An event study for news: what the average reaction hides
Posted: Sat Oct 31, 2026 6:58 pm
To understand how a pair behaves around releases, an event study aligns many events on the release time and examines price paths. Collect all instances of a given release, say US jobs, normalize time relative to the event, and compute average movement, range, spread, and slippage.
The average path often shows nothing, because positive and negative surprises cancel. The absolute move is the informative measure. Plot the median absolute change by second after the release, and also its 90th percentile.
The analysis reveals how quickly volatility decays, when spreads normalize, and how often the initial direction reverses. Conditioning on the size of the surprise splits the sample into meaningful groups.
For scalpers, the outcome is a protocol: how many seconds or minutes to stay out, when conditions return to normal, and which releases justify the effort.
With tick data and a calendar, it takes a few hours to build.
The sample size is limited, since each release occurs once a month, so combine similar events and treat the results as indicative.
Which event study has been done, and what changed in the trading plan as a result?
The average path often shows nothing, because positive and negative surprises cancel. The absolute move is the informative measure. Plot the median absolute change by second after the release, and also its 90th percentile.
The analysis reveals how quickly volatility decays, when spreads normalize, and how often the initial direction reverses. Conditioning on the size of the surprise splits the sample into meaningful groups.
For scalpers, the outcome is a protocol: how many seconds or minutes to stay out, when conditions return to normal, and which releases justify the effort.
With tick data and a calendar, it takes a few hours to build.
The sample size is limited, since each release occurs once a month, so combine similar events and treat the results as indicative.
Which event study has been done, and what changed in the trading plan as a result?