XJO 0.33% 8,045.1 s&p/asx 200

Hi Martis, thanks for taking the time to do some comparative...

  1. 8,720 Posts.
    lightbulb Created with Sketch. 81
    Hi Martis, thanks for taking the time to do some comparative charts to explain your approach.

    I do agree with you that drawing a trendline from as many data points as possible is desirable - the higher the number the more powerful the trendline becomes in importance. However it is very clear that we chart using different methods.

    I also agree those points you have circled should not in "no mans land". Many of those points would be used either for:

    1. Drawing the highs and lows of waves in a shorter term EW formation (as per example by Sandash yesterday)
    2. Drawing shorter term or intermediate impulse & reaction trendlines during the course of that longer term trend
    3. Drawing an Andrews Pitchfork version of the long term trend, where many of those points would form the middle tyne of the pitchfork

    Unfortunately your "best fit" trendline drawing method ends up (imo) being more arbitrary than I would prefer in my analysis, i.e. the ability to interpret any given candle in relation to a trendline that is best-fit to non-exactly aligned points amplifies the level of interpretation required to analyse its meaning. For example, if the most recent candle closes close to the trendline, would it be just above or just below the trendline if your arbitrary trendline placement were slightly higher or lower than where it is.

    On the other hand, even if only two data points were used to draw a trendline, such as the March 2009 and October 2011 lows, but the trendline is aligned exactly to those two points, then it is very clear whether a given candle might close above or below that line - because their is no dispute about the positioning of that line.

    In other words, while I agree with you that it is desirable to align a trendline to as many data points as possible, if you start to include data points that aren't exactly on that line, then you introduce arbitrariness to the positioning of that line, and the downside is you start losing precision in your analysis with each additional non-exact data point added. In science this can be overcome by using for example a least squares regression type technique, and you will get a "best fit" line with an associated "R squared" correlation factor somewhere between 1 (exact correlation or alignment) and 0 (no correlation or alignment).

    So on the one hand it is desirable to obtain precision in charting by aligning to only data points that fit exactly, on the other hand it is also desirable to obtain a degree of strength from a trendline by aligning it to as many data points as possible even if not exactly. Unfortunately there is a trade-off between those two goals. Clearly I prefer precision in my trendlines whereas you appear to prefer a regression-type "best fit" picture.

    The reason I prefer the former approach is because it offers an ability to assign more meaning to an individual candle, like last week's (or this month's), whereas using the "best fit regression-type approach it is more difficult to do that - i.e. is a given candle's position just typical "noise" around the best-fit trendline or is it actually outside that - really difficult to tell (imo).

    And so according to my methodology I can at least say that:

    1. This week's weekly candle on the DJI clearly closed above that March 2009 trendline, on very strong volume (buying support) - see weekly chart below

    2. That this month's candle, so far (with one day to go) looks likely to close above that trendline.

    Now the above commentary notwithstanding, no single candle is definitive even if precise, continued confirmation is always desirable...so as you always say..."we await".

    2015-08-29 DJI weekly.png

    Cheers, Sharks.
 
watchlist Created with Sketch. Add XJO (ASX) to my watchlist
arrow-down-2 Created with Sketch. arrow-down-2 Created with Sketch.