Data dictionary

Every field in the archive files (177) and every LAB column (120), grouped by the dealer-gamma book it reads; magnitudes are never comparable across books.

VOLUME default

cp x cumulative session volume (cp = +1 call, -1 put). Spotted by: suffix _volcp or _vol in the payload, _vol in LAB.

Payload fields 19

field	meaning	unit	since
`ngv_vol` Minute series	Net cp x volume gamma at spot, 0DTE only. The minute twin of `sec.ngv_volcp`.	$M per index point	all sessions (backfilled 2026-10-03 from `sec.ngv_volcp` at each minute's :59 second)
`gpct_vol` Minute series	The same percentile computed on `ngv_vol` against the same minute of the prior 20 sessions that carry it.	percentile	all sessions; all null on the first 10 sessions (2019-06-19, 2022-04-14 to 2022-05-02: pool under 10)
`flip_vol` Minute series	Volume-book zero-gamma flip, 0DTE, raw directional root (+/-2%, then +/-6%): `sec.flip_volcp` at the end of the minute. The default flip. `null` when the book is one-sided.	index points	all sessions (backfilled 2026-10-03 from `sec.flip_volcp` at each minute's :59 second)
`call_wall_vol` `put_wall_vol` Minute series	Peak positive / negative volume-book gamma strike within +/-2.5% of spot, 0DTE: `sec.cr0_volcp / ps0_volcp` at the end of the minute. The default walls.	strike	all sessions (backfilled 2026-10-03 the same way)
`hhi` Minute series	Removed from the payload 2026-10-08 (sessions published before carry it; later ones do not). Herfindahl concentration of abs(volume-book gamma) across strikes within +/-2.5% of the median strike, narrow scope, x100. High = a few strikes hold the book. Not the hold band (that is `frames[].hhi`).	percent (0 to 100)	sessions to 2026-10-08
`flowvel` Minute series	Flow velocity: sum over strikes of the absolute 15-minute change in volume-book gamma, narrow scope.	$M per index point	all sessions
`now.gpct_vol` `now`	Last finite `gpct_vol` (backfilled files) or the last minute's (new builds).	percentile	all sessions; null on the first 10
`levels.flip_vol` `levels.call_wall_vol` `levels.put_wall_vol` `levels`	`flip_vol` / `call_wall_vol` / `put_wall_vol` at the last frame's minute, the instant `levels.flip` is read at.	index points; strike	all sessions (backfilled 2026-10-03)
`frames[].gv` `frames[].gv0` `frames[]`	cp x volume gamma per strike. `gv0` is the object `ngv_vol`, `flip_volcp` and the volume walls are built from; `gv` adds the later expiries.	$M per index point	all sessions
`frames[].vv` `frames[].vv0` `frames[]`	cp x volume vanna per strike.	$M of delta per vol point	all sessions
`frames[].cv` `frames[].cv0` `frames[]`	cp x volume charm per strike.	$Bn of delta per day	all sessions
`frames[].vc` `frames[].vp` `frames[].vc0` `frames[].vp0` `frames[]` · volume (contracts)	Cumulative session volume per strike, calls (`vc`) and puts (`vp`) apart, no greek and no sign. The page's VOL COUNT ladder. On a session published before these keys the page reads the 0DTE count from the `vol` block (cb+cs+cu, pb+ps+pu) and says so. Not the per-second `sec.vc` / `sec.vp`, which are the whole chain's totals per second.	contracts	SPX sessions built from 2026-10-08 (live and rebuilt). SPX only: feedd's ticker frames (src/gateway/feedd/src/book.rs) do not carry them (payload_schema.json frame_spx_only)
`sec.flip_volcp` `sec`	Volume-book zero-gamma flip, 0DTE, raw directional root (+/-2%, then +/-6%). The default flip.	index points	all sessions
`sec.cr0_volcp` `sec.ps0_volcp` `sec`	Peak positive / negative volume-book gamma strike within +/-2.5% of spot, 0DTE. The default walls.	strike	all sessions
`sec.ngv_volcp` `sec`	Net volume-book gamma at spot, 0DTE.	$M per index point	all sessions
`sec.vc` `sec.vp` `sec.vcn` `sec.vpn` `sec`	The split of the printed contracts by right and expiry: `vc` the call contracts of today's expiry printed in that second, `vp` its puts; `vcn` / `vpn` the same for the near expiries the engine streams (the three nearest SPXW expiries). Their difference per bar is the VOLUME DELTA the terminal draws in the scope of the chart's expiry switch (0DTE: vc − vp; NEXT: vcn − vpn; ALL EXP: the sum), the volume book's own sign (a call bought is a call), never a guess at the aggressor. Live recordings since 2026-10-07; for every other day the `{day}.vol.json.gz` sidecar rebuilt from the raw tapes carries the same series (contracts/vol-sec.json); a sidecar without a near tape carries today's pair alone.	contracts	live recordings since 2026-10-07; sidecars rebuilt from the tape
`sec.dc` `sec.dp` `sec.dcn` `sec.dpn` `sec`	The same split as `vc` / `vp` / `vcn` / `vpn`, each print weighted by its option's delta at the print: `dc` = sum of size x \|delta\| over today's calls printed in that second, `dp` the same over its puts, `dcn` / `dpn` the near expiries'. Absolute deltas per side, so `dc - dp` is the prints netted into index delta (a put's delta is negative; the minus restores it), the DELTA unit of the volume delta band (0DTE: dc − dp; NEXT: dcn − dpn; ALL EXP: the sum). Delta is Black-Scholes at the contract's carried IV and the parity spot (common.print_delta), the zero-vol limit (e^{-qT} in the money, 0 out of it) where the contract has no IV yet. Live: spot and IV of the second; sidecar: of the minute (the surface's parity spot and the IV of the print's own NBBO, carried per contract). Not multiplied by 100: one contract at delta 1 is 1. Live recordings since 2026-10-08; sidecars rebuilt since then.	delta-contracts (size x \|delta\|)	live recordings since 2026-10-08; sidecars rebuilt from the tape
`orders[].flip` order ledger	Distance of the order's top-gamma strike from the volume flip at that minute, the flip the chart draws (the measured flip until 2026-10-03; every archived ledger was re-pointed that day). `null` when the session carries no volume flip at that minute.	index points (absolute)
`orders[].side` order ledger	`over` / `under` the volume flip.	string

