A Flexible Approach Towards Multi-frequency Re-engineering of the Moving Average Convergence Divergence Indicator

Abstract

The study develops an innovative and flexible methodology for re-defining the traditional convergence–divergence indicators in the light of multi frequency trading behaviour of the heterogeneous agents. The developed indicator is labelled as multi-resolution convergence divergence indicator (MRCD). In contrast to the traditional moving average convergence divergence (MACD), the MRCD is “flexible” as it reacts to fluctuations arising at any frequency interval and is thereby capable of adapting to a wide variety of future possibilities. The “innovative dimension” underpinning this methodology is the replacement of the traditional trend extractor (moving-average) with a more novel methodology—the multi-resolution analysis. The forecasting ability of this newly engineered indicator is examined by structuring a neural network based MRCD–NARX model. The performance of this model is bench-marked against that of a similar model developed using the traditional MACD indicator. Out-of-the sample mean square error and the Diebold–Mariano test are used to examine the statistical accuracy of the forecasts. The profitability of the indicator is ascertained using the correlation measure and the hit ratio. A “long-short trading rule” is developed and back-tested on the testing data-sample to validate the practical applicability and “reproducibility” of the methodology.