CandleCode alive! .
Similarly, in the construction SandleCode: in one category - for example, the average body - are all the candles, body size, which range from the lower to the upper threshold. As with the P & F, most of the market movement is not different, then there is a more rough idea of the graph. It is clear that the choice of threshold can not be arbitrary, in every market, these thresholds will be his. To make informed choices of thresholds necessary to analyze the size of the candles. Initially, I used the histogram dimensions of bodies and shadows, built on certain time intervals (for hourly charts currency used at intervals of 4 weeks).

Incidentally, these histograms show that the size distribution of candles are not even log-normal, and perhaps even worse - exponential. The rapids on the histogram I was setting equiprobability reasons: to about 33% of all candles have small bodies, 33% - the average body, 33% - more. Found thresholds fixed for some time to calculate indicators CandleCode. Then histogram built for the new 4-week intervals and were new thresholds.

The problem is not that periodically conduct a statistical analysis - it tiresome. Most importantly, it turned out that the resulting thresholds have very high resistance: they change little over time. Thresholds for different rates differed by 10-20%, and the change in the threshold of this magnitude almost no effect on the behavior of average indicators that are based on CandleCode.

This, incidentally, shows statistical stability (robustness) procedures based on average indicators CandleCode, which is a nice feature for financial decisions.

It became clear that the selection process should be automated thresholds for what I used the Bollinger band. Their use for the selection of thresholds seems to me a good solution. In itself, the calculation of the mean values and variances within the Bollinger band is not associated with any specific properties of the normal distribution, they are suitable for the analysis of any measurement and usually give a good estimate.

If a trader wants to get the best fit indicators CandleCode its market, it can create histograms for selected plots and for them to find the optimal thresholds. This setting indicators for a particular market, would make it possible to obtain better results in the trade. However, the automatic selection of the threshold is universal, suitable for all markets, and in this form CandleCode work that has already proven!

Coding groups of candles

There was a note: in CandleCode considered only a single candle, while it is necessary to take into account neighboring candles. It is in the analysis of groups of candles - the essence of Eastern wisdom. With this, I really agree. CandleCode assigns a numerical weight of a single candle. For this purpose he intended. The justification I can give two reasons.

First, to build the code of one candle LED Candle-Code still analyzes the market for quite a grand historic time interval (window averaging Bollinger). Therefore, in the code of each candle laid much statistical information about the market.

Second, coding groups of candles - it's just another challenge. Present a number not a candle, and a group of candles - even more complex problem than weighing isolated candles. Perhaps, CandleCode is part of the solution of this problem, maybe someone will go a completely different way. Be that as it may, CandleCode corresponds to the purpose for which it was created. And averaging, or otherwise combining codes neighboring candles, you get a description of it for some groups of candles!

And this objection opponents were: quantification of candles - too difficult, even the Tushar Chande not tackled it!

Now nothing can be done. While working on CandleCode I was not known to the book The New Technical Trader, but later I met with the proposed indication there Qstick (Quantitative Candle-Stick), which is the average of the difference between opening and closing prices (close-open).

Of course, as a numerical representation of the candle the light is impressive. It's like that in Figure throw out scenarios covering everything that does not fall within the range of the body (between horizontal lines drawn through the open and close). Important piece of information will remain (candle body), but the picture of the market as a whole will be lost.


Measure the number one market sentiment, his psychological motives inherent in the graphic picture of the behavior of the market - a very difficult task. We could not agree more. This problem is not the sole and final decision. Different approaches here have a right to exist, and research into the candlestick chart analysis, no doubt, will continue. Coding algorithm based CandleCode and its variants are as simple as possible in a difficult task. And as derived indicators clearly able to generate the desired signals, they deserve to be studied and used in more than further complications. Over the years CandleCode indicator and trading systems based on its basis have been implemented staff from software developers. CandleCode indicator and trading systems based on it are now available in these packages as Equis MetaStock, Omega Research Trade Station, NeuroShell (Ward Systems), TradingSolutions (Neuro-Dimensions), Wealth-Lab, Investor / RT, TechniFilter Plus. It's safe to say that they have already been tested in various financial markets.