MODEL_PREDICTABILITY

Volatility Forecasting Evaluation Framework

Coordinatore	ATHENS UNIVERSITY OF ECONOMICS AND BUSINESS - RESEARCH CENTER    more  Organization address address: KEFALLINIAS STREET 46 city: ATHENS postcode: 11251 contact info Titolo: Ms. Nome: Maria Cognome: Marinopoulou Email: send email Telefono: 302109000000
Nazionalità Coordinatore	Greece [EL]
Totale costo	45˙000 €
EC contributo	45˙000 €
Programma	FP7-PEOPLE Specific programme "People" implementing the Seventh Framework Programme of the European Community for research, technological development and demonstration activities (2007 to 2013)
Code Call	FP7-PEOPLE-2010-RG
Funding Scheme	MC-ERG
Anno di inizio	2011
Periodo (anno-mese-giorno)	2011-11-10   -   2014-11-09

 Partecipanti

#	participant	country	role	EC contrib. [€]
1	ATHENS UNIVERSITY OF ECONOMICS AND BUSINESS - RESEARCH CENTER  Organization address address: KEFALLINIAS STREET 46 city: ATHENS postcode: 11251 contact info Titolo: Ms. Nome: Maria Cognome: Marinopoulou Email: send email Telefono: 302109000000	EL (ATHENS)	coordinator	45˙000.00

Mappa

 Word cloud

Esplora la "nuvola delle parole (Word Cloud) per avere un'idea di massima del progetto.

 Obiettivo del progetto (Objective)

'Predicting volatility is of great importance in measuring and managing risk more accurately. Volatility is measured, estimated and predicted by a vast number of model frameworks. Though, researcher has to select a specific model for his/her forecasting purposes. Thus, the models are evaluated in order to extract the most adequate model for forecasting purposes. Under the Marie Curie Intra-European Fellowships Grand, we define a method of evaluating the predictability of the models, which assumes that the distance between the actual volatility and its forecast is normally distributed. However, empirical applications provide evidence that the distribution of this distance is better described via a leptokurtotic and asymmetric distribution. Under this grand we intend to construct a model evaluation method that presumes a leptokurtic and asymmetric distributed distance between the actual volatility and its forecast. The development of a volatility forecasting evaluation framework, with assumptions closer to reality, is of great importance in producing accurate forecasts of risk measures (specially the last years with the deep crisis caused in financial sectors).'

Introduzione (Teaser)

An EU team experimentally compared various models predicting financial volatility. Assessing the models' performances in European stock and currency markets, the study, based on a volatility forecasting evaluation framework, was able to determine one particular model yielding the lowest error.

Descrizione progetto (Article)

Accurately managing risk, in financial markets, depends on predicting volatility. While many different models make such predictions, defining the appropriate framework of choosing the right model is challenging.

The EU-funded MODEL_PREDICTABILITY (Volatility forecasting evaluation framework) project compared and evaluated a set of competing models. Researchers worked to develop a model selection framework, incorporating the evaluations and method for evaluation, for use in volatility forecasting. The three-year project concluded in November 2014.

Team members defined a set of models for forecasting volatility, which included certain assumptions about errors' distribution. After calculating the errors for each model, the team selected the one yielding lowest errors. The group also confirmed that its methodology for evaluating error was suitable.

Models were next applied to estimating volatility of major European stock market indices, plus certain currency exchange markets. For each trading day, the models were re-estimated and their error rates compared. In general, a certain ARFIMA-GARCH model proved most accurate.

The project also provided training and career development for its research staff.

MODEL_PREDICTABILITY estimated the volatility forecasting value of various models. The best performer was identified.