Spread trading offers some of the most dependable opportunities in trading and is one of the best ways for traders to get started with a small account. Finding good trades, however, can be a daunting task. For a sample of twenty commodities, there are hundreds of intra-commodity spreads and over ten thousand inter-commodity spreads. This article presents a technique that can be used to help in the process of sifting through these many possible spreads to find the most profitable trades.

To demonstrate the technique, consider the fictitious seasonal chart below for the January contracts of commodity X (figure 1).

Figure1

What would the seasonal chart look like for the March contracts of commodity X if those contracts had the exact seasonal tendencies of the January contracts? It would be the same as figure 1 except that the curve would be shifted down and to the right as shown in figure 2.

Figure 2

Notice that the vertical distance between the two curves remains constant in time. This only occurs because of the assumption that the different months have identical seasonal patterns. As long as this vertical distance remains constant, a spread of these two contracts will, on average, remain constant. So, there’s no profit opportunity.

In order to find a profit opportunity, one should look for contracts that have different seasonal patterns. One method for doing this is by looking for divergences between the seasonals. For example, consider the seasonal chart for the January and May contracts of commodity Y in figure 3.

Figure 3

In this example, the divergence between the seasonals for the January and May contracts can clearly be seen. Beginning in August, the seasonal for the May contracts begins decreasing while the seasonal for the January contracts continues to increase. This alerts us to the possibility of profiting from a spread formed by going long the January contract and short the May contract. The position should be initiated at the beginning of the divergence and closed when either the divergence ends or one of the contracts gets too close to expiration.

Have all the divergences between the seasonals for the January and May contracts for commodity Y been found? No, they have not. Figure 3 gives no information about spread opportunities during the period from December through May because of the way the seasonals were superimposed in the figure. To see all the opportunities, it is necessary to plot two cycles of each seasonal as shown in figure 4.

Figure 4

This figure shows another divergence between the seasonals. Starting in January, the seasonal for the May contracts increases while the seasonal for the January contracts decreases. This divergence suggests the possibility of profiting from a spread formed by going short the January contract and long the May contract. This position should be initiated in January and held until the May contracts expire.

Now, let’s apply this technique on some real data. Consider the seasonals for the August and October contracts of Feeder Cattle shown in figure 5 (generated by the Multiple-Month Seasonal program).

Figure 5

The rectangle shows a significant divergence. Figure 6 shows this divergence in detail.

Figure 6

The figure shows the seasonal for the October contracts gaining on the seasonal for the August contracts starting around October 4 and continuing until the August contracts end. But, before determining the historic profit/loss for this spread (long Oct/short Aug), let us first see in how many of the past years these contracts actually traded during this time period. This can be determined from the chart shown in figure 7 which was generated by the Spread Calculator program.

Figure 7

This chart was generated by shifting the past spreads forward in time so that they all expire in the same year (this is actually the first step in creating seasonal charts). It shows that the spread did trade from October 4 until the spreads expired. If we zoom in on the expiration as shown in figure 8, we can see that if a close date of October 20 is chosen, then the spread will have traded for the all of the last 10 years (only the past ten years are considered since ten-year seasonals were examined).

 

Figure 8

The profit/loss for this trade can now be determined. Table 1 shows the results for the past 10 years.

Table 1

spread		open			close			profit/loss
LC2002V/LC2003Q		10/04/2002	-0.97500		10/18/2002	-0.90000		0.07500
LC2001V/LC2002Q		10/04/2001	-3.38000		10/19/2001	-2.03000		1.35000
LC2000V/LC2001Q		10/04/2000	-1.25000		10/20/2000	-1.33001		-0.08001
LC1999V/LC2000Q		10/04/1999	0.51999		10/20/1999	3.87000		3.35001
LC1998V/LC1999Q		10/05/1998	-2.94999		10/20/1998	0.58000		3.52999
LC1997V/LC1998Q		10/06/1997	-2.75000		10/20/1997	-1.53000		1.22000
LC1996V/LC1997Q		10/04/1996	8.64999		10/18/1996	9.14999		0.50000
LC1995V/LC1996Q		10/04/1995	2.77000		10/20/1995	4.50000		1.73000
LC1994V/LC1995Q		10/04/1994	2.82999		10/20/1994	3.89999		1.07000
LC1993V/LC1994Q		10/04/1993	1.69999		10/20/1993	1.97000		0.27001

number of trades: 10

average profit/loss: 1.30150

percent up: 90.0

At $400 per point, this amounts to an average profit value of 1.30150($400) = $520.60 per trade. Not a bad record especially considering that 90% of the trades were profitable.