IIM AHMEDABAD · NSE CASH MARKET · 2019–2021

How much does
a trade move
the price?

A study of execution size, market liquidity and the price response that follows. Explore three years of reconstructed quotes and matched annual stock cohorts.

Explore the evidence
At the comparison size30 seconds
6.39 bps

Illiquid minus liquid response

95% date-bootstrap interval
6.01 to 6.76 bps

Executions of 0.01%–0.02154% of prior average daily volume. Equal stock-day weights.

739 / 741

Usable source dates

236.9 million

Executions in the 30-second model

12,482

Accepted 30-second stock-days

10 + 10

Liquid and illiquid stocks per year

THE SIZE–RESPONSE RELATIONSHIP

Explore the impact curves.

Change the horizon and the unit of observation. Hover over a point for its support and uncertainty.

Response horizon
Signed midpoint response · basis points
Comparison interval0.01%–0.02154%
Illiquid—
Liquid—
Difference—
Date-bootstrap interval—

Loading the frozen results…

The definition of a trade matters. At the comparison size, grouping fills into incoming order events reduces the 30-second gap from 6.39 to 0.92 basis points. Grouping changes the observation unit, weighting and bin membership.

TIME AND SCALE

What happens after the trade?

Responses develop over seconds. Their apparent size also depends on the price scale used.

From one second to five minutes

Individual executions · share-based size

Horizon-specific samples can differ. These finite-horizon responses do not identify a permanent component.

A different price scale

30 seconds · 0.01%–0.02154% of prior ADV

The denominator is half the median stock-day quoted spread for each group. This aggregate scaling does not identify the contribution of spreads to the response.

SIZE × TIME × RESPONSE

Step inside the response surface.

Rotate the three-dimensional view and inspect the observed bin means. Compare the two liquidity groups on the same vertical scale.

Drag to orbitScroll to zoomHover over a point
Signed midpoint response—

The grid contains 18 size bins supported at all six horizons in both groups. In the 3D view, size-bin centres and elapsed seconds use logarithmic spacing. The 2D view uses a categorical horizon axis. Each vertex is an observed bin mean. Connecting faces interpolate visually between those means. Horizon-specific samples can differ.

THE STOCKS BEHIND THE ESTIMATES

Three annual cohorts.
A transparent selection rule.

Ten liquid–illiquid pairs are selected at the start of each year using prior turnover, price and volatility. Membership and labels stay fixed within each year.

2019 matched stock pairs
PairLiquidIlliquidPrior turnover ratio

MEASUREMENT AND INTERPRETATION

From messages to evidence.

The executable code and processing appendix record the reconstruction rules, filters, exclusions and estimation details.

SOURCE & RECONCILIATION

Start with the trading record.

NSE Capital Market orders and trades are reconciled against official daily quantity, value and execution counts. Two unusable sources are excluded explicitly.

QUOTE RECONSTRUCTION

Compare two replay paths.

An order-only and an execution-informed book must agree on eligible quotes. Internal checks include exact execution audits and strict quote-coverage gates.

ESTIMATION

Give each stock-day a voice.

Curves average stock-day means. The controlled model uses firm-year, date and time effects, with standard errors clustered by date and firm-year.

Signed midpoint response
R(h) = 10,000 × s × m(t + h) − m(t−)m(t−)

s is the aggressor sign. m is the bid–ask midpoint. Responses are in basis points.

What these estimates can support

The findings describe associations in the selected annual cohorts. News, trade timing and persistent order flow can influence both execution size and later prices. Independent exchange quote verification remains pending. The internal checks establish consistency, while external quote accuracy remains a measurement limitation.