LAB columns 18

column	meaning	unit	built from
`flip_vol_dist` levels · volume	spot - the VOLUME-book zero-gamma flip, points (+ = above it, the positive-gamma side): THE flip, the one the chart draws	points, signed	`spot` `sec.flip_volcp` spot - the last sec.flip_volcp inside the minute
`gamma_pos_vol` levels · volume	1 when spot is at or above the VOLUME-book flip (long gamma), 0 below	flag	`spot` `sec.flip_volcp` 1 when flip_vol_dist >= 0, 0 below, NaN when either is missing
`call_wall_vol_dist` levels · volume	spot - the VOLUME-book call wall (the 0DTE strike with the most positive gamma), points (+ = above it)	points, signed	`spot` `sec.cr0_volcp` spot - the last sec.cr0_volcp inside the minute
`put_wall_vol_dist` levels · volume	spot - the VOLUME-book put wall (the 0DTE strike with the most negative gamma), points (+ = above it)	points, signed	`spot` `sec.ps0_volcp` spot - the last sec.ps0_volcp inside the minute
`call_wall_strength_vol` levels · volume	share of the near-money 0DTE VOLUME book standing on the volume call wall's own strike, 0..1 -- a thin wall is drawn like any other but may behave like none	share	`frames[].gv0` `frames[].strikes` `sec.cr0_volcp` \|gv0\| on the strike nearest the minute's cr0_volcp (within 3 pt) / sum \|gv0\| over the frame; each 5-minute frame carried forward
`put_wall_strength_vol` levels · volume	same for the VOLUME-book put wall, 0..1	share	`frames[].gv0` `frames[].strikes` `sec.ps0_volcp` \|gv0\| on the strike nearest the minute's ps0_volcp (within 3 pt) / sum \|gv0\| over the frame; carried forward
`call_wall_agree` levels · all three	how many of the three books (volume, open interest, measured) name the SAME call wall, 1..3. Two or more agree in 45.5% of minutes, all three in 7.1% (store through 2026-09-23)	count	`sec.cr0_volcp` `sec.cr0_oi` `sec.cr0` of the three books' call walls at the minute, how many name the most-named strike
`put_wall_agree` levels · all three	same for the put wall: two or more books agree in 29.9% of minutes, all three in 1.8%	count	`sec.ps0_volcp` `sec.ps0_oi` `sec.ps0` of the three books' put walls at the minute, how many name the most-named strike
`call_wall_agree_dist` levels · all three	spot - the call wall the most books name, points (+ = above it)	points, signed	`spot` `sec.cr0_volcp` `sec.cr0_oi` `sec.cr0` spot - the most-named call wall
`put_wall_agree_dist` levels · all three	spot - the put wall the most books name, points (+ = above it)	points, signed	`spot` `sec.ps0_volcp` `sec.ps0_oi` `sec.ps0` spot - the most-named put wall
`net_gamma_vol` book · volume	net gamma at spot, VOLUME book, $M -- 'how much gamma' unqualified	$M	`sec.ngv_volcp` the last value inside the minute (the same object as the minute series ngv_vol)
`net_gamma_vol_flow` book · volume	1-minute change of net_gamma_vol, $M per minute -- 'gamma flow' unqualified	$M per minute	`sec.ngv_volcp` 1-minute difference of net_gamma_vol
`gamma_pctl_vol` book · volume	net_gamma_vol ranked against the SAME minute of the prior 20 sessions, 0..100 (the chart's GAMMA PCTL). Read it as a decile: 20 sessions give ~11 points of error at the median	percentile	`sec.ngv_volcp` recomputed over the store from net_gamma_vol: the share of the same clock minute in the prior 20 sessions strictly below, x 100; NaN under 10 comparands. Not copied from gpct_vol
`total_change_15_vol` book · volume	sum over strikes of \|gamma change\| of the 0DTE VOLUME book against 15 minutes earlier, $M -- 'book being rebuilt / flow velocity' unqualified	$M	`frames[].gv0` `frames[].strikes` sum over strikes of \|gv0 now - gv0 in the latest frame at least 15 minutes older\|, strikes matched by value; carried forward
`concentration_vol` book · volume	Herfindahl of \|gamma\| across the near-money strikes of the 0DTE VOLUME book, 0..1: high = a few strikes hold it and pin hard -- 'concentrated book / pin risk' unqualified	Herfindahl	`frames[].gv0` sum of (\|gv0\| / sum \|gv0\|)^2 over the frame's strikes; carried forward. Not the minute series hhi (narrow scope, x 100)
`gamma_0dte_vol` expiry · volume	VOLUME-book gamma, today's expiry only ($M)	$M	`frames[].gv0` sum of gv0 over the frame; carried forward
`gamma_next_vol` expiry · volume	VOLUME-book gamma past today ($M)	$M	`frames[].gv` `frames[].gv0` sum of (gv - gv0); carried forward
`gamma_next_share_vol` expiry · volume	share of the VOLUME book that outlives today (0..1)	share	`frames[].gv` `frames[].gv0` sum \|gv - gv0\| / sum \|gv\|

MEASURED

the dealer's net position, signed print by print off the trade tape. Spotted by: neutral payload names (flip, cr0, ps0, gpct, ngv_meas, hold_lo/hi, dex), _meas in LAB.

