Approaching a large expiry, participants who want to keep exposure close the expiring contract and open a further-dated one. Open interest in the near contract falls; open interest further out rises.
Net positioning has not changed. Only its location has.
What it looks like if you watch one expiry
A collapse. The chart of that contract’s open interest falls steeply into expiry every single time, which reads as capitulation and is nothing of the sort.
This is the same error as reading open interest without knowing whether contracts were opened or closed — the number moved, and the movement means something different from what it appears to.
How to tell a roll from a real unwind
Look at the whole board rather than one expiry. A roll shows near-dated open interest falling and further-dated rising by a comparable amount; a genuine unwind shows the total falling with nothing picking it up.
The aggregate across expiries is the number that answers “is there less positioning now”, and the per-expiry series answers “where is it”. Confusing the two is how a routine calendar event becomes a narrative.
What the roll actually tells you
Something modest and real: it says how much conviction extends past the near term. A roll that moves most of the open interest forward is participants choosing to keep exposure; one where much of it simply expires is exposure being retired.
And the shape of where it lands is informative — a roll concentrating into the next quarterly is a different statement from one spreading across weeklies.
Why it matters for the map
Because the gamma profile is rebuilt around the new concentration. Walls near the expiring strike stop mattering and structure appears around the strikes the roll landed on — which is the discontinuity described in what happens at 08:00 UTC, seen from the days before rather than the minute after.