Ανάρτηση Ερευνητικών Δοκιμίων no 01/26 & 02/26

    Ερευνητικό Δοκίμιο no 01/26 με τίτλο "The Tell-Tale Clock! Speed and Agent Composition in High Frequency Trading"

    των Ιορδάνη Καλαϊτζόγλου και Στυλιανού Αρβανίτη

    Περίληψη

    We introduce an instantaneous, intensity-based metric for estimating the number, type, and probability of different agent-types over any time interval. It is founded on the idea that differences in trading motives are reflected in agent specific arrival rates, which are modeled using conditional hazard functions. This novel framework exhibits superior empirical performance and theoretical properties over its interval-based — and thus slower — counterparts. In addition, it enables the derivation of a local measure for HFT reflexivity that is found to be highly correlated with the presence of variance spikes and extreme events, such as flash crashes. Therefore, it can act as a propensity measure for HFT market ”stress”

    O Ιορδάνης Καλαϊτζόγλου είναι Kαθηγητής στο Audencia Business School και ο Στυλιανός Αρβανίτης είναι Καθηγητής του τμήματος Οικονομικής Επιστήμης του Οικονομικού Πανεπιστημίου. Αθηνών.


    Ερευνητικό Δοκίμιο no 02/26 με τίτλο "Sentiment Utopia Index: Rhetorical Structure and Corpus-Relative Financial Sentiment"

    του Στυλιανού Αρβανίτη

    Περίληψη

    This paper introduces the Sentiment Utopia Index (SUI), a functional and corpus relative measure of financial sentiment that explicitly incorporates rhetorical structure within texts. Each document is represented as a sentiment trajectory over normalized rhetorical time, and sentiment extremeness is defined as a weighted shortfall from corpus-level bullish and bearish envelopes constructed under real-time Information constraints. Rhetorical emphasis is treated as a structured form of model uncertainty and explored through a finite, interpretable family of parametric weighting schemes. An empirical application illustrates the use of the SUI as a predictive signal for daily realized volatility in a heterogeneous autoregressive framework using SEC filings and an OHLC-based volatility proxy.

    O Στυλιανός Αρβανίτης είναι Καθηγητής του τμήματος Οικονομικής Επιστήμης του Οικονομικού Πανεπιστημίου Αθηνών.