Payload fields 41

field	meaning	unit	since
`flip` Minute series	Measured zero-gamma flip: the root of net measured gamma over a grid of hypothetical spot (+/-2%, widened once to +/-6%), on a curve smoothed with a 3-minute half-life EMA and tracked to the previous root. Lags the raw book by about 4 minutes. `null` when the book is one-sided over +/-6%.	index points	all sessions
`flip_se` Minute series	Crossing standard error of `flip` (Mueller 1985): trailing curve noise over the slope at the root. `null` until 5 innovations exist or above 99.	index points	since 2026-08-24 (29 sessions)
`ngv_meas` Minute series	Net measured dealer gamma at spot, narrow scope.	$M per index point	all sessions
`gpct` Minute series	Percentile (0 to 100) of `ngv_meas` against the SAME minute of the prior 20 sessions; `null` under 10 comparands. A decile, not a rank: the sampling error is about +-11 points at the median. No measured forward edge once the time of day is held fixed.	percentile	all sessions (sessions before 2026-08-24 recomputed to this definition on 2026-10-03, see Known caveats)
`hold_lo` `hold_hi` Minute series · measured only	Removed from the payload 2026-10-08 (sessions published before carry it; later ones do not). The hold band: the contiguous interval of hypothetical spot around current spot where measured dealer gamma stays positive, searched over +/-6%. `null` when spot itself sits in negative measured gamma. An edge that reached the end of the grid reads as spot +/-6%. No volume-book band exists.	index points	sessions to 2026-10-08
`charm` Minute series	Measured charm pressure: the change in dealer dollar delta per day from time decay, narrow scope.	$Bn of delta per day	all sessions
`dex` Minute series	Measured dealer delta inventory, narrow scope. Millions of index units ("shares"), not dollars: multiply by `spot` for $M.	millions of index units	all sessions
`vex` Minute series	Measured dealer vega, narrow scope.	$M per vol point	all sessions
`exp_range` Minute series · measured (adjustment)	FirmTape's forecast of the range left to the close: `rv30`-based sigma x spot x sqrt(minutes left), scaled by exp(-0.147 z), z = `ngv_meas` z-scored against the prior 20 sessions' pooled minutes. Measured to add nothing over the straddle; kept for comparison.	index points	all sessions
`now.gpct` `now`	Last minute's `gpct`.	percentile	all sessions
`now.exp_range` `now` · measured (adjustment)	`exp_range` at the second-to-last minute; near zero at the close by construction.	index points	all sessions
`levels.flip` `levels`	`frames[-1].flip`.	index points	all sessions
`levels.cr0` `levels`	Strike with the largest positive measured gamma within +/-2.5% of spot (the measured "call wall"), narrow scope.	strike	all sessions
`levels.ps0` `levels`	Strike with the most negative measured gamma within +/-2.5% of spot (the measured "put wall").	strike	all sessions
`levels.hold_lo` `levels.hold_hi` `levels`	Removed from the payload 2026-10-08 (sessions published before carry it; later ones do not). The hold band at the last frame.	index points	sessions to 2026-10-08
`ladder.g_flow` `ladder`	`frames[-1].g`.	$M per index point	all sessions
`ladder.v_flow` `ladder`	`frames[-1].v`.	$M of delta per vol point	all sessions
