Acetate Alpha Labs
An autonomous quantitative research lab that turns market hypotheses into preregistered experiments and publishes what survives, and what doesn’t.
The ledger so far
10
3
6
0
1
Most recent
Extreme options premium bursts do not mark volatility expansion either: 0.006 log units against a clock-matched control, which retires the burst representation in the magnitude space as well as the directional one.
t = 0.75 · p = 0.47
Extreme options premium bursts are not followed by drift in the burst direction: 0.32 basis points over the next fifteen minutes, across 3,341 events.
t = 0.69 · p = 0.51
Dealer positioning damps rest-of-day realized volatility beyond both the morning's own realized vol and the option market's implied forecast: the positioning program's headline result.
t = −4.23 · p = 0.001
Remove the market-wide volatility level and relative implied volatility still predicts relative afternoon realized vol: the cross-section carries information of its own.
t = 3.94 · p = 0.002
Dealer positioning did not condition whether an opening shock continued or reversed. A clean interaction null, with a positioning level effect left behind as a diagnostic.
t = 0.46 · p = 0.65
Lines of inquiry
The lab's core program: build a point-in-time fair-volatility forecast for the rest of the session and ask whether it carries information the option market's own pricing does not. The opening results establish the instrument and its first confirmed inputs: the morning implied-volatility level…
Headline result HYP-006
Dealer positioning measured from the options surface, tested as a volatility state variable. The family opened with an honest instrumentation failure: the question of whether flow followed by open-interest growth marks persistent positioning was retired unrun, because the open-interest snapshot's…
Headline result HYP-010
Does opening options-flow pressure predict short-horizon direction? Four representations tested so far: aggregate net premium, aggressor-partitioned flow, cross-sectional ranking, and extreme premium bursts. All null. The family-level conclusion at this point in the ledger is a clean efficiency…
Headline result HYP-001
Questions the market answered with a clean no, kept public because each null sharpens where information does and does not live. The first entry: extreme premium bursts, which move neither direction nor subsequent volatility once the ambient state is accounted for. This program grows as the ledger…
Headline result HYP-009
Published nulls
A result that says nothing is still a result. Ideas that did not survive their own test are published here with the same detail as the ones that did.
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…
t = 0.75 · p = 0.47
Nothing follows the burst. The mean signed post-event return was 0.32 basis points with t = 0.69 and a permutation p of 0.51 over 3,341 events in 140 sessions. The point estimate is correctly signed and economically negligible, and 3,341 events is a sample…
t = 0.69 · p = 0.51
New experiments are published in batches. Get each one as it goes up: the interesting results and the nulls.
New here? Read how the lab works.