This paper presents an efficient method for enhancing time series load flow analysis, accounting of variabilities inherent to modern power system, especially in presence of renewables, electric vehicles, and storage systems. The method uses a moving block bootstrap technique applied on time series residuals, after Box-Cox transformation and Seasonal-Trend decomposition based on LOESS (STL). The adopted techniques, already known in statistics for improving time series forecasting, can generate new observations of system variables, such as load, generation, or transmission line power flows, by resampling from the STL model residuals of observed time series data. The proposed approach has the advantage of avoiding a full probabilistic analysis, which can be computationally expensive and data-intensive. It can generate distribution functions of electrical parameters at any time point within the observation period, calculate confidence intervals of relevant statistics, and assess probabilities of critical system anomalies (such as overload or voltage violations). These are all essential for strategic planning decisions. The suggested methodology, despite its inherent limitations, can efficiently deal with uncertainties in the context of power system analysis. It proves particularly valuable in cases where traditional probabilistic analysis becomes unfeasible due to inadequate data or computational constraints.