Tick Data

Quick Answer

Tick data records every trade and quote update with price, size and timestamp, so it is the only data that lets a backtest walk a market order up the book to measure slippage. A liquid Bank Nifty feed can exceed a million quote updates a day, so it is reserved for fill modelling and microstructure despite its terabyte-scale cost.

Definition: Tick Data

Tick Data is the finest-grained market record of individual trades and quote updates, each timestamped with price and size, preserving the intrabar path that bar data discards.

Key takeaways: Tick Data

• Tick data is the raw, ordered stream of trades and quotes — the source of truth bars are derived from
• Fill and slippage modelling require quote ticks, not just trade ticks
• Its detail comes at terabyte-scale cost, so use it selectively where the detail matters
• Out-of-order, duplicate and crossed ticks are common and silently corrupt results unless cleaned

Tick Data at a glance

Tick Data in simple words

A tick is a single event on the exchange: a trade printing at a price and quantity, or a change in the best bid or offer. Tick data is the raw stream of all these events in order. If an OHLC bar is a daily weather summary, tick data is the continuous thermometer reading every fluctuation as it happens.

What Tick Data is for

Tick data exists so that strategies and simulations can see exactly what happened and when — the real sequence of trades and quotes — which is essential for microstructure research and for realistically modelling how an order would have filled.

Tick Data — professional explanation

Trade ticks versus quote ticks

There are two distinct kinds of tick, and conflating them is a common error. A trade tick (or 'last' tick) records an executed transaction: price, traded quantity and time. A quote tick records a change in the order book — typically the best bid and best ask with their sizes (Level 1), or the full depth of five or more price levels each side (Level 2 / market-by-price). Trade ticks tell you what changed hands; quote ticks tell you what was available to trade. Fill modelling needs quotes, because you fill against the bid or the ask, not against the last trade.

Size and the scale of tick data

Tick data is enormous. A single liquid NSE underlying like Bank Nifty can generate hundreds of thousands to millions of quote updates per day once you include depth, and the full options chain across strikes and expiries multiplies this dramatically. Storing a year of tick data for a broad universe runs to terabytes, which is why it is kept in compressed columnar formats (such as Parquet) and why most retail research downsamples to bars. The sheer volume also makes tick data slow to backtest over, so it is used selectively where its detail is genuinely required.

Uses: market microstructure

Microstructure research studies how the mechanics of trading — the spread, order-book depth, order flow, the tick-by-tick arrival of trades — affect price. Signals such as order-flow imbalance, the ratio of buy-initiated to sell-initiated volume, quote-update intensity and effective spread can only be computed from tick data. These are the raw materials of market-making, latency-sensitive strategies and short-horizon prediction. On NSE, tick data also reveals the pre-open auction dynamics and the microstructure of the last-half-hour closing session that bar data completely obscures.

Uses: realistic fill modelling

The most valuable practical use of tick data for a systematic trader is honest execution simulation. Instead of assuming your order fills at a bar's close, you replay the actual quote stream: a market buy fills by walking up the ask side of the book, consuming size at each level until filled, which directly models slippage and market impact. A limit order fills only if a trade prints through your price, or if you can prove you were at the front of the queue — queue position itself is estimable only from tick data. This turns slippage from an arbitrary backtest fudge factor into a measured quantity.

Volume, tick rule and trade classification

Tick data carries true traded volume per print, but it usually does not label whether a trade was buyer- or seller-initiated. Analysts infer this with the tick rule (a trade above the previous trade is buy-initiated) or the Lee-Ready algorithm (compare the trade price to the prevailing mid-quote). This classification underpins order-flow metrics but is itself an estimate with error, especially around fast markets and locked or crossed quotes. Treating inferred trade direction as ground truth is a subtle but real source of model error.

How bad tick data silently corrupts results

Tick data has its own pathologies that quietly poison analysis. Out-of-order timestamps (ticks arriving with earlier times than their predecessors due to feed handler quirks), duplicated ticks, zero or negative prices from feed glitches, and crossed quotes (bid above ask) all occur in raw NSE feeds. A microstructure signal computed on unsorted or duplicated ticks is meaningless, and a fill simulation on a crossed book will invent impossible fills. Because these errors are buried in millions of rows, they never announce themselves — they simply make the backtest lie with great precision.

Worked example: Tick Data

Trade ticks vs quote ticks

Advantages of Tick Data

• Preserves the exact sequence of trades and quotes — no intrabar information lost
• Enables realistic fill, slippage and market-impact modelling against the actual book
• The only basis for microstructure signals like order-flow imbalance and effective spread
• Can be aggregated to any bar interval, so it is the ideal single source of truth

Limitations of Tick Data

• Enormous storage and compute cost — terabytes per year for a broad universe
• Retail NSE depth is capped at five levels, so queue position and true depth are only estimated
• Trade direction (buy vs sell initiated) is inferred, not given, and the inference has error
• Prone to out-of-order timestamps, duplicates and crossed quotes that silently corrupt analysis
• Slow to backtest over, so unusable for large parameter sweeps without downsampling

How professionals treat Tick Data

Institutional quants and market-makers store full tick and order-book data as their canonical record and rebuild every bar and every backtest from it, so research and execution share one lineage. They invest heavily in feed-handler correctness — sequence numbers, gap detection, timestamp normalisation to a single clock — because a microsecond of disorder can invert a microstructure signal. Fill simulators replay the historical book event-by-event to estimate queue position and impact, and slippage is a measured distribution, not a constant. Retail traders rarely need this depth, but should understand that any serious claim about execution quality ultimately rests on tick data.

Common misconceptions about Tick Data

Common mistakes with Tick Data

• Using trade (last) ticks to model fills instead of quote ticks, so slippage is systematically understated
• Assuming tick timestamps are monotonic — failing to sort and de-duplicate before computing microstructure signals
• Treating inferred buy/sell trade direction from the tick rule as ground truth rather than an estimate
• Backtesting a whole strategy on tick data when only the fill layer needs it, wasting compute and time
• Ignoring crossed or locked quotes (bid >= ask) that create impossible fills in simulation
• Forgetting that retail NSE feeds show only five depth levels, so large-order impact is underestimated

Tick Data: frequently asked questions

Voice search: how people ask about Tick Data

Natural-language questions people ask about Tick Data.

Sources & references

• Irene Aldridge, “High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems”, 2nd ed., Wiley, 2013
• Ernest P. Chan, “Quantitative Trading: How to Build Your Own Algorithmic Trading Business”, 2nd ed., Wiley, 2021