seatrout – Please use the ‘Insert PRT Code’ button when posting code rather than attaching an image of it.

When back testing you can speed it up by ignoring the main code or quitting in tests that don’t involve use-able times. I would use two variables. For example when testing every hour – one variable is starttime (000000 to 230000) and the other is addedhours (010000 to 240000). You can code something along similar lines when dealing with minutes or seconds. Here is the hourly example.

endtime = starttime + addedhours 

if endtime > 240000 then
quit
endif

if endtime = 240000 then
endtime = 000000
endif

Here is an example of the theory applied to a simple strategy to show how it works. To test every 15 minutes set up four optimised variables.

• StartTime from 0 to 23 step 1
• AddedHours from 1 to 24 step 1
• AddedEndMinutes from 0 to 45 step 15
• AddedStartMinutes from 0 to 45 step 15

This will then test every possible start time at 15 minute intervals and every possible end time at 15 minute intervals and ignore and quit any test that has an end time that is not a possible end time to speed up the back test. This obviously gives 9216 possible optimisations which does not leave much capacity for optimising other variables!

I would always advise against optimising time unless there are very sound reasons to such as market hours because optimising time will often turn out to be pure data mining.

once starttime = starttime * 10000
once addedhours = addedhours * 10000
once addedendminutes = addedendminutes * 100
once addedstartminutes = addedstartminutes * 100

once starttime = starttime + addedstartminutes

once endtime = starttime + addedhours + addedendminutes

if endtime > 240000 then
quit
endif

if endtime = 240000 then
endtime = 0
endif

if time >= starttime and time < endtime then
if not onmarket and rsi[14] < 30 then
buy 1 contract at market
endif
endif

if onmarket and rsi[14] > 70 then
sell at market
endif

graph starttime
graph endtime

Added as Log 178 to here …

Snippet Link Library