`ladder.c_flow` `ladder`	`frames[-1].c`.	$Bn of delta per day	all sessions
`frames[].g` `frames[].g0` `frames[]`	Dealer gamma per strike.	$M per index point	all sessions
`frames[].v` `frames[].v0` `frames[]`	Dealer vanna per strike.	$M of delta per vol point	all sessions
`frames[].c` `frames[].c0` `frames[]`	Dealer charm per strike.	$Bn of delta per day	all sessions
`frames[].d` `frames[].d0` `frames[]`	Dealer delta per strike, in dollars (delta x 100 x spot), unlike the aggregate `dex`.	$M of index exposure	all sessions
`frames[].e` `frames[].e0` `frames[]`	Dealer vega per strike.	$M per vol point	all sessions
`frames[].cr0` `frames[]`	Peak positive measured gamma strike in this frame (`argmax g`).	strike	all sessions
`frames[].ps0` `frames[]`	Peak negative measured gamma strike (`argmin g`).	strike	all sessions
`frames[].flip` `frames[]`	`flip` at this minute.	index points	all sessions
`frames[].hlo` `frames[].hhi` `frames[]`	Removed from the payload 2026-10-08 (sessions published before carry it; later ones do not). `hold_lo` / `hold_hi` at this minute. `hhi` here is the hold band's upper edge, not the top-level `hhi` concentration series.	index points	sessions to 2026-10-08
`vol.cb` `vol.pb` `vol` · measured (signing)	Customer BUY contracts, calls / puts, per strike, cumulative.	contracts	all sessions
`vol.cs` `vol.ps` `vol` · measured (signing)	Customer SELL contracts, calls / puts.	contracts	all sessions
`sec.flip` `sec`	Measured flip (smoothed, tracked, minute cadence).	index points	all sessions
`sec.flip_se` `sec`	Its crossing standard error.	index points	since 2026-08-24
`sec.cr0` `sec.ps0` `sec`	Peak positive / negative measured gamma strike within +/-2.5%, narrow scope.	strike	all sessions
`sec.ngv_meas` `sec`	Net measured gamma at spot.	$M per index point	all sessions
`sec.hold_lo` `sec.hold_hi` `sec` · measured only	Removed from the payload 2026-10-08 (sessions published before carry it; later ones do not). The hold band per second.	index points	sessions to 2026-10-08
`sec.charm` `sec`	Charm pressure.	$Bn of delta per day	all sessions
`sec.vanna` `sec`	Vanna pressure (the only aggregate vanna series).	$M of delta per vol point	all sessions
`sec.mc.<w>.strike` `sec.mc.<w>.value` `sec`	Max-change gamma, w in `1 5 10 15 30` minutes: the strike whose measured gamma moved most over the trailing w minutes, and the signed change. `null` for the first w minutes.	strike; $M per index point	all sessions
`orders[].g` order ledger	Dealer gamma the order added (measured sign); `null` if it could not be priced.	$M per index point
`stats.no_side` `stats.no_gamma` order ledger	Orders the tape could not side / could not price.	count
`stats.gamma_up` `stats.gamma_dn` order ledger	Orders that added positive / negative dealer gamma.	count
`stats.<event name>` calendar file · measured (the book statistics)	What usually followed that kind of event: `window`, `ratio` (range in the window over the trailing-move expectation, event days over other days), `n`, `study`, `recent` [[date, value]], `book` {`flip_crossed`, `at_wall`, `short_gamma`, `short_vs_long`}. The `book` statistics use the measured flip and the measured hold band; `at_wall` means "ended within a few points of a hold-band edge", not of `cr0` / `ps0`.

