Weekend Charting and Chat - 21st October 2016, page-5

@paddington bear
Dragonfly study
I’ll start the post with a very quick discussion on developing an “edge”. IMO the only reason we do this “stuff” (TA, FA or dart boards) is in search of an edge …. So what is it?

If you toss a coin 10 times logic … and statistics …. says you should get 5 heads and 5 tails. Easy.
If you get 7 heads does it mean you are GOOD at tossing heads? Have you discovered an EDGE? Maybe, but I doubt it.

If you toss a coin 1,000,000 times and you get 700,000 heads does that mean you are GOOD at tossing heads? I suspect so. You have an EDGE (or a dodgy coin?)

So what? So I propose that so much of what we do in TA is dominated by a Confirmation Bias.

Whaaattttt; confirmation bias; noun:
the tendency to interpret new evidence as confirmation of one's existing beliefs or theories.

So confirmation bias is what I am always concerned about in charting; recently @paddington bear and I have been chatting about dragongly doji’s …. for a long time I have kept an eye on doji's generally and their ability to predict turning points .... never traded them tho'

Whats a dragonfly doji? Investopedia: “A dragonfly doji is a type of candlestick pattern that signals indecision among traders. The pattern is formed when the stock's opening and closing prices are equal and occur at the high of the day. The long lower shadow suggests that the forces of supply and demand are nearing a balance and that the direction of the trend may be nearing a major turning point.
Source dragonfly-doji

Interestingly Bulkowski does not rate the d’fly very highly; see here Bulkowski Dragonfly

My observations over the years rate it much higher than Bulkowski and instinctively a day where the Bulls overtake the bears and pull price back after a loss has Bullish all over it …. Like a one days VSA test!

Whats my interest?
…. Trying to determine how reliable they are in predicting a market low. Is there a set of conditions that could reliably trigger a buy on the open after a dragonfly? Can I get an "edge"?


So I set up scan for dragonflys in PRT; looking for d’flys that are within a down trend (hopefully predicting the end).

Here is Fridays scan:



I have now got 7 days of scans (see xso thread for previous scans) and have identified 23 shares that met my criteria. Each day I have checked the charts and done a “viual” to see whether it meets certain criteria (its hard to program every criteria so I just write “Like” on the ones that meet the downtrend criteria to my liking and/or if there are other features that “appeal ). I “liked” 12 of them and have now done a check of how they performed to date;

Here are the results (they are a screen shot of an excel spreadsheet):




And here are 2 of the more interesting charts …. To my eye ….




I intend to keep assessing these for a while and will report any interesting outcomes …. In the coin tossing analogy I think I have 6 heads from 10 tosses at this stage…. Need much more data!

Any comments welcome. (or mention any addition dragonflys that I might have missed and you want tracked)

PS I have a similar process for tracking evening star doji’s as a predictor of tops …. Another day if any interest

Cheers g