Tick data
Also written tick-by-tick
Individual market events with timestamps, rather than prices summarised into bars: every trade, and usually every quote change behind them. The word is not precise about which of those is included, and the difference is roughly thirty-fold in volume, so two vendors selling tick data can be selling products that differ by an order of magnitude in size and price.
How it works
A tick is one event: a trade printed, or a quote changed. Tick data is the sequence of those events with timestamps, as opposed to a bar, which is a summary of everything that happened in a minute or a day. Nothing about the word says which events are in it, and that ambiguity is the main thing to know about it.
The published sizes make the point better than any description. NYSE's Daily TAQ client specification dated 3 March 2026 gives typical figures for one trading day of consolidated US equities — the whole market, not one venue: trades 2.4 GB across 69 million records, quotes 38 GB across 1.9 billion records, and the NBBO file 11 GB across 330 million. All three are gzipped figures. So "tick data" meaning trades is a couple of gigabytes a day, and "tick data" meaning trades and quotes is forty, from the same market on the same day.
Uncompressed, it is worse. In NYSE's public samples directory, the 26 quote files for 1 April 2026 total 47.6 GB; a sampled unzip ran about 5.2 to 1, which puts a day of quotes near 250 GB raw and a year near 60 TB. Options are the extreme case and nobody publishes a byte figure: what the plan publishes is messages, and the revised OPRA capacity projections of 15 September 2025 project 13.575 billion messages a day effective July 2026, rising to 14.964 billion by July 2027 — for one of two redundant streams.
What differs between two vendors both selling ticks
Quotes or not. The thirty-fold difference above. Ask first.
Venues, or the consolidated view. A venue's own ticks show that venue. The consolidated tape shows the market and carries no depth. Both are called tick data.
Timestamps. Whose clock, at what resolution, and whether the venue's own timestamp survives alongside the vendor's receipt time. For anything about sequence, this is the field that decides whether the dataset can answer the question at all.
Condition codes. A trade print carries flags saying what kind of trade it was. Dropping them makes a series tidier and makes any volume or price statistic computed from it subtly wrong.
Depth. Level 2 and market-by-order are separate products again, priced separately, and a vendor that has them will say so — AlgoSeek carries futures depth and explicitly does not carry it for US equities, because the tape does not.
Why it matters here
Two practical consequences, and they run in opposite directions.
First, most projects do not need ticks. If the question is measured in minutes, bars are smaller by orders of magnitude, cheaper to license and cheaper to query, and the tick storage category exists precisely because the ones that do need ticks hit an engineering problem rather than a data-shopping one. Little of that cost is storage — 12 TB a year on object storage is a modest bill — and most of it is query time and the licence.
Second, if you do need them, the size figures above should drive the decision about where they live before you buy a byte. Storing tick data works through the formats and engines; how to backfill minute bars is the cheaper path that covers most of what people actually want ticks for.
One licensing note worth carrying over. Historical tick files are normally sold for internal use, and a program acting on live ticks is non-display use, priced per firm rather than per screen. The real-time fee stack applies to ticks with the largest multipliers on this site.
Where you will meet this
The cards where this changes a decision, then the rest that use the word.
Sources
- Daily TAQ Client Specifications, version 4.3 — New York Stock Exchange,
- Daily TAQ historical data samples directory — New York Stock Exchange, read
- Revised OPRA Capacity Projections, notice to OPRA Multicast Data Subscribers — Securities Industry Automation Corporation (OPRA),
FAQ
Do I actually need tick data?
Usually not. Anything measured in minutes or days is better served by bars, which are smaller by three or four orders of magnitude and far cheaper to license. Ticks earn their cost when the question is about the spread, the queue, execution quality or the sequence of events inside a second.
What is the first question to ask a tick data vendor?
Whether quotes are included, and which venues. Trades alone are a modest dataset; trades plus every quote change is roughly thirty times larger. After that, ask whether the timestamps are the venue's or the vendor's, and whether trade condition codes are carried through.
Updated