Data topic hub

High-frequency data

∞structural

Microstructure as it is recorded: the trade/quote/book stream, its volume, and the statistical quirks that break the assumptions slower data lets you make.

Open the map The data you'd need →
What's in this topic

6 pages, each linking down to the ideas inside it.

How data is recorded01
Trades, quotes and book updates, each stamped with a sequence number and a clock time. Getting the timestamps and ordering right is the difference between a real backtest and a fiction.
∞structural
Data volume & engineering02
Terabytes per day per market. High-frequency data is enormous, irregularly timed, and unforgiving of naive storage: the engineering is half the battle.
∞structural
Bid-ask bounce03
The artificial up-down sawtooth in trade prices caused by trades alternating between hitting the bid and lifting the ask, with no change in fair value. Roll (1984) backs out the effective spread from exactly this serial covariance.
∞structural IX-BOUNCE
Fat tails04
High-frequency returns are not normal. The tails are far heavier than a Gaussian predicts, so the six-sigma event happens monthly, not once a millennium. Pretend otherwise and your risk model lies to you.
∞structural IX-FATTAILS
Irregular time & point processes05
Trades arrive in clustered bursts, not on the clock. Engle–Russell’s ACD model (1998) treats the durations between events as the object of study, the maths a mathematician will recognise immediately.
∞structural IX-DURATION
Trade-sign inference06
Most feeds don’t label who was the aggressor, so you infer it. The tick rule signs by price change; Lee–Ready (1991) compares the trade to the prevailing mid. Both are imperfect, and accuracy drops in fast markets.
∞structural IX-TICKRULE

Where this fits

↔ Across · composes with Market microstructure ↔ Across · composes with Systems & building ↔ Across · composes with Performance & capacity → Apply · makes money in Markets ⤓ Build / Buy · tool needed Datasets & tools

Questions this topic answers

Where do I get limit-order-book data to practise on?
Crypto exchanges publish full L2/L3 order-book data over free public WebSocket APIs, the most accessible real high-frequency data in 2026. Prediction-market venues (e.g. Polymarket) expose their books too. Equities tick/depth data is gettable but costly. For learning, capture a crypto feed yourself; for serious backtesting you need a clean, point-in-time L2/L3 dataset, the kind of resource on our datasets waitlist.