Sílvio M. Duarte Queirós

Partially aiming at having larger datasets in complex systems analysis, we are frequently pushed to either consider the data quasi-stationary or apply some filter and jettison nonstationarities. Nevertheless, that sort of dynamical evolution can contain important information about the system under study as well as it is now understood that the analysis of higher-order statistical moments is crucial for an accurate description, namely when we are talking about complex systems.

In my presentation, I will revisit the classical problem of intrady seasonalities in financial trading introducing a new perspective: the analysis of higher-order moments and the assessment of the nonstationarity of those statistical features.

The results of that analysis allow understanding to what extent the market has modified its tradingdynamics since 2003 and how changes in regulation (namely short-selling/uptick rules) and the 2008 subprime crisis have affected opening and closure behaviour of market agents.

Moreover, I show that the morning part of the trading session is dominated by the trading of a couple of stocks that input information and volatility in the market whereas the second part of the business day is ruled by significantly more concerted behaviour between stocks.

Last, I will lift the veil on the impact of liquidity on these results, especially regarding structure of the correlation matrices.