Parabolic SAR

[bpcooper]

I would really appreciate any feedabck on this. I use the parabolic SAR for a trailing stop off either 30min or 60min charts. I notice that the SAR values keep changing. As the price moves towards the SAR level, it will back off slightly. As I am pretty sure that Wilder's formula is not based on the current bar but only the previous bars, how does this keep happening? Does anybody have a copy of the formula eSignal is using so I can check it? Has anybody written their own version . I keep getting stopped out at what was previously reported as the SAR level before it turns around only a tick or two from the new SAR level.

[eSignal_JasonK]

Hello bpcooper,

I see the behavior you are describing with version 7.8. I know there has been some work done on the Parabolic SAR study for the 7.9 version. I checked the current beta version of 7.9 and the study appears to be plotting correctly. Also, as far as my understanding goes, the value of SAR can change during the bar interval and it can back away from the price. However, it's not supposed to go past the previous bar's SAR value.

Version 7.8


Version 7.9

[bpcooper]

Thanks for your response Jason. However, I am still confused as how it can move the stop further away ( or at all for that mater). From my research, I have found the formula to be:

if Long,
SAR = ([high(-1) - SAR(-1)] * AF) + SAR(-1)

if Short,
SAR= ([Low(-1) - SAR(-1)] * AF) -SAR(-1)

AF = aceleration factor.

As this formula only references previous bars, it would not allow the SAR level to move the stop further away during the course of the bars movement.
Is there anyway I can see the eSignal code for the Parabolic SAR. I would really like to more confident in using this as I am risking large sums of money on it.

[eSignal_JasonK]

Hello bpcooper,

The code for the SAR is not in EFS form. Its built-in to our software so I can't show you that code. I don't know the exact calculations for the study, but I'm pretty sure they are much more involved than the equations you posted. I'll forward a message to development to investigate this. However, due to our current schedule with the upcoming 7.9 release they most likely will not be able to look at this right away.

If you have any on-line references for the research you've mentioned, please post some links here. That would be very helpful.