LAB columns 33

column	meaning	unit	built from
`flip_meas` levels · measured	the MEASURED-book zero-gamma flip as a price (use flip_meas_dist in rules)	index points	`flip` copied
`flip_meas_dist` levels · measured	spot - the MEASURED-book flip, points (+ = above it)	points, signed	`spot` `flip` spot - flip
`gamma_pos_meas` levels · measured	1 when spot is at or above the MEASURED-book flip, 0 below	flag	`spot` `flip` 1 when spot >= flip, 0 below, NaN when either is missing
`call_wall_meas_dist` levels · measured	spot - the MEASURED-book call wall, points (+ = above it)	points, signed	`spot` `sec.cr0` spot - the last sec.cr0 inside the minute
`put_wall_meas_dist` levels · measured	spot - the MEASURED-book put wall, points (+ = above it)	points, signed	`spot` `sec.ps0` spot - the last sec.ps0 inside the minute
`call_wall_strength_meas` levels · measured	share of the near-money MEASURED book on the measured call wall's strike, 0..1	share	`frames[].g` `frames[].strikes` `frames[].cr0` \|g\| on the frame's cr0 / sum \|g\| over the frame; carried forward
`put_wall_strength_meas` levels · measured	same for the MEASURED-book put wall, 0..1	share	`frames[].g` `frames[].strikes` `frames[].ps0` \|g\| on the frame's ps0 / sum \|g\| over the frame; carried forward
`net_gamma_meas` book · measured	net gamma at spot, MEASURED (tape-signed) book, $M	$M	`ngv_meas` copied
`net_gamma_meas_minus_oi` book · measured − open interest	net_gamma_meas - net_gamma_oi, $M: the tape-signed book against the open-interest book. For measured vs VOLUME compare net_gamma_meas with net_gamma_vol	$M	`ngv_meas` `ngv_conv` ngv_meas - ngv_conv
`net_gamma_meas_flow` book · measured	1-minute change of net_gamma_meas, $M per minute	$M per minute	`ngv_meas` 1-minute difference of ngv_meas
`gamma_pctl_meas` book · measured	net_gamma_meas ranked against the same minute of the prior 20 sessions, 0..100	percentile	`ngv_meas` recomputed over the store from net_gamma_meas, the same way as gamma_pctl_vol. Not copied from gpct, so one definition on every session
`dealer_delta_meas` book · measured	dealer delta exposure, MEASURED book, millions of SPX units of delta; + = dealers long delta. No volume-book twin	M SPX units of delta	`dex` copied
`dealer_vega_meas` book · measured	dealer vega exposure, MEASURED book, $M per vol point. No volume-book twin	$M per vol point	`vex` copied
`charm_meas` book · measured	charm pressure, MEASURED book: the dollar delta the book sheds per day from time decay alone, $Bn per day	$Bn of delta per day	`charm` copied
`vanna_meas` book · measured	vanna pressure, MEASURED book: the dollar delta the book gains per 1 vol point of IV, $M per vol point	$M of delta per vol point	`sec.vanna` the last value inside the minute
`dealer_delta_meas_flow` book · measured	1-minute change of dealer_delta_meas, millions of SPX units	M SPX units per minute	`dex` 1-minute difference, within the session
`dealer_vega_meas_flow` book · measured	1-minute change of dealer_vega_meas, $M per vol point	$M per vol point, per minute	`vex` 1-minute difference, within the session
