Technical analysis is one of the most frequently used analysis and trading methods in the markets. In academic research, on the other hand, this form of analysis has often been and still is ridiculed and accordingly neglected. Although some models make certain assumptions about so-called "noise trading", they fail to capture the true essence of technical analysis.
The study "The Interaction Between the Aggregate Behaviour of Technical Trading Systems and Stock Price Dynamics", on the other hand, describes a model that takes rational technical analysis and its interaction with stock prices into account. 
The starting point for these considerations is the assumption that technical trading is an integral part of market activity. In this function, it can cause an excess of supply or demand if different strategies produce corresponding clusters of rectified signals. Thus, initial price movements, which are triggered by news, for example, can be reinforced by sequences of trades based on trend-following strategies. If there are many signals in the same trading direction, a (destabilizing) surplus of buy or sell orders can occur. If this is the case, a feedback process occurs between the movements of the stock prices and the signals or transactions of the strategies: If the prices rise, the technical models increasingly produce buy signals (and vice versa).
Trading signals are not always exogenous
Stephan Schulmeister examined a total of 2580 trading strategies. Each of these models, which indicated a long position, was rated +1, each short position -1 and each neutral rating 0. Then he calculated a net position index every 30 minutes from the sum of these numbers across all strategies. In this way, he was able to track the aggregated behavior of the strategies over time and compare it with the price development. He also examined whether the signals of the various models offset each other by analyzing the number of new long and short signals for each 30-minute interval.
On the basis of his investigations, he arrived at the following findings:
Often the majority of the signals are on the same side of the market (long or short). The aggregated indicator is hardly ever in the zero line range for a longer period of time, which would be expected with a random walk.
The process of changing existing positions in response to a new price trend usually begins one to three periods (here 30 minutes each) after a local high or low. If the new trend continues, it takes 10 to 20 periods until the positions of almost all strategies have turned from long to short (or vice versa)
Once 90 percent of the technical strategies have given an appropriate signal, prices tend to move towards these positions. If the movement loses momentum, counter-cyclical technical strategies contribute to reversing the trend.
What is particularly exciting is the conclusion that the individual models hardly balance each other out. The author writes that, on average, only 2.3 percent of all the strategies examined trade with each other, i.e. trigger opposing signals at the same time.
Interaction of trend and signal
The procedure for a technically driven trend continuation is as follows - in case of an upward trend: