The often reported empirical success of trend-following technical timing strategies remains to be puzzling. In previous academic research, many authors admit some prediction power but struggle to substantiate their findings by referring vaguely to insufficient market effciency or unknown hidden patterns in asset price processes. We claim that empirical timing success is possible even in perfectly efficient markets but does not indicate prediction power. We prove this by systematically tracing back timing success to the statistical characteristics of the underlying asset price time series, which is modeled by standard stochastic processes. Five major impact factors are studied: return autocorrelation, trend, volatility and its clustering as well as the degree of market efficiency. We use trading rules based on different intervals of the simple moving average (SMA) as an example. These strategies are applied to simulated asset price data to allow for systematic parameter variations. Subsequently, we test the same strategies on real market data using non-parametric historical simulations and compare the results. Evaluation is done by an extensive selection of statistical-, return-, risk-, and performance figures calculated from the simulated return distributions.