`charm_meas_flow` book · measured	1-minute change of charm_meas, $Bn per day	$Bn per day, per minute	`charm` 1-minute difference, within the session
`max_change_5_meas` book · measured	the largest single-strike gamma change of the MEASURED ladder over the last 5 min, $M, signed -- how hard the busiest strike just moved	$M, signed	`sec.mc.5.value` the last value inside the minute
`max_change_15_meas` book · measured	the same over 15 min, $M	$M, signed	`sec.mc.15.value` the last value inside the minute
`max_change_30_meas` book · measured	the same over 30 min, $M	$M, signed	`sec.mc.30.value` the last value inside the minute
`concentration_meas` book · measured	the same Herfindahl on the MEASURED book, 0..1	Herfindahl	`frames[].g` sum of (\|g\| / sum \|g\|)^2 over the frame's strikes; carried forward
`abs_gamma_meas_over_gross_oi` gross · measured ÷ unsigned OI	sum \|measured gamma\| / sum unsigned OI gamma: how much of the convexity the tape-signed book accounts for. Low = the sides cancel. Changes scope with book_width	ratio	`frames[].g` `frames[].ga` sum \|g\| / sum ga over the frame
`abs_gamma_below_meas` gross · measured	sum \|measured gamma\| on strikes within 30 pt BELOW spot, $M/pt	$M/pt	`frames[].g` `frames[].strikes` `spot` sum \|g\| on strikes in [spot - 30, spot); carried forward
`abs_gamma_above_meas` gross · measured	the same within 30 pt ABOVE spot, $M/pt	$M/pt	`frames[].g` `frames[].strikes` `spot` sum \|g\| on strikes in (spot, spot + 30]; carried forward
`abs_gamma_below_meas_chg15` gross · measured	change of abs_gamma_below_meas against the frame 15 min earlier (negative = the book below thinned)	$M/pt, signed	`frames[].g` `frames[].strikes` `spot` abs_gamma_below_meas minus its value in the latest frame at least 15 minutes older
`abs_gamma_above_meas_chg15` gross · measured	change of abs_gamma_above_meas over 15 min (positive = the book above thickened)	$M/pt, signed	`frames[].g` `frames[].strikes` `spot` abs_gamma_above_meas minus its value in the latest frame at least 15 minutes older
`abs_gamma_below_meas_chg30` gross · measured	change of abs_gamma_below_meas over 30 min	$M/pt, signed	`frames[].g` `frames[].strikes` `spot` the same against the frame at least 30 minutes older
`abs_gamma_above_meas_chg30` gross · measured	change of abs_gamma_above_meas over 30 min	$M/pt, signed	`frames[].g` `frames[].strikes` `spot` the same against the frame at least 30 minutes older
`gamma_0dte_meas` expiry · measured	MEASURED gamma, today's expiry only ($M)	$M	`frames[].g0` sum of g0 over the frame; carried forward
`gamma_next_meas` expiry · measured	MEASURED gamma past today -- the book that survives the close ($M)	$M	`frames[].g` `frames[].g0` sum of (g - g0); carried forward
`gamma_next_share_meas` expiry · measured	share of the MEASURED book that outlives today (0..1)	share	`frames[].g` `frames[].g0` sum \|g - g0\| / sum \|g\|
`range_left_fcst` vol · measured (adjustment)	FirmTape's FORECAST of the range left to the close, index points: realised vol over the last 30 min scaled to the minutes left, adjusted by how high net_gamma_meas stands against the prior 20 sessions. A model, not an implied number	index points	`exp_range` copied

