Last week the S&P moved a total of 0.4%. That certainly looks like an indecisive market. The believers in candlesticks call that a doji, when the body of the candle is tiny compared to the full bar. They would argue that it signals a turn in the market. I would love to believe, but I have to test, it is my training.
Here is a chart of SPY, the second to last candle is last week’s. (Click on the chart to enlarge)
Definition of Doji
So I took weekly data for the SPY, as a proxy for the S&P 500, back to 1993. As a definition for a doji, I took that the closing price – opening price should be less than five percent of the high price – low price for the week. That way the body would be small compared to the entire bar.
Test of the Predictive Power
I tested by comparing the close of the week before the doji candle to the close of the week with the indecisive market; and the week after the doji. If the week before was lower than the week of the doji so that it was an up trend, the next week should be a down week if the candle predicts a change of market direction. Similarly, if the week before the doji was higher, than the close of the doji week, the week after should be higher. It should look like V with the doji at the bottom of the V.
There were 41 dojis since the last week in January 1993, for this weekly version of the test, there is no predictive power.
But that is not how the eye would judge the trend. You wouldn’t look at the nearby week on either side of the doji, you would look at a few weeks on either side of the candle to judge the trend. So I redid the comparison with the close four weeks on either side of the doji.
Of the 40 cases, 26 of them were in the same direction and 14 signaled a change in direction. The reason that there are 40 instead of 41 is that the doji two weeks ago is too recent to count, since I am looking four weeks on either side.

0 Responses
Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.