Lots o' Ticks: Realtime high performance time series queries on billions of trades and quotes

Arthur Whitney, Dennis Shasha

Research output: Contribution to journalArticle

Abstract

Financial mathematicians think they can predict the future by looking at time series of trades and quotes (called ticks) from the past. The main evidence for this hypothesis is that prices uctuate only by a small amount in a given day and more or less obey the mathematics of a random walk. The hypothesis allows traders to price options and to speculate on stocks. This demonstration presents a query language and a parallel database (50-way parallelism) to support traders who want to analyze every tick, not just end-of-day ticks, using temporal statistical queries such as time-delayed correlations and tick trends. This is the first attempt that we know of to store and analyze hundreds of gigabytes of time series data and to query that data using a declarative time series extension to SQL (available at www.kx.com).

Original languageEnglish (US)
Pages (from-to)617
Number of pages1
JournalSIGMOD Record
Volume30
Issue number2
DOIs
StatePublished - 2001

Fingerprint

Time series
Query languages
Demonstrations

Keywords

  • Finance
  • Parallel databases
  • Time series

ASJC Scopus subject areas

  • Software
  • Information Systems

Cite this

Lots o' Ticks : Realtime high performance time series queries on billions of trades and quotes. / Whitney, Arthur; Shasha, Dennis.

In: SIGMOD Record, Vol. 30, No. 2, 2001, p. 617.

Research output: Contribution to journalArticle

@article{190ed0c1bf5049b186a32d45d4d1ad1a,
title = "Lots o' Ticks: Realtime high performance time series queries on billions of trades and quotes",
abstract = "Financial mathematicians think they can predict the future by looking at time series of trades and quotes (called ticks) from the past. The main evidence for this hypothesis is that prices uctuate only by a small amount in a given day and more or less obey the mathematics of a random walk. The hypothesis allows traders to price options and to speculate on stocks. This demonstration presents a query language and a parallel database (50-way parallelism) to support traders who want to analyze every tick, not just end-of-day ticks, using temporal statistical queries such as time-delayed correlations and tick trends. This is the first attempt that we know of to store and analyze hundreds of gigabytes of time series data and to query that data using a declarative time series extension to SQL (available at www.kx.com).",
keywords = "Finance, Parallel databases, Time series",
author = "Arthur Whitney and Dennis Shasha",
year = "2001",
doi = "10.1145/376284.375783",
language = "English (US)",
volume = "30",
pages = "617",
journal = "SIGMOD Record",
issn = "0163-5808",
publisher = "Association for Computing Machinery (ACM)",
number = "2",

}

TY - JOUR

T1 - Lots o' Ticks

T2 - Realtime high performance time series queries on billions of trades and quotes

AU - Whitney, Arthur

AU - Shasha, Dennis

PY - 2001

Y1 - 2001

N2 - Financial mathematicians think they can predict the future by looking at time series of trades and quotes (called ticks) from the past. The main evidence for this hypothesis is that prices uctuate only by a small amount in a given day and more or less obey the mathematics of a random walk. The hypothesis allows traders to price options and to speculate on stocks. This demonstration presents a query language and a parallel database (50-way parallelism) to support traders who want to analyze every tick, not just end-of-day ticks, using temporal statistical queries such as time-delayed correlations and tick trends. This is the first attempt that we know of to store and analyze hundreds of gigabytes of time series data and to query that data using a declarative time series extension to SQL (available at www.kx.com).

AB - Financial mathematicians think they can predict the future by looking at time series of trades and quotes (called ticks) from the past. The main evidence for this hypothesis is that prices uctuate only by a small amount in a given day and more or less obey the mathematics of a random walk. The hypothesis allows traders to price options and to speculate on stocks. This demonstration presents a query language and a parallel database (50-way parallelism) to support traders who want to analyze every tick, not just end-of-day ticks, using temporal statistical queries such as time-delayed correlations and tick trends. This is the first attempt that we know of to store and analyze hundreds of gigabytes of time series data and to query that data using a declarative time series extension to SQL (available at www.kx.com).

KW - Finance

KW - Parallel databases

KW - Time series

UR - http://www.scopus.com/inward/record.url?scp=85024251781&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85024251781&partnerID=8YFLogxK

U2 - 10.1145/376284.375783

DO - 10.1145/376284.375783

M3 - Article

AN - SCOPUS:85024251781

VL - 30

SP - 617

JO - SIGMOD Record

JF - SIGMOD Record

SN - 0163-5808

IS - 2

ER -