Profitable Parabolic SAR

2019년 2월 13일

Introduction
As you know, Parabolic Stop and Reverse (SAR, PSAR) was originally developed by J. Welles Wilder and was described in his book "New Concepts in Technical Trading Systems" (1978). It derives its name from the fact that when charted, the pattern formed by the points resembles a parabola.

Mr. Wilder described it as "one of my favorite systems because it squeezes more profit out of an intermediate move than any method I know".

Interpretation
PSAR follows price and can be considered a trend following indicator. Once a downtrend reverses and starts up, PSAR follows prices like a trailing stop. Same is true for the opposite direction.
Due to its nature, PSAR continuosly protects on long and short positions.

Parameters
One of the key components of PSAR is the Acceleration Factor (AF). The AF is one of a progression of numbers beginning at 0.02 and ending at 0.2. The AF is increased by the increment of 0.02 each time that a new high is made until a value of 0.2 is reached.

Mr. Wilder used the next parameters

• Start: 0.02
• Increment: 0.02
• Maximum: 0.2

and they are default for the built-in PSAR indicator and its strategy.

But are these params really profitable? Mr. Wilder noticed that "I have tried many different acceleration factors on this system and have found that a consistent increase of 0.02 works best overall...the range for the incremental increase is between 0.018 and 0.021".
That was then, in 1978. Other times have come. Is our grandpa still right in his recommendations?

I made this tool to figure it out.

What is this tool?
This tool is a performance scanner that uses a decision tree-based algorithm under the hood to find the most profitable settings for PSAR. It analyzes a bunch of different Start (between 0.001 to 0.02) and Increment (between 0.001 to 0.03) parameters and backtests each combination across the entire history of an instrument. If the more profitable parameters were found, the indicator will switch its values to the found ones immediately.

Instead of manually selecting parameters, just relax - the algorithm will do it for you.

It doesn't touch the last parameter, Maximum, for two reasons.
First, as Mr. Wilder noticed in his book, "...the number of increases it takes to reach at least 0.2, but do not exceed 0.22". That is, the parameter sits in a very narrow range.
Second, I tested different maximums and I came to the conclusion that this parameter has a minimal impact on net profit, compared with the more significant parameters of start and increment.

Alerts
It has an alert that notifies when the more profitable settings were detected.

NOTE: It does not change what has already been plotted.

Good luck!

2019년 2월 14일

릴리즈 노트

• Add a minor optimization fix
  

2019년 3월 10일

릴리즈 노트

• Now it analyzes Start parameter between 0.001 to 0.04 with a decimal step of 0.001
• Now it analyzes Increment parameter between 0.001 to 0.04 with a decimal step of 0.001
• Made optimization fixes
  

2019년 8월 12일

릴리즈 노트

• Added date range
  

2019년 8월 26일

릴리즈 노트

• Security update
  

2019년 9월 1일

릴리즈 노트

• Security update
  

2019년 9월 3일

릴리즈 노트

• Security update
  

2019년 9월 4일

릴리즈 노트

• Security update
  

2019년 9월 5일

릴리즈 노트

• Security update
  

2019년 9월 7일

릴리즈 노트

• Removed the end date for analysis
  

2019년 9월 25일

릴리즈 노트

• Security update
  

2019년 10월 26일

릴리즈 노트

• Security update
  
• Added commission customization
  

2019년 12월 23일

릴리즈 노트

• Update
  

2019년 12월 23일

릴리즈 노트

• Update
  

2020년 6월 1일

릴리즈 노트

• Update
  

2020년 9월 30일

릴리즈 노트

• Update
  

2020년 9월 30일

릴리즈 노트

• Update. One input name fix
  

2020년 12월 15일

릴리즈 노트

Update

• Added an option to choose a direction for optimization (Longs | Shorts | Longs & Shorts)