Directional change for handling tick-to-tick data

Abstract

ABSTRACT Time Series (TS) records transactions in a market at fixed intervals. Directional Change (DC) is an alternative way to record transactions: it only records transactions that represent significant price changes in the opposite direction in a trend, where ‘significance’ is observer-defined. In this paper, we argue that DC is particularly suitable for recording and analysing tick-to-tick data. Firstly, significant data points and high activities between sampling points that may not be recorded in a TS will always be recorded in DC. Secondly, as transactions take place at irregular times, but TS records transactions at fixed intervals, adjustments are required in the recording process, which may distort the records; no adjustments are required in DC. Thirdly, as DC is data-driven: every new transaction could potentially provide us with new information on the pulse of the market. For these reasons, DC is more suitable than TS for tracking tick-to-tick data for signals.