OPEN INTEREST

cp x morning open interest. Spotted by: struct, conv or _oi in the payload (flip_struct, ngv_conv, cr0_oi), _oi in LAB.

Payload fields 12

field	meaning	unit	since
`ga_backfill` Top level	Provenance stamp of the 2026-09-02 backfill that added the unsigned book `ga/ga0` to older sessions: `{at, oi, expiries, contracts, applied}`.	object	sessions to 2026-08-21
`ngv_conv` Minute series	Net cp x OI gamma at spot, narrow scope (0DTE plus three expiries, not 0DTE only).	$M per index point	all sessions
`frames[].go` `frames[].go0` `frames[]`	cp x OI gamma per strike.	$M per index point	all sessions
`frames[].vo` `frames[].vo0` `frames[]`	cp x OI vanna per strike.	$M of delta per vol point	all sessions
`frames[].co` `frames[].co0` `frames[]`	cp x OI charm per strike.	$Bn of delta per day	all sessions
`frames[].ga` `frames[].ga0` `frames[]` · open interest (unsigned)	Gross book: abs(gamma) x OI, calls and puts added. No sign, so no flip and no walls.	$M per index point	all sessions (backfilled 2026-09-02 on sessions to 2026-08-21)
`frames[].oc` `frames[].op` `frames[].oc0` `frames[].op0` `frames[]` · open interest (contracts)	Open interest per strike, calls (`oc`) and puts (`op`) apart, no greek and no sign: the plain count of the chain. The page's OI COUNT ladder (net, calls beside puts, or the sum). Whole numbers.	contracts	SPX sessions built from 2026-10-08 (live and rebuilt). SPX only: feedd's ticker frames (src/gateway/feedd/src/book.rs) do not carry them (payload_schema.json frame_spx_only)
`sec.flip_struct` `sec`	OI-book flip, narrow scope, raw root.	index points	all sessions
`sec.ngv_conv` `sec`	Net OI gamma at spot, narrow scope.	$M per index point	all sessions
`sec.cr0_oi` `sec.ps0_oi` `sec`	Peak positive / negative OI gamma strike within +/-2.5%, 0DTE only.	strike	all sessions
`sec.cr0_oi_book` `sec.ps0_oi_book` `sec`	The same peaks over the narrow-scope OI book (they can pick a 3-5 DTE wall).	strike	all sessions
`orders[].oi` order ledger	Each leg's open interest that morning, aligned with `lg` (four at most); a leg the OI file does not carry is `null`, and the field is `null` when no leg is known. The ORDERS panel marks an order larger than one of them.	contracts	ledgers rebuilt from 2026-10-11

