Skip to content
All library documents

Handling Overnight Gaps in Intraday Tick-Data Models

Article Quant Q&A · Author: F.F.
Read the original

Summary

The document asks whether a month of NYSE tick data should be modeled day by day when price changes and time between trades are explained by their recent lags, in a GARMA-like setup. The answer recommends separating trading sessions because the overnight interval is absent from the sample and may contain trading activity or relevant market movement. Treating the final trade one day and the first trade the next day as adjacent observations can therefore misrepresent the elapsed time and the process being modeled.

Combining sessions may be defensible if there was no relevant overnight market activity, news, or price movement. The answer offers this as a conceptual modeling consideration, not an empirical comparison or a universal rule. The document does not discuss how to handle opening auctions, overnight data, or session-specific effects, so those choices remain outside its scope.

Key ideas

  • Overnight gaps can break the continuity assumptions of a model built from trade-to-trade intervals.
  • Unobserved overnight trading or news may affect the first observed trade of the next session.
  • Modeling each trading day separately is a reasonable default for regular-session tick data.
  • Joining sessions may be reasonable when the missing interval contains no relevant activity or movement.

Tags

EquitiesStatisticsMarket microstructure
Full text
# Should I analyze the tick data day by day?


# Should I analyze the tick data day by day?












Let assume that we have one month of tick data which were traded at NYSE. We want to model the price changes as a function of the last p lags of price changes and the last q lags of the time duration between trades ( this is similar to the GARMA(p,q) model). Each day we use the data from 9:30 until 16:00. My question is: in order to analyze the data should I analyze each day separately?

## Answer by SCallan (score 2)

https://quant.stackexchange.com/a/14213

Probably yes. There is a time gap between the two days. During that time gap it's likely that there was a market being made (and trading going on) for that security outside of what your data captures. If there was no market and no relevant news or market movement taking place in between days, then you could argue for keeping the data together.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.