Talk:Parabolic SAR

Review
Article is generally understandable. I think it needs more references. This is an advanced topic that may not be understandable to those who are not involved in technical analysis of commodities or stocks.

 Comment begins here...

The "Statistical Results" section states that "The parabolic SAR showed results at a 95% confidence level in a study of 17 years of data.[3] ".

This could be misleading for a person ignorant of statistical concepts. The confidence level merely states the margin of error accepted by the data analyst in order to draw conclusions, and is not an indicator of the performance of the object of study, in this case, the Parabolic SAR. In other words, the performance could be determined with a 95% confidence level independently of it being good or bad. The section should be rewritten indicating the measured performance in the study (including the bands determined for that confidence level). — Preceding unsigned comment added by 201.231.131.39 (talk) 14:08, 4 January 2022 (UTC)

GT Comment begins here...

The mathematics is extremely sloppy.

If alpha is a time-varying parameter with bounds, it ought to be expressed as such - not displayed as if it is a parameter. According to the text, $$\alpha\!$$ starts at 0.02 and rises at a rate of 0.02 per iteration.

So in fact the appropriate formula is


 * $$SAR_{n+1} = SAR_n + \alpha_n( EP - SAR_n ) \!$$

where


 * $$\alpha_n=max(0.2, 0.02(n+1)) \!$$

If the seed value, maximum value and time-adjustment for $$\alpha\!$$ can be user-defined, then the true (non-time-varying) parameters are, say, σ,τ and μ, and


 * $$\alpha_n= max(\mu, \sigma + n \tau) \!$$

It is annoying when people write variables as parameters.

77.206.39.149 (talk) 15:25, 15 August 2008 (UTC) Geoffrey Transom, France Hello, I have a non trading background and I find the following phrase confusing "today's or yesterday's price range". This is mainly because "price range" in not define anywhere. Cheers Mrphonon (talk) 16:21, 19 February 2009 (UTC)
 * Prices range between the highest high and the lowest low, during a time period.

-oo0(GoldTrader)0oo- (talk) 16:15, 22 February 2009 (UTC)

Gbwi Comment begins here...

Some of the formulae by GT (above) are not entirely correct.

Since 0.2 is the maximum, the function in the formula should be MIN, not MAX, like this:
 * $$\alpha_n=min(0.2, 0.02(n+1)) \!$$
 * $$\alpha_n= min(\mu, \sigma + n \tau) \!$$

--Gbwi (talk) 09:15, 26 June 2009 (UTC)

Note that n is incremented only when a new EP (extreme point) is recorded.

--Gbwi (talk) 10:33, 16 July 2009 (UTC)

recursively calculated
Since "calculated" means a process, let us speak the truth and say that it is an iterative one.--VKokielov (talk) 16:10, 16 December 2010 (UTC)