Cycle Trading

[JLS]

Gravestone Doji,

The following, in italics, are Bulkowski's words:

Many people look at a gravestone doji with alarm, especially during an uptrend. They believe that it signals a bearish reversal. It does, but only 51% of the time. I call that random. The trend after the reversal does not amount to much, placing the performance rank at 77 out of 103 candles where a rank of 1 represents a trend that really moves.

I agree with that. Then he goes on to say:

There are exceptions, of course, and a gravestone doji can signal a lasting trend change. The analysis I conducted (see my Candlestick Encyclopedia book) highlights several tips to identify gravestone doji candles that outperform.

http://www.thepatternsite.com/Gravestone.html

In his first paragraph he labels the doji's value as random then in the second paragraph says there are exceptions. I'm not exactly sure how to chew on that.

I used to communicate frequently with Bulkowski via email quite often, but I don't anymore. My reason is that, even though we both come from an engineering background (he in software only, and mostly management, and me in both software and hardware) he does not follow the scientific process all the time, and he introduces errors when he does (such as in things that are random but have exceptions).

For instance, he discusses NR4 and NR7 patterns and their relative value at his web site. He likes the NR7 pattern quite a bit, but not the NR4. I was also very interested in those patterns as general market health indicators but I could not replicate his NR7 results (so I didn't even try his NR4 pattern). So I sent him some data along with my results and asked what my problem was. Well, this was the first time he took a long time to respond so I assumed he was having trouble figuring out what the problem was. When he finally responded, he said that in his counts of days he was counting weekends. Well, that says it all. Therefore, I have to also assume he counts holidays. I've never discussed anything with him since that time.

As publicly exposed, there is a general problem with candlestick theory; and in fact, pattern theory of all kinds. For instance, when discussing candlestick theory, have you ever seen anybody numerically define the terms Large, Small, Long, and Short? I've done a lot of searching and I've never found any specific numerical definitions. Without specific numerical definition relative to some anchor, those terms are all worthless. Without those definitions there is no science. Without those definitions it's only religion.