Skip to content

Latest commit

 

History

History
132 lines (91 loc) · 5.7 KB

NEWS.md

File metadata and controls

132 lines (91 loc) · 5.7 KB
Raw

highfrequency 1.0.0

  • Updated citation as a paper on our package will be published in the Journal of Statistical Software

highfrequency 0.9.5

  • Fixing KaTeX note

highfrequency 0.9.4

  • Fixing reverse dependency issues w.r.t. FKF
  • Removed option based on BMS package because it got removed from CRAN

highfrequency 0.9.3

  • HEAVY model test less strict
  • Corrected bitwise '&' with boolean operand

highfrequency 0.9.2

  • Export noZeroPrices

highfrequency 0.9.1

  • Bug fix in ReMeDI estimator

highfrequency 0.9.0

  • Fix typo where in spotVol, the stochastic periodicity was erroneously written as stochper instead of stochPer.
  • Better consistency with the choices in spot* functions
  • Add warning instead of print in detPer in spotVol. The periodic component is no longer set to 1 either, just warned that results may be weird.
  • Implement Beta-Adjusted realized covariance estimation in the rBACov function
  • Add plotTQData function to plot trades and quotes.
  • Add ICov and IVar documentation pages that work as lists of implemented estimators of the integrated covariance and the integrated variance, respectively.
  • Deprecated loads of realized measure estimators and added new versions, so we have a higher degree of uniformity in function names.
  • More or less complete rework of HEAVY model and related functions
  • Fix bug in HARCJ model, thanks to @SKMASON (Zhixi Shao) for reporting the bug and providing pointers for the fix.
  • Better formatting of NEWS.md
  • aggregateTS now accept both data.table and xts objects

highfrequency 0.8.0

  • Various bugfixes and performance improvements
  • Added lowercase "r" in front of realized measures that did not have it, so it can easily be found in IDEs with highfrequency::
  • Added drift burst code
  • Added Backwards - Forwards matching algorithm to cleaning step. Thanks to Kim Christensen for providing his Matlab code.
  • removed various datasets, outdated TAQ format, simulated 5-minute prices, returns, and prices with jumps
  • Added European sample data, anonymized one-minute data with anonymous stock and market data, as well as home-grown dataset with realized measures
  • The arguments "on" and "k" in data handling is now "alignBy" and "alignPeriod" respectively. This means that the realized measures and data handling functions have similar notation
  • Improved UX in data cleaning functions by having more clear report on the trades when prompted and adding defaults that follow the standard in the literature
  • Added data.table support to (most) realized measures
  • Added functions makeRMFormat and makeOHLCV to convert data from a long format to a format that can be used for realized measures, and to make arbitrary period bars.
  • HARmodel has less 'RV' in the types - much easier to type and read, also it now supports an external regressor and has robust standard errors reported in the summary.
  • add asymptotic variance estimator for ReMeDI estimation - thanks to Merrick Li for contributing his Matlab code
  • In general, improved documentation. Also, better documentation of methods which were, in most cases undocumented and not clearly exported

highfrequency 0.7.0

  • New naming convention
  • Bugfix in BNSjumpTest, JOjumpTest, AJjumpTest. These functions behaved in an unexpected and inconsistent manner when the input spanned more than one day
  • Bugfix in aggregateTS function which in edge cases returned data from AFTER the input data
  • Implement intradayJumpTest function which allows for flexible Lee-Mykland style jump tests
  • Implement rankJumpTest to test for the rank of the jump matrix
  • Implement new features in spotVol. Now the local volatility can be estimated with realized measures, they can also be used with pre-averaged realized measures.
  • Implement a wrapper around quantmod's getSymbols.av function
  • harModel now includes Newey-West standard errors in the output
  • Bugfix for refreshTime function and large performance improvement
  • Implement CholCov estimator in rCholCov
  • Bugfixes in data handling functions, which sometimes produced different results depending on the options(digits.secs) setting. Most data handling functions now run considerably faster as a consequence of internally using numerics for timestamps.
  • Implemented new realized semi-covariance estimator in rSemiCov
  • Implemented new lead-lag estimation in leadLag
  • Implemented ReMeDI estimation in ReMeDI
  • More transparently handle the lagging of quotes when matching these with trades, now the user has control of this.
  • Add business time sampling
  • Changes to the included datasets. The microseconds quote datasets have been thinned out aggressively for exchanges != "N"

highfrequency 0.6.5

  • bug fix for kernelCov if cor = TRUE
  • compatibility with lubridate 1.7.8

highfrequency 0.6.4

  • bug fix in refreshTime (affected rMRC for n > 2)
  • one additional test for rMRC
  • updated realized library file until end of 2019

highfrequency 0.6.3

  • aggregateTrades size aggregation bug fix

highfrequency 0.6.2

  • spotVol and spotDrift do not assume naming convention for univariate time series anymore
  • bug fix tpv and finite sample corrections

highfrequency 0.6.1

  • bug fix for Fedora compilation

highfrequency 0.6.0

  • all new backend
    • documentation via roxygen2
    • testing via test_that
    • covr integration on github
  • microsecond compatibility for WRDS files
  • improved documentation
  • new options in harModel
  • updated data sets
  • updated references
  • cleanup of code basis

highfrequency 0.5

  • converted so that it would work with Cran
  • added missing data files
  • compressed data files

highfrequency 0.4

  • update package code github to version on rforge
  • to do: print more output in tradesCleanup about the different filters
  • correction to implementation AJjumptest by Giang