AbnormalReturns Data Structure

The key to the performance in this package is the underlying data structure. These rely on a combination of BusinessDays.jl and Tables.jl to provide fast access to slices of data based on dates.

DataVector

These structures provide strongly typed data that is easy to slice based on a range of dates. The data is always stored as Float64, even though it accepts elements of type Missing. In storing the data, Missing values are converted to 0.0, and the missing_bdays is a SparseVector that is true when that value is missing. dates are the minimum and maximum dates for the data.

Data in a DataVector is stored in an OffsetVector from OffsetArrays.jl, but this data is rarely accessed directly.

MarketData

    struct MarketData{T}
        calendar::MarketCalendar
        marketdata::Dict{Symbol, DataVector} # column names as symbols
        firmdata::Dict{T, Dict{Symbol, DataVector}} # data stored by firm id and then by column name as symbol
    end

This struct is made up of a set of DataVector. The main purpose of this is to provide efficient storage of the underlying data along with the calendar so that dates are easily translated to access the underlying data.

FixedTable