Originally published on lavkesh.com
An at-the-money weekly option loses about 60% of its remaining time value in the final two trading days before expiry, according to Zerodha Varsity's options chapter. The decay accelerates instead of running flat, and the same curve shows up inside a two-week sprint. Most of what ships in the last two days was scheduled for the first eight, and it shows in the quality of the diff.
We tracked merged pull request size by day of the sprint for six sprints running. The median PR filed on day one or two ran about 140 lines. The median PR filed on day nine or ten ran under 40 lines, and nearly half of those skipped a second reviewer because the sprint review was already booked on the calendar.
That pattern is not an accident. It is the same mechanism that burns time value off an option. Work committed early still has room to be done right, and work committed late gets compressed until only the smallest, safest slice survives the deadline.
Standups do not catch this, because a standup measures whether a ticket moved columns, not whether the ticket that moved was the right size for the days left on the clock. A blocker announced on day nine gets the same thirty-second slot as a blocker announced on day two, even though one of them has no runway left to fix.
The fix that worked for us was not a longer sprint. It was a mid-sprint checkpoint on day six, where every ticket still in the backlog got re-estimated against the days actually remaining, not the days originally planned. Tickets that no longer fit got cut in that meeting, not discovered as a surprise on day ten.
Cutting a ticket on day six feels like admitting failure in the room. Shipping a rushed version of it on day ten feels like delivery, and the second one is worse for the codebase, but it gets rewarded anyway because nobody measures the review debt a compressed PR leaves behind.
Retrospectives that count velocity miss this entirely. A team can hit its story point target every sprint while quietly moving all the risk into the last forty-eight hours, the same way an option seller collects premium every week until the one week the underlying moves and the position that always worked stops working.
The mid-sprint checkpoint is now a fifteen-minute standing item on day six of every sprint we run. It has cut exactly one kind of failure: the ticket that ships on the last day because nobody looked at the calendar until it was too late to do anything but rush it.
Top comments (0)