In one of Niederhoffer's books he points out that dojis occur far more often than chance would predict. In my judgment, dojis tend to occur at turning points in stocks, fwtw. But here is the hypothesis: dojis occur because a buyer has a mandate to buy UP TO the point at which the stock opened. In other words, if the stock is 5c below the opening price, the trader can buy the available shares up to the opening price, which results in a doji. This is a bullish formation because it is the result of a buyer trying to conceal the buying activity/not wanting to stick neck out/not bid against himself etc.
A similar phenomenon occurs in real estate, here is a post from a real estate forum: A seemingly artificial price ceiling for a neighborhood is a very interesting phenomenon in the real estate market. Logically speaking one would assume that if a house is selling for $X than a supperior house with an extra bathroom or a bit more space should be able to sell for an appropriate premium of 2% to 5% depending on how much ânicerâ it is. However in reality neighborhoods tend to form a ceiling which properties have trouble breaking through. This is a perfect example of that phenomenon.
more scientifically stated and dealing with market utility, a doji level on a slower fractal is a sentiment indicator on all faster fractals. Please do not confuse TF with fractal. I use this concept on all turns to indicate : 1. the type of trend (one of four exclusive types), 2. the type of turn (one of thre exclusive tpes), and 3. what action card (a thru f below) to use at that moment to continue to take full offer of market. the action types include: a. going "IN". b. early entry c. Hold Thru d sideline, and e. Wait f. exit. If the established doji on slower fractal changes color (symbol for sentiment), it indicates a re entry into a trend not yet taken or a trend where profit taking was already maximized. What remains is the action called "opposite early entry"; that is not within the scope of ET contributions. All early exits (not the CW type but the scientific types) are also there and the doji is always "in line" with optimum profit taking at trend extremes.
Or one could hypothesize that the doji is simply an illustration of the efforts of traders to seek equilibrium. If the doji were "opened" in a smaller interval, this could easily be seen: price opens at a certain level, rises until no more buyers can be found, falls until no more sellers can be found, then returns to a level where the greatest number of contracts or shares can be traded. This may be in the median, or it can be just north or south of it. Whether this is illustrated as a single bar or as a series of waves depends on the interval being plotted: 1m, 5m, 30, 60, daily, whatever.
I don't understand how this paragraph differs from "stock prices follow a random walk." But that's not the assumption of technical analysis. Many charts are hard to distinguish from randomness, but not all. Often when there are embedded sellers or buyers at a certain level (the old high or low) then prices will observe those levels more than randomness would predict.