Overview
Intraday event studies analyze market reactions at sub-daily frequency, capturing price movements within minutes or seconds of an event. This is critical when events have immediate, short-lived effects that daily data would miss.
The R package v0.40.0 provides IntradayEventStudyTask with POSIXct timestamp support and a dedicated non-parametric significance test.
When to Use Intraday Analysis
- Scheduled announcements: Earnings releases, Fed rate decisions, macro data
- Flash events: Flash crashes, circuit breakers, sudden halts
- High-frequency trading: Algorithmic trading responses
- Precise timing: When you know the exact timestamp of the event
- Short-lived effects: Reactions that are absorbed within hours
IntradayEventStudyTask
library(EventStudy)
# Data must include POSIXct timestamps
intraday_task <- IntradayEventStudyTask$new(
data = intraday_data, # minute-level price data
event_time = "2020-03-15 14:30:00",
estimation_window = c(-120, -11), # in minutes
event_window = c(-5, 5) # in minutes
)
model <- MarketModel$new()
model$fit(intraday_task)Data Requirements
- Timestamps: POSIXct format with timezone information
- Frequency: Typically 1-minute or 5-minute intervals
- Market hours: Data should cover trading hours only
- Synchronization: Event and market data must be time-aligned
Non-Parametric Significance Test
The package implements the Rinaudo & Saha (2014) non-parametric test specifically designed for intraday event studies:
# Non-parametric test for intraday data
stats <- IntradayNonParametricTest$new()
stats$compute(intraday_task)This test is preferred over standard parametric tests because:
- Intraday returns often exhibit stronger non-normality
- Microstructure effects (bid-ask bounce, price discreteness) violate parametric assumptions
- The test is robust to irregular spacing of observations
Window Design for Intraday Studies
| Window | Duration | Purpose |
|---|---|---|
| Estimation | 60–120 minutes | Calibrate normal return model |
| Pre-event gap | 5–10 minutes | Buffer to avoid information leakage |
| Event window | 5–30 minutes | Capture the market reaction |
Best Practices
- Align timestamps across stock and market data carefully
- Exclude non-trading hours from the analysis
- Account for market microstructure effects
- Use robust tests given the non-normality of intraday returns
- Consider bid-ask bounce when using transaction-level data
References
- Rinaudo, J.B. & Saha, A. (2014). A non-parametric test for event studies with intraday data.
- Barclay, M.J. & Litzenberger, R.H. (1988). Announcement effects of new equity issues and the use of intraday price data. Journal of Financial Economics, 21(1), 71-99.