How Seasonal Charts Are Created

It is not necessary to have a background in computer programming to understand the manner in which the statistical data for this study was prepared. In order for you to more fully comprehend the results and the intensity of effort that was involved in their production, you may wish to familiarize yourself with the methodology. Here, in step form, is the basic procedure used in our computer analyses of the Weekly Seasonal Charts:

  1. Read the daily history file for each market and month for every year on file, i.e., June Live Cattle ’67, ’68, ’69, ’70...’07.
  2. Line them all up by week using Friday as the end of the week. a. If there are 2 or fewer trading days by the first Friday of the month then regard the week ending that Friday as being part of the previous month.
  3. Line up last X months of data to end of last pre-delivery month. If we’re doing the August Live Cattle contract, for example, then stop at the end of July. a. The only exception to this is Sugar. Stop 2 months prior to the expiration month for Sugar. So if the contract is for July Sugar then the last month of the chart should be May.
  4. Normalize the data: Max = highest price; Min = lowest price. All other prices X = (X – Min) / 100 * (Max – Min).
  5. Calculate the differences from week to week. Start with a closing price for the first week. Say that it was 10. If at the end of the next week the closing price is 11 then the week was a winning week and the gain was +1. Count up all the winning weeks and all of the losing weeks. (Do not count weeks with no change.) In the % square at the bottom of the table for each week show the % winners or the % losers depending on whether there were more winners or losers. In the square above the % squares show +/- depending on whether there were more winning years or losing years.
  6. In the squares above the +/- squares show an arrow if the % value is greater than 60. Show a down arrow for <= -60 and an up arrow for >= +60.
  7. The trend line is actually the cumulative gain/loss values over all the weeks. If the close at the end of the first week is 10 then the first plot on the chart is 10. If the gain for week 2 is +1 then the next point on the chart is 11. If the gain for the next week is -3 then the next point on the line is +8, etc. Given that the data is already normalized, the charts have a min value of 0 and a max of 100.

There are other ways in which the data could have been analyzed. If you plan to replicate this study with your own database, you may wish to experiment with different techniques of aligning the data, or with different indexing methods.

No price charts are 100% accurate. We have, however, validated the data, both in real time and on computer, to test for accuracy. To the best of our knowledge the data is reliable and reflects the true seasonal situation for each commodity.

Note that the weekly UP and DOWN percentages for a given contract month are applicable only to that contract month. It is not advised, for example, that you use the plot for November beans to trade July beans.

 

Next Topic: How To Read Charts