Global Commodities Strategy

Last actualized 16 July 2020
Our global commodity strategy every minute recalculates potentials of the commodity companies selecting the most undervalued companies given the current commodity prices and exchange rates environment.
The strategy has worked great on backtest and since 31 May 2020 is working on the real accounts. The strategy is performing well since launch showing +19% result in just 2 months
In our view strong performance of the strategy is likely to continue due to
  • Ability of the strategy to recalculate potentials and pick the best companies in each market environment
  • Rising commodity prices due to increasing money supply / money printing by FED
  • Falling expenses of global commodity companies due to devaluation of currencies in South Africa, Brazil, Argentina, Russia in 2020
In this article we outline key rationale behind the strategy and show actual and backtest results
Introduction — necessity to take into account current market environment, nowcasting
Our basic approach to investing in commodity companies is to estimate fair value of the company every moment, taking into account last market environment changes and time passed since the last reporting date.

For example, if average palladium price in the last reporting period was $ 1600/oz, and the company had profit $ 1 bn last reporting period and it current Market Cap is $ 5 bn, it is much more convenient to be invested in this company in situation when current nickel price is $ 1800/oz compared to situation when it is $ 1300/oz. Moreover, fundamental value of the company in $ 1800/oz market prices environment would be higher.

Why higher? Traditional big banks research would say that fair value of the company is determined base on the DCF approach and on the long term price forecasts, while daily commodity price movements should have no influence on fair value of the company. Still, in our view the fundamental value would be different due to 2 factors:
  1. There is some probability that occurred commodity price movement is fundamental (e.g. connected with growing world demand for electric vehicles) and not temporary (like rising coal prices during floods in Australia) (in mathematical terms we could say that posterior mathematical probability of the prices in future period would be higher because commodity prices are auto-correlated as can be seen below)
  2. In fact, the company is earning more in current market environment and probably has earned more since the last reporting date, which means its net debt would be lower compared to situation of negative market environment

In this respect we believe that it is important to estimate 2 parameters: how much the company is earning in current market environment? What is estimate of its net debt taking into account of market environment and number of days passed since the reporting date (and dividends paid after the reporting date)?
In some other fields it is called Now Forecasting (or Nowcasting) — prediction not of the far future, but forecasting of the current state
So, if the company is traded cheaply on multiples of its current market value to profits in current market environment (compared to historical levels and other players), it might be undervalued, otherwise — overvalued.

For calculations we use Enterprise Value / EBITDA multiple as a multiple most often used in M&A transactions and investment analysis. In a few words, compared to pure P/E multiple It shows better results eliminating some one-off effects from earnings and valuating effect of the debt of the company on the capitalization level (rather than at the earnings level). More information on the multiple can be found for example in Toby Carlisle book "The Acquirer’s Multiple: How the Billionaire Contrarians of Deep Value Beat the Market".

In fact, quite often markets either overreact or underreact to market environment changes, so it is important to estimate effect of such changes in the same systematic approach.

Based on above we have developed the system that daily (and every minute) recalculates the potentials of commodity players, analyzing how its revenue and costs will likely change taking into account market price changes.
Small theoretical research of long-term commodity prices since 1960
While we don’t fully believe in the ability of traditional market analysts to predict commodity price movements, it is interesting to think of the following:

If certain commodity moves substantially up (or down) — in interest of long-term model, what is better — to imply commodity prices to stay and current prices, or to imply that it would return to historical levels (or to use some average approach)?

To answer that firstly we need to understand what parameter we are forecasting. Probably, if we believe that value of the company is discounted NPV of its future cash flows, we are interested in forecasting discounted price for the commodity for the next period; let’s take 10 years for this period in frames of our analysis.

We took Worldbank data for key commodity prices since 1960 and tested which approach would better forecast future discounted commodity price (the price which is important for the company valuation)

It is quite obvious that forecasting error is quite high — it is impossible to predict what would happen in the next 10 years.
For most commodities it turned out to be reasonable to use average between current and historical prices
However, it turned out that for all the resources it is worthwhile to use current market prices in the forecast (compared to using just historical levels). And for most of the commodities it is reasonable to take average for current prices and historical levels. In additional to increased precision it also reduces risks — when the commodity prices are elevated the system averages them with historical levels, which helps to avoid buying at maximums (which makes sense given cyclical nature of the commodity prices).

Using just current prices (compared to averaging current and historical prices) turned out to be more reasonable for oil and gold, which means prices for these resources historically showed more trend-following rather that cyclical nature
Comparison of precision of forecasting further commodity price depending on the approach – using average price for the last 5 years, using current spot price, using average price between current spot price and historical 5-years level

In green the approach arriving at the lowest deviation is highlighted

Currencies devaluation effects
Usually currencies tend to devaluate in negative market environment (economic problems, falling oil prices, etc). In these moments quite often market indices and commodity prices are falling as well.

But based on our experience market substantially underestimates positive effects of devaluation: while selling its production in dollars, expenses of the companies are nominated in local currencies, which makes commodity companies earn a lot more after devaluation.

The is why many Russian companies in our portfolio showed ~100% price growth in 2015 (after Russian ruble 2x times devaluation in 2014). Or after COVID shares fall on Mar 2020 many commodity players showed substantial overperformance.

In 2020 many currencies of developing countries substantially devalued (as can be seen below) which (combined with rising commodity prices) form favourable environment for relevant commodity players and for strategy as a whole.

The strategy back-test:
Year Strategy S&P 500
2015 -18.4% -0.7%
2016 115.9% 9.5%
2017 25.3% 19.4%
2018 -4.4% -6.2%
2019 92.9% 28.9%
2020 27.1% -2.2%
Average 34.5% 8.4%
Total 416.7% 56.4%
Backtest return by year (as of the date of the article)
Actual results