Transformations  in
Business & Economics

PARETO OPTIMISED MOVING AVERAGE SMOOTHING FOR FUTURES AND STOCK TREND PREDICTIONS
Aistis Raudys, Zidrina Pabarskaite

ABSTRACT. The most common way to forecast the trend direction is to use moving averages. It is very popular in finance. In this study a Pareto optimised custom moving average is suggested, as it is the most suitable for financial time series smoothing. Suitability criteria are defined by smoothness and accuracy, the criteria often used by practitioners. Previous research has mostly concentrated on only one of the two criteria in isolation. We define this as the multi-criteria Pareto optimisation problem. The essence of the proposed method is weight optimisation, so that for every level of smoothness we obtain the best accuracy. We compare the proposed method to the five most popular moving average methods on 1000 synthetic and 2000 real world stock data on smoothness levels equivalent to the smoothness of simple moving average of 5, 10, 21, and 63 days. The comparison was performed using out-of-sample, unseen data. Weights optimised on one stock are very similar to weights optimised for any other stock and can be used interchangeably. The new method outperforms other methods in the majority of cases. It allows better time series smoothing with the same level of accuracy as traditional methods, or better accuracy with the same smoothness. Traders can use the new method to detect trends earlier and to avoid unnecessary trading and increase the profitability of their strategies. The concept is also applicable to sensors, weather forecasting, and traffic prediction where both the smoothness and accuracy of the filtered signal are important.

KEYWORDS: moving average filter, smoothness and accuracy, weight optimisation, triple smoothed exponential moving average (TSEMA), custom moving average.

JEL classification: G10, C61, C22, C63.