A recent article on the contango histogram has brought many comments. One of our readers, Adam, wrote a very good comment directly under the article. He asks what is the right way to compare the true value of the spread with historical results. This is a really good question, and at the same time a suggestion to write today’s article. In the previous article, I have shown you an incorrect approach to analyse the spread value. And today I will describe the right one.
“Hello Eli, thank you for the very useful article. Indeed, I had been using the histogram exactly as as described in the article as a wrong way. Personally, I am intuitively tending to use the histogram in such way, so I would ask if you plan to include in the Spreadcharts app a distribution that would focus on a particular spread? If I focus on the current Contango tab (for example, last 5 years – F2-F1), I see the development of the individual spreads that had been traded in this position for a given time period. This information is very important for viewing the market as a whole and for identifying volatile periods / combinations, but actually (as you write) it does not help me with the distribution of the price of the currently traded spread combination in the time. Such information would be (in my point of view) very important, because I could not only find the bottom and the price top for the chosen period but, and above all, compare it with the current price. I do not know if it makes sense, but if there are more of us who use the histogram incorrectly, then it is obviously there are more of us who tend to search for the same. Of course, thank you for clarifying the things in the article, as it can mean avoiding many losses. Good luck, Adam. “
The spread value is relative
From what Adam writes in the comment, I understand that he would like to compare the actual price of spread with some historical period and create the distribution of a particular spread from this price history. This would de facto mean applying a histogram to the price curves made from stacked seasonality charts of a particular spread. This would give us some idea about the distribution of the price of a particular spread, but not real values. Again, I think that this would also be a misuse. The resulting conclusions would be distorted, and this could have a negative impact on our trades and thus cause unnecessary losses.
The reason is that comparing the spread price over a longer period is not correct. The spread value depends on external influences such as physical storage costs, financing costs, risk premium, etc. The amount of these costs is slowly changing. For a short period of days, weeks, or even a few months, it will not change much. For a longer period of several years, however, it can change fundamentally. Therefore, it does not make sense to compare the absolute price of the spread today and price few years ago, because the situation can be quite different. In other words, a corn interdelivery spread which is traded for 20 cents, can now be considered expensive. But it is quite possible that the same spread at the same price would be considered very cheap a few years ago.
Grains and oil plants
The best way how to get the idea of the spread value in the historical context is to compare its relative value. So, to compare its price in relation to the abovementioned external influences. On grains and oil plants, we are fortunate that SpreadCharts has already a ready-made model that tracks the value of the mentioned external influences. As a result, we compare spread price ratios to full carry, which reflects the maximum cost of storage, financing, etc. Please see the example of the following chart for corn spreads:
On this chart we can easily compare the spread value (as a share of full carry). This is the best and most accurate way how to compare the value of the spread with the historical results. Based on the comparison of the current blue curve with the curves from previous years, we can precisely conclude if the spread is underrated or overvalued.
But what about the other commodities where the full carry model is not available? There’s not much to work with here. Yet there are bad and “less bad” approaches to do it. The bad approach is to look at the stacked seasonality chart. I explained this reason few paragraphs above. The price does not take into account changing storage costs, financing, etc., which fundamentally affects the spread price. Still, the stacked seasonality chart is used by the vast majority of traders.
But there is also a “less bad” way. If we cannot calculate a full carry model, it is better to express the price of the spread by contango. We can at least partly balance the effect of financing costs. Contango for a specific spread can be displayed in the Spreads tab in the application, and then we can compare it with the historical – stacked seasonality chart. So how to display it? Contango always expresses the relative difference between two contracts in relation to the closer one. If we want to see the percentage of the contango, it is enough to multiply it by 100. Therefore, in the application, we use the formula: 100 * (F2-F1) / F1. The following chart is contango between January and February gas contract (100 * (NGG19-NGF19) / NGF19):
And how about the histogram?
So, we already know that it is not a good idea to compare the regular spread price curves with the stacked seasonality charts. We also know that it is the best to compare curves on full carry. Where the full carry model is not available, the best of all the wrong options is to express spreads through contango.
From the original Adam’s query, however, it comes that he would like to take the curves of the stacked seasonality charts, merge them together, create a histogram from all of these data and compare the current spread value with it. Well, we know that just seasonality cannot be applied. But why not to make a histogram from a stacked full carry chart or contango chart for a specific spread? … simply because we would get a very distorted result again. I’ll explain why.
The key factor in contango histogram made from continuous contract is that the given curves consist of a combination of contracts that always have approximately the same expiration time. This is probably the most important parameter of each futures contract, regardless of whether it is futures on a physical commodity or a financial derivative (such as VIX). However, physical commodities may be distorted by different harvests that ripple term structure in general.
If we would make a histogram from contango curves of a particular spread or full carry, we would include in the calculation both the spreads with a yearly expiration, but even spreads with one month to expiration. Again, we would compare apples with oranges. For example, the spread of the ZCN19-ZCZ18, which I have traded this year, behaves completely differently a year before the expiration and a few months before the expiration. Simply put, one year before the expiration a commodity contract is made for a commodity that does not yet exist and not many things can affect it. While a few months before the expiration, it is a growing plant, which is affected by the weather and other things.
Therefore, on SpreadCharts you will find a histogram only from the contango of continuous contracts and not from the contango of a particular spread or from full-carry ratios. Because only continuous contracts keep the condition of approximately constant expiration time.