I got to thinking about seasonality again and this got me thinking that I should test to see if there are any sections of a year that if you had been long or short in every year would have returned good results and a nice equity curve. So I coded two very simple strategies that count week numbers and and buy on one week number and sell on another week number. Optimise the two strategies with the variables startweek and endweek with values between 1 and 53 and it will tell you which blocks of weeks have given the best return during the historical data. Then I created a third strategy that you can enter the best returned values as startweeklong, endweeklong, startweekshort and endweekshort so as to see what the equity curve would have looked like had you been short during the best weeks and long during the best weeks. On my tests I found that the best long weeks rarely clashed with the best short weeks but if you find an instrument where they do then you can’t put those values in the ‘Both’ strategy. The strategies only work on the weekly chart. The question up for general discussion is….. is this seasonality or is it just the ultimate curve fit? I have attached images of Live Cattle and Soybeans. No spread was used in the tests. defparam cumulateorders = false if openyear <> openyear[1] then wn = 0 endif wn = wn + 1 if wn = startweek then buy 1 contract at market endif if onmarket and wn = endweek then sell at market endif defparam cumulateorders = false if openyear <> openyear[1] then wn = 0 endif wn = wn + 1 if wn = startweek then sellshort 1 contract at market endif if onmarket and wn = endweek then exitshort at market endif defparam cumulateorders = false startweekshort = 17 endweekshort = 39 startweeklong = 39 endweeklong = 17 if openyear <> openyear[1] then wn = 0 endif wn = wn + 1 if wn = startweeklong then buy 1 contract at market endif if onmarket and wn = endweeklong then sell at market endif if wn = startweekshort then sellshort 1 contract at market endif if onmarket and wn = endweekshort then exitshort at market endif
.
.
