I recently spent around $90 on a book about candlesticks called High Profit Candlestick Patterns by Stephen W. Bigalow. I was very excited someone was finally going to show me all the benefits of these mysterious creatures called Candlesticks, but as I read on I realized his book had nothing to do with Candlesticks at all. Rather, it had to do with chart patterns. You see, I am a systems trader, and I code everything I hear about and read. I can’t help it; I have done that nonstop for close to 20 years. So, when I read something about things I have already tested, but called something else, then I know it isn’t that thing necessarily that leads to any success, it is the pattern itself that gets coded, not the name.
Stephen Bigalow is a successful trader as I understood at the time I bought the book. I would not have bought the book if he was not. In fact, I won’t even buy a trading book unless I have reason to believe (and verifiable reason) that person has done something remarkable trading. So I will listen to what he has to say.
As I went on reading, the text stressed how to use different patterns such as a Doji, Hammer, Hang Man, Harami, Shooting star, Morning Star etc. in conjunction with different moving averages and oscillators and other technical trading tools. If you have followed my other articles on oscillators and moving averages, you already know where I stand with these technical indicators. Does the existence of a Doji in some relation with a moving average portend a successful trade? Can a Doji (or any Candlestick pattern for that matter) be defined in some rigorous terms, or is it some funny notion that one cannot really define?
A Doji on a chart looks like this:
It is defined as a bar in a chart where the close and the open of the bar are very close together.
Now, here’s the thing. The open and close could be at the bottom of the bar, or at the top of the bar. Like this (in the bottom of the bar):
Now this particular bar is fairly short in height. It could be much taller. This bar is also a bit thicker (the difference between the open and the close is a little bigger than the previous image). What would happen if that thickness started getting thicker by any degree?
Well then that is called a Spinning Top. A Spinning Top then, is a Doji that is a little bit different. What specifically that difference is is not disclosed. It is a mystery.
None of these relationships seem to get too concerned with the amplitude of the bar except on two or 3 bar Candlestick patterns. Now, this article is not intended to be an exhaustive look into Candlesticks, because I could go on and on about where one candle relationship blends into another. If I am going to truly test such things, the computer will require I define what I am testing. For candles, this will be difficult.
I think the problem with candles as a study is the name and all the mystery turns it into something it is not. If I really want to define a Doji, I would have to do it by saying, a Doji is a bar where the close is within X percent of the open of the bar. Now if I fail to define it further, then I will not be differentiating, as we had discussed between the Doji and the Spinning top. As a result, it might be good to define that the Doji could, for example, only allow the open and close to be between the upper quarter of the bar and the lower three quarters of the bar. We might go one step further and define how much variance we would allow for the Doji to have between the open and the close as a function of the range. This could go on and on until we had completely defined the Doji mathematically. From that starting point, we could begin to test the efficacy of a Doji in various circumstances.
Let me address another topic while we are here in this important area that puzzled me for many years about technical analysis. If I create a 5 minute chart and have a Doji on the chart, by simply shifting the time one minute forward or back, it would create, in most cases, a completely different Candlestick pattern.
If I had put so much effort into defining the pattern only for it to be dramatically changed by a subtle shift in time, then what, I must ask myself, am I really defining?
I love the notion that patterns in data that can reveal what is going on in a larger time frame. I believe there is value in this, but unfortunately, such patterns become quickly meaningless when contextualized as they are with Candlesticks.
There are many two bar Candlestick patterns that are said to be indicative of a market turning up or down. The problem with these is the charts that point them out, have the same darned patterns occurring all over the chart in places that did not turn at all. So it becomes a matter of picking out what you want to see for that specific case. After working with computers for many years coding such things, any glimpse of hopefulness you might have in your eye is quickly extinguished when it comes to such scrutiny.
It is said that when certain Candlestick patterns show up, and a close occurs above a previous bar high, for example, that is a Candlestick buy (as in the third bar up from the bottom in the image below) . This is such a (very bullish) pattern:
Of course what really happened in this case is the market continued abruptly lower after a short rally upward. But here is what I really want you to see. On a shorter time frame, the chart really looks like this:
As you can see, the pattern of one set of Candlesticks from the above, longer term Candlestick chart, just blended into a whole other set of different candles on this shorter term Candlestick chart. To the left of center a bit, we see another buy before continuing lower.
You may wish to dispute my claim Candlesticks as a method of technical analysis are questionable (particularly) if not better defined. That is up to you. I do not intend to insinuate they are not of any value (afterall, they do help me to see my charts), but one should certainly not accept the conventional wisdom when it comes to these mysterious creatures (and their patterns) we call candlesticks without sufficient investigation.
The book High Profit Candlestick Patterns goes on to cover all kinds of areas from options to wave counting to Fibonacci retracements, support and resistance, moving averages, gaps etc. all in the context of candles. I am not really sure (actually I am) it would (not) hold up to the mathematical scrutiny of being coded into a computer even though it is all important stuff to know.
Maybe on a future article, I will dig into my old code and rigorously test some of these patterns using moving averages and Candlestick patterns associated with other technical indicators.