Hello, I've been developing an EA since the beginning of the year that places trades based off trendlines it places on the chart itself. At the program start, the EA engine looks at the last 500 bars before the current time and draws trendlines, never looking farther ahead than the current bar in the loop. Once it reaches the current trading bar, the EA starts placing trades based off the most recent trendline.
Its a relatively smart program. It knows to look from the max high or low of the last trend to draw the trendline, but it also can look further back that. And it recalculates on each new bar, in case a better trend is appearing.
What I'm interested in is two things. The first is if anyone else has taken this approach to EAs. I've heard of 1000s of EAs based off indicators but 0 based off trendlines. And this seems strange since all the Pros slam indicators but praise trendlines.
The second thing that I'm interested in, is improving the way I rank trendlines. Right now I'm using a medieval formula where the EA simply rates a trendline by counting its intersects(high or low goes through the trendline, but not the open or close). However I know I need a much smarter formula. I know the length of a trendline needs to be factored in so I'm considering implementing something along the lines of assigning each intersect a score based on its average distance from all the other intersects, and then tallying these scores to give the whole trendline a rank. However I already feel that this still may be too crude and I made need to add some sort of coefficient to the formula.
Again has anyone else tried ranked trendlines by a formula/algorithm before?
- Nicholas
Its a relatively smart program. It knows to look from the max high or low of the last trend to draw the trendline, but it also can look further back that. And it recalculates on each new bar, in case a better trend is appearing.
What I'm interested in is two things. The first is if anyone else has taken this approach to EAs. I've heard of 1000s of EAs based off indicators but 0 based off trendlines. And this seems strange since all the Pros slam indicators but praise trendlines.
The second thing that I'm interested in, is improving the way I rank trendlines. Right now I'm using a medieval formula where the EA simply rates a trendline by counting its intersects(high or low goes through the trendline, but not the open or close). However I know I need a much smarter formula. I know the length of a trendline needs to be factored in so I'm considering implementing something along the lines of assigning each intersect a score based on its average distance from all the other intersects, and then tallying these scores to give the whole trendline a rank. However I already feel that this still may be too crude and I made need to add some sort of coefficient to the formula.
Again has anyone else tried ranked trendlines by a formula/algorithm before?
- Nicholas
