[pcabc]

The Stages of Stage Analysis are defined in more of a descriptive than a simple mathematical manner, with entry to Stage 2 being from a breakout from Stage 1 and entry into Stage 4 on a breakdown from Stage 3.  At best we can make an algorithm that identifies a reasonable approximation to the stages.  Picking breakouts for humans is easier than for computers - as typically for the computer to do so we need to plot a channel (e.g. a Donchian channel) and identify a breakout from it.  However, what is the approrpiate time period for the channel varies from chart to chart.  I've ignored channels and used an algorithm that works something like this...

I take it that you haven't seen the Stage Analysis Technical Attributes (SATA) indicator that I created on Tradingview, which gives a good automated guide of the Stages, and uses a scoring system based on 10 key characteristics from the method, and was expanded from my original work on the SATA concept 10 years ago when I first launched the site, when I was looking for common characteristics of the Stages.

I saw your SATA system a few years ago, I'd not picked up on your latest revision or the links to the other sites, thank you, appreciated.

Over the years I've written a number of routines to estimate stages.  They seem to have fairly similar results and I think you rapidly hit a wall of diminishing returns as you add complexity to try to refine them, especially with respect to picking out Stage 1 & 3 boundaries.  For breath calculations, based on each stage, you have to calculate the stage for every stock in the pool of stocks for that breadth data.  Adding complexity slows it down.  For the purposes that I'm calculating breadth for I think my Stage estimator is good enough.

   

I note I could probably speed up my coder considerably with a major re-work.  However, the speed up would be when undertaking bulk calculations on historic data. I suspect this would make little difference when just updating the latest figures for the last day or two.  So I'm probably better off being patient and just being patient when I'd doing bulk calculations on historic data.


I note your quote from the Tradingview article you posted:

There's no easy automated way to define the four stages, but by using a scoring system is one of the most accurate ways that we've seen, especially at the Stage 2 breakout and Stage 4 breakdown points. But it is still only a rough guide, as you still need to use your eyes to define the key levels.

Obviously it would be madness if one were to manually cross check the stage when making a breadth indicator!  Not that you are suggesting that.




However, you also mentioned a screener.  My screener does not rely on may stage estimation, but it does rely on various parameters including price in relation to moving averages, breakouts, volume action etc.  So it is picking up the various factors in identifying a strong Stage 2 breakout.  It produces a score and sorts stocks by that score.  Noting that, as well as doing an overall screen it is worthwhile running screens on specific strong sectors as despite having put a lot of work into it, it sometimes gives lower scores to generally good charts.

I'm currently looking at whether the % stocks which have the Mansfield Relative Strength above the zero line and whether the zero line is rising is a useful breadth measure.  These additional calculations are very slow and I'm trying to work out whether I have bugs in my code or whether this is not a very useful parameter to measure?