How (not) to Search for Profitable Commodity Spreads

In our courses of Working with SeasonAlgo platform we always focus a lot on practical search and filtering of spread strategies that have a high winning and profitability potential. These search and filtering criteria get explained and demonstrated in different sections of Strategies module as well as in Backtest and Optimize sections of Analyse module.

In our courses we often meet traders who already trade commodity spreads for a number of years and have been actively using the SeasonAlgo platform. They watched introductory video tutorials available on the internet and realized that the SeasonAlgo platform has a very strong potential that the traders cannot fully exploit without getting more profound training. Therefore, these traders take part in our Courses of Using the SeasonAlgo platform which are authorized by the SeasonAlgo developers.

In today's article we shall elaborate on the topic that we often encounter our courses. It concerns searching for spreads by percentage of historical winning trades. In other words, the search is based on the highest possible number of historical years in which the strategy resulted in profit. Most often the traders perform the search on a history of last 15 years. In case the strategy was profitable in 14 out of 15 years, the SeasonAlgo displays a 93% winning of the strategy. If the strategy resulted in profit in 13 years, then its profitability is 87%.

The vast majority of traders focus on searching for spreads with 100% profitability in a given historical period. However, they intend to forget, that the strategy is listed as profitable even if the profit in a given year was only $ 1. The overall strategy performance may provide a high Total Profit figure and a high percentage of profitable years. However, such figures tell nothing about the stability of the strategy. Therefore it may very well be that a 100% winning strategy will gain $ 100 annual profit in 13 years and extreme profit of $ 2 000 in remaining 2 years.  The Total Profit of all the 15 years would be $ 5 300.  The Average Profit in 15 years would be $ 353 but such profit would actually only be reached twice. As you can see, at first sight the strategy may seem very promising, because it was profitable in each year during past 15 years. But such a figure does not allow us to reveal the fact that these profits were very unstable.

Of course we can find more detailed data in the Backtest section of SeasonAlgo. But much more convenient and efficient is to search for a stable strategy directly in the Strategies module by using appropriate filtering criteria. Otherwise we would have to manually analyze one strategy after another.

Information about stability of reaching a certain level of profit is provided by SeasonAlgo filter %APW (Average Profit Win). This value represents percentage of trades which reached the Average Profit. In other words, it is a percentage of trades meeting the condition that the highest value (Best) reached by the strategy in a given year was higher than the Average Profit (PLØ).

Let's have a look at two examples of spreads with 100% profitability in a period of 15 years but with different values ​​of APW%.


The spread WN15-WK15 has a APW% value of 33%. It means that only in five years the strategy reached the Average Profit. This can be checked in backtest. The Average Profit in 15 years was $ 986.67. The relevant years are shown on the image bellow. It is also very clearly visible on cumulative equity curve which shows the price jumps instead of the smooth curve.


On the other hand, there is the second spread LHG-LHZ with a smooth 45° cumulative equity curve. APW% is 87% which means that the Average Profit of $ 760 was reached in the 13 out of 15 years. And again the Backtest shows in which years was the Best value higher.

All of these examples clearly demonstrate that the winning percentage is often overestimated when searching for profitable spread strategies. Moreover, the criterion of stable profitability gets usually completely overlooked. It is always very important to monitor other factors and to assess them in a context of the given market and in a combination with other Strategy Performance Data.

In no case this article recommends to search for spread strategies only by filters Win% and APW%. The issue of searching, filtering, analyzing and optimizing commodity spreads is a complex subject which can be taught in the following trading courses:

Commodity Spread Trading - group or individual training

Using the SeasonAlgo platform - group lessons or individual training or online course

Workshop of Spread Trading with the Advanced Bactester of SeasonAlgo



You might also be interested in

In our previous articles we explained why it is not possible to use historical ...

In this article we will look at the futures spread, which SeasonAlgo platform provides ...

Are you curious about results of our survey of bi-weekly analyses of futures spreads? ...

In the previous article we explained the principle of calculating historical patterns in the ...