Rafał Kowalski, Paweł Oświęcimka, Robert Gębarowski, Stanisław Drożdż

For today’s investors electronic trading platforms together with algorithmic trading constitute “natural environment” for carrying out efficient and transparent transactions. Advanced technology has displaced traditional trading floor makes the trading available not only for professional traders but also for retail investors. Moreover, especially in the case of high-frequency trading, many transactions are executed not by human but automated trading systems using complicated algorithms of trading. These technology innovations are also core driving force of coupling between markets from distant geographical locations and developing globalization of financial markets. However, the algorithmic trading can also carry some serious risk for investors. The best example of the danger related to the automated electronic trading is the stock market crash in 1987 where electronic technology is considered to be one of the main factors responsible for the index value plunge. This, in turn, forced the introduction of some financial regulatory instruments protecting investors against possible dramatic losses.

Natural question which arises in this context is the impact of the technology revolution and related trading regulations on the stock market dynamics. In this contribution we used multifractal formalism to rolling-window multiscale analysis of the indices values coming from the major world stock exchanges. We showed that the stock market dynamics changed significantly in the second half of the 80s which coincides with the period of growing popularity of the electronic trading. Distortion of the hierarchical organization of the data is clearly visible on the level of multifractal spectra which are characterized by strongly asymmetric shape for the period after 87 in contrast to almost symmetric shape for the earlier one. Heterogeneity of multiscaling properties related to the asymmetry of the multifractal spectrum indicates imbalance in fractal complexity between the price fluctuations of different amplitude. It has been demonstrated that in the era of electronic trading large price fluctuations reveal much more pronounced hierarchy then the small ones which can be considered as a kind of noise. Moreover, temporal variability of the multifractal properties of the financial data offers the possibility of its practical application. In particular, our findings suggest that this approach can be potentially useful in predicting stock market crashes.