Options premium bursts predict volatility expansion over the next 15 minutes

A burst of options premium had already failed to predict which way price would move. That leaves the weaker and more plausible version of the same intuition: perhaps a large slug of premium does not say where price is going, but does say that something is happening, and volatility is about to pick up. Direction is the hard claim; magnitude is the one most people would expect to survive. A yes would have made bursts a usable short-horizon volatility trigger even though they carry no directional content, and would have given the volatility program an event-driven input alongside its level-based ones.

HYP-005 asked whether extreme premium bursts predict WHICH WAY price moves (NULL). This asks whether they predict HOW MUCH price subsequently moves: is post-burst realized volatility elevated relative to the pre-burst level, beyond what the same clock-time in an ordinary adjacent session shows? Burst definition reused frozen from HYP-005; this is a new outcome, not a parameter variation. A null does not falsify other horizons, trade-level burst definitions, or IV-response variants.

Nine liquid names (two index proxies and seven mega-caps) over seven months of sessions. The burst definition is reused frozen from the directional test and not re-tuned: a one-minute bucket whose absolute net options premium reaches eight times the causal within-session running baseline, eligible between 10:00 and 15:29, with a refractory period that prevents overlapping observations. Only the outcome is new. For each event the statistic is the log ratio of realized volatility over the fifteen buckets after the event to realized volatility over the thirty buckets before it, minus the same ratio computed at the identical clock position in that symbol's previous session. The matched subtraction is the point of the design: volatility has a pronounced intraday shape, so a raw post-versus-pre ratio would measure the time of day as much as the event, and pairing against the same minute of an ordinary neighbouring session cancels that shape exactly. Inference flips the sign of every event in a session together, 999 draws, preregistered positive direction. Pairs missing either side are dropped with accounting rather than filled.

Bursts do not mark volatility expansion. The mean pair difference was 0.006 log units, roughly half a per cent of relative volatility, with t = 0.75 and a permutation p of 0.47 across 3,319 matched pairs in 139 sessions. Correctly signed, far too small to matter, and well inside the permutation distribution. Read with its parent, this is the more useful null of the pair. Reusing the same frozen event definition and changing only the outcome splits the question cleanly into direction and magnitude, and the burst representation is now closed in both: extreme aggregate premium tells you neither where price will go nor how far it will travel. Whatever the largest prints of the day represent, it is absorbed by the time the premium is visible in aggregate, and the lab stopped building on the event-burst idea rather than searching for the horizon at which it might work.

Formation windows close roughly a minute before outcome windows open, and measured feed publication latency consumes most of that slack. Live availability of each signal at the formation boundary is therefore tight rather than generous.

The clock-matched control assumes the previous session is an ordinary one at that minute. It usually is, but a genuinely eventful neighbouring session inflates the control leg and shrinks the measured difference, and the design has no mechanism to detect that beyond the sheer number of pairs. Forty-two candidate events dropped during construction where one side of a pair was unavailable, and the reported count reflects those drops.

The burst is defined on aggregated premium with all trade types pooled, so a null does not falsify trade-level burst definitions such as sweeps or blocks, other horizons, or variants measured on the implied-volatility response rather than the realized one. Like every result the lab has published so far, it is a discovery-substrate result: preregistered, but not yet confirmed on data the research process has never seen.