A release surprise compares the published actual with the consensus estimate. A revision instead changes a previously published observation. Subtract consensus from actual using the original statistic units. An upside surprise doesn't automatically imply a favorable market reaction; that depends on context outside this arithmetic exercise.
You are covering a fictional analyst desk. Read the requested units before running a command, then inspect the returned table or value. Keep the timestamp, observation window and position sign with your answer. A precise number can still answer the wrong question. The final task uses a different security or data selection so you have to apply the definition again. All prices and records are training data; this terminal cannot place orders or connect to a live feed.
Subtract consensus from the actual release.