Doji Lines and the PSEi

Of all the trading literature I’ve read, perhaps the most colorful and evocative terminology comes from the field of Japanese Candlestick Theory. This is because of its distinctive oriental jargon and near-mystical quality in describing and predicting market movements.

There are many critics of Candlestick patterns, primarily because it attempts to formulate insight on market trends based on very few inputs–most candlestick patterns rely on market action for the last 3 or less data points on a chart (3 days, 3 weeks, 3 months). Logically, this would have a tendency of oversimplifying developing market action, and draws weighty conclusions on very little basis.

I will not go at length to defend or critique all of Japanese Candlestick Theory here, however I brought up the topic because of the next pattern I will discuss, which also just happens to be a popular subset of Candlesticks. In general, we all know that prices do not move in a linear fashion, but instead move in rapid bursts and spurts, trends and consolidations. Perhaps more predictable than price direction, the more certain thing you can say about markets is they alternate between high and low volatility periods, between periods of certainty and indecision. If our assumption of this dynamic is accurate, then this presents an opportunity for traders. If periods of high volatility follow periods of low volatility–then identifying low volatility periods is a key to positioning ahead of a future trend in prices.

Here’s where Candlesticks are convenient because one pattern in particular may be exactly the indicator we need to identify these low volatility points prior to a trend direction. This is the doji line or doji star. I’ll quote the definition of such a pattern from StockCharts.com:

Doji are important candlesticks that provide information on their own and as components of in a number of important patterns. Doji form when a security’s open and close are virtually equal. The length of the upper and lower shadows can vary and the resulting candlestick looks like a cross, inverted cross or plus sign. Alone, doji are neutral patterns. Any bullish or bearish bias is based on preceding price action and future confirmation. The word “Doji” refers to both the singular and plural form.

Ideally, but not necessarily, the open and close should be equal. While a doji with an equal open and close would be considered more robust, it is more important to capture the essence of the candlestick. Doji convey a sense of indecision or tug-of-war between buyers and sellers. Prices move above and below the opening level during the session, but close at or near the opening level. The result is a standoff. Neither bulls nor bears were able to gain control and a turning point could be developing.

You can’t get any more descriptive than that. However, this is where the art of candlestick reading takes over–because like all other visual chart patterns, identifying and interpreting the pattern can be a subjective exercise.

To convert the art into a science, we need objective criteria:

• First a criteria to identify the pattern in question.
• Secondly, a hypothesis on the results to be expected after the pattern has been identified.

The whole idea behind my introducing the idea of a doji has been to form an anchor to test our assumption of market dynamics alternating from periods of indecision to trending periods. If a doji can be said to represent periods of indecision, then we can expect the market action following a doji to be trending action.

But again, outside of sharp eyes, how do we identify dojis objectively? Simply sticking to the definition, we could just select days where the open and close are equal–but we already know that this ideal situation rarely happens, so we make an objective allowance. For the purpose of this exercise, let’s identify dojis as days when the difference between the open and close for the day is less than 0.2%. Then lets get all those days and examine the market action after them to see if the doji according to our definition can really be said to identify key points prior to a trend.

Quickly scouring the PSE Index data since 1997 yields some interesting examples:

On August 4 1997, the market opened and closed at 2,638 and 2,639 respectively, forming a doji according to our definition. The next day, the market opened higher at 2,642 and gained 30 points for the day.

On August 27, 1997, the market opened and closed at 2,286 and 2,284, forming a doji. The next day, the market opened lower at 2,262 and lost 190 points by the close.

On October 3, 1997, the market opened at 2,031 and closed at 2,029–practically unchanged and forming a doji. This happened on a Friday and after the weekend break the market opened lower at 2,023 and lost 62 points by the close.

On July 1, 1998, the market opened at 1,780 and closed unchanged at 1,781, forming a doji. The next day, the market opened at a gap at 1,819 and gained 36 points by the close.

On November 29, 2007, despite having a wide 100 point range for the day, the market opened at 3,581 and closed unchanged at 3,578 forming a doji. The next day the market opened up at 3,604 and gained 20 points by the close. 3 days later on December 5, the market opened at 3,644 and closed unchanged at 3,648 forming another doji. On December 6 the market opened up at 3,676 and closed 56 points higher. 2 weeks later on December 11, the market opened at 3,679 and closed slightly lower at 3,672. The next day the market opened at 3,632 and lost 26 points by the close.

December 7 in the middle of the last example (encircled in grey) also preceded large price action, but did not qualify as a doji according to our definition (0.2% open vs. close).

The examples are quite interesting right? Although we’re not forming any ideas about predicting direction, it does seem evident that dojis which signify market indecision, can be good anticipation points for a strong move the following day.

Let’s test the idea, which we will formulate this way:

• Identify doji lines where open and close are less than 0.2% apart.
• Expect a significant move in the market the day following the doji and the opening of the following day will determine the trend for that day.

Simple enough. So we now scour our data for all days qualifying as dojis under our definition and check the market return the next day based on the opening price the day following the doji:

Of the 2,724 days tested, 426 were dojis according to our definition. On days following these dojis, if the market opened down, those days cumulative lost 1,980 points. The winning rate for these down days are a dismal 33% and the expectation decidedly negative. On the days following dojis where the market opened up, these days cumulatively earned 1,737 points with a 62% winning percentage, and a positive expectancy. Both cases contrast sharply with all other days with no prior dojis, which cumulatively broke even.

Caramba those dojis are sneaky predictors!

To test the robustness of the opening price following the doji as a good predictor of the market returns that day (because we know the market loves to fool people), we can draw a correlation between the gain/loss at opening and the total gain/loss of the market by the close:

The distinctively upward sloping trendline confirms that the opening action the day following a doji is positively related to the close of the market that day. The correlation in this case is 0.62 which shows a strong tendency for the opening of the day to indicate the closing direction.

Insight and Application

Without realizing it, what we are actually doing here is already going into the micro minutiae of market structure. Spotting periods of indecision (as signified by the doji) is a powerful tool to identify crucial points prior to an explosive move by the market. Market prices are the perfect indicator to gauge buyers and sellers intentions. Whenever the two opposing forces are equal, this invariably leads to periods where one party gets the advantage.

Short-term and long-term traders alike can benefit from doji patterns. Short term traders can base their entries and exits on market indication immediately after a doji to capitalize on the tendency of the market to trend shortly after. Longer-term traders can use market action after doji points to confirm if their original bullish or bearish bias in the market is changing. Since the market tendency is to break away from the original doji level, a doji can be a good base point to add positions or cut losses.