample standard deviation (S) of the

**two**populations is not big enough to be statistically significant.

** 2. P-value**
The p-value equals

**0.1424**, ( p(x≤F) = 0.9288 ). It means that the chance of type I error, rejecting a correct H0, is too high: 0.1424 (14.24%). The larger the p-value the more it supports H0.

**The test statistic F equals**

__3. The statistics__**1.1422**, which is in the 95% region of acceptance: [0.8392 : 1.1945]. S1/S2=1.07, is in the 95% region of acceptance: [0.9161 : 1.0929]. The 95% confidence interval of σ12/σ22 is: [0.9562 , 1.361].

__F test for variances, using F distribution (dfnum=585,dfdenom=399) (two-tailed)__** (validation)**

** 1. H0 hypothesis**
Since p-value < α, H0 is rejected.
The sample standard deviation (S) of

**Silver's**population is considered to be

**not equal to**the sample standard deviation (S) of

**Copper's**population. In other words, the difference between the sample standard deviation (S) of the two populations is big enough to be statistically significant.

** 2. P-value**
The p-value equals

**3.902e-10**, ( p(x≤F) = 1 ). It means that the chance of type I error (rejecting a correct H0) is small: 3.902e-10 (3.9e-8%). The smaller the p-value the more it supports H1.

**The test statistic F equals**

__3. The statistics__**1.804**, which is not in the 95% region of acceptance: [0.8364 : 1.1993]. S1/S2=1.34, is not in the 95% region of acceptance: [0.9146 : 1.0951]. The 95% confidence interval of σ12/σ22 is: [1.5042 , 2.1568].

## Comments