Seize the (intra)day: Features selection and rules extraction for tradings on high-frequency data

Trading rules for high-frequency financial data based on hybrid clustering

“…At the same time, verify the validity of the SOM&K-means based trading rules algorithm in actual stock trading, this section also compare it to the traditional B&H strategy [12] and trading rules which are proposed by Resta [11].…”

“…In order to verify the validity of SOM&K-means based trading rules, we use statistical indicators and financial indicators which are used in documents [11]. These indicators are listed in Table Ⅱ, Table Ⅲ.…”

NPL disposal treatment in the LGD estimates

“…Thus, the dependent variable is represented by a vector  composed of (n+m) rows, where every row assume the value 1 in the case of a fraud event, and 0 in the case of a legal transaction. It is a question of finding those relations between (current and lagged) data included in  and  that define the multidimensional hyper-surface (not necessary a hyperplane) that divides the estimation space between fraud events and legal transactions, as every transaction can be related to the actual observable parameters (including time, IP address, 13 browser, operating system, HTTP header, 14 etc.) corresponding to an event.…”

“…In questa sezione vengono presentati sinteticamente i principi di funzionamento delle reti neurali dinamiche e le relative procedure di addestramento [15], [1]. Successivamente, vengono presentate le misure di performance impiegate per comprovare la qualità della previsione in applicazione al back-testing finalizzato alla validazione della rete [14]. Per approfondimenti teorici più specifici inerenti le reti neurali statiche e dinamiche, si rimanda alla letteratura scientifica di settore citata in bibliografia [4].…”

An Object-Oriented Bayesian Framework for the Detection of Market Drivers

“…Table 8 shows the performance measures used to asses the goodness of the trading signals in each period. Their selection is driven by the choices documented in Dempster et al (2001) and Resta (2009). To ease the readability of the table, please note that all the performance measures assume working on time-series of length ν = n − 1 whose corresponding mean value is denoted byμ; when necessary, it has been made use of the indicator function:…”