Poznámka k běžně používaným vzorcům PIN na základě vygenerovaných dat

Translated title of the contribution: Note to commonly used PIN formulas based on the generated data

Research output: Chapter in Book/Report/Conference proceedingConference contribution

12 Downloads (Pure)

Abstract

This study tests randomly generated series of buys and sells to test nine various algorithms applied in SAS proc NLP. We suppose that the algorithms applied on generated data series with high PIN (PIN>0.2) will work approximately well in estimating PIN. But for data generated with zero-PIN or low-PIN (PIN<0.2) the algorithm showed overestimated results. PIN estimates are significantly affected by delivering (or not delivering) the initial parameters in the estimation while the effect of factorization is less significant. Not delivering the initial parameters overestimates PINs for zero-PIN and low-PIN data. Zero-PIN data appear to have data close to 0.26 while low-PIN data show their PIN to be overestimated by 0.05 to 0.10.
Translated title of the contributionNote to commonly used PIN formulas based on the generated data
Original languageOther
Title of host publication10th International Scientific Conference Financial management of Firms and Financial Institutions
Publication statusPublished - Sep 8 2015
Event10th International Scientific Conference Financial management of Firms and Financial Institutions, VŠB-TU of Ostrava, Faculty of Economics, Department of Finance - EF VSB-TU, Ostrava, Czech Republic
Duration: Sep 7 2015Sep 8 2015

Conference

Conference10th International Scientific Conference Financial management of Firms and Financial Institutions, VŠB-TU of Ostrava, Faculty of Economics, Department of Finance
Country/TerritoryCzech Republic
CityOstrava
Period9/7/159/8/15

Keywords

  • PIN
  • probability of informed trading
  • algorithm

ASJC Scopus subject areas

  • Finance

Fingerprint

Dive into the research topics of 'Note to commonly used PIN formulas based on the generated data'. Together they form a unique fingerprint.

Cite this