LAB columns 15

column	meaning	unit	built from
`flip_oi_dist` levels · open interest	spot - the OPEN-INTEREST-book flip, points (+ = above it)	points, signed	`spot` `sec.flip_struct` spot - the last sec.flip_struct inside the minute
`call_wall_oi_dist` levels · open interest	spot - the OPEN-INTEREST-book call wall at 0DTE, points (+ = above it)	points, signed	`spot` `sec.cr0_oi` spot - the last sec.cr0_oi inside the minute
`put_wall_oi_dist` levels · open interest	spot - the OPEN-INTEREST-book put wall at 0DTE, points (+ = above it)	points, signed	`spot` `sec.ps0_oi` spot - the last sec.ps0_oi inside the minute
`net_gamma_oi` book · open interest	net gamma at spot, call-minus-put OPEN-INTEREST book over four expiries, $M (not comparable with the 0DTE books' level)	$M	`ngv_conv` copied
`gross_oi_total` gross · unsigned OI	sum of \|gamma\| x open interest over the near-money window, $M per index point. Its scope follows book_width (3 expiries, 21 on widened sessions)	$M/pt	`frames[].ga` sum of ga over the frame; carried forward
`gross_oi_at_spot` gross · unsigned OI	the unsigned OI book on the strike nearest spot, $M/pt	$M/pt	`frames[].ga` `frames[].strikes` `spot` ga on the strike nearest spot, NaN when it is more than 3 pt away; carried forward
`gross_oi_above` gross · unsigned OI	the heaviest unsigned OI strike ABOVE spot, $M/pt	$M/pt	`frames[].ga` `frames[].strikes` `spot` the largest ga on a strike above spot; carried forward
`gross_oi_below` gross · unsigned OI	the heaviest unsigned OI strike BELOW spot, $M/pt	$M/pt	`frames[].ga` `frames[].strikes` `spot` the largest ga on a strike below spot; carried forward
`gross_oi_above_dist` gross · unsigned OI	points from spot up to that heaviest strike above, always positive	points, always positive	`frames[].ga` `frames[].strikes` `spot` that strike - spot
`gross_oi_below_dist` gross · unsigned OI	points from spot down to the heaviest strike below, always positive	points, always positive	`frames[].ga` `frames[].strikes` `spot` spot - that strike
`gross_oi_near_share` gross · unsigned OI	share of the unsigned OI book within +-10 pt of spot, 0..1 -- how much convexity sits on top of price	share	`frames[].ga` `frames[].strikes` `spot` sum of ga within 10 pt of spot / sum of ga
`gross_oi_top3_share` gross · unsigned OI	share of the unsigned OI book on its three heaviest strikes, 0..1	share	`frames[].ga` the three largest ga / sum of ga
`gamma_0dte_oi` expiry · open interest	OPEN-INTEREST gamma, today's expiry only ($M)	$M	`frames[].go0` sum of go0 over the frame; carried forward
`gamma_next_oi` expiry · open interest	OPEN-INTEREST gamma past today ($M)	$M	`frames[].go` `frames[].go0` sum of (go - go0); carried forward
`gamma_next_share_oi` expiry · open interest	share of the OI book that outlives today (0..1) -- the OI book is roughly half next-expiry, so this one really moves	share	`frames[].go` `frames[].go0` sum \|go - go0\| / sum \|go\|

NO BOOK

market quantities, calendar and bookkeeping: no dealer weight. Spotted by: spot, iv, straddle, volume, events, axes.

Payload fields 52

field	meaning	unit	since
`day` Top level	Session date, ET.	`YYYY-MM-DD`	all sessions
`built` Top level	Build stamp of the code that wrote the file; a `-live` suffix means the live engine wrote it.	string	all sessions
`minutes` Top level	The minute axis every minute series is aligned to.	`HH:MM`	all sessions
`step` Top level	Minutes between ladder frames.	5	all sessions
`exp_n` Top level	Post-0DTE expiries the session was archived with (14 to 21; 21 on nearly every session since early 2023).	count	all sessions
`lad_n` Top level · volume, OI	Post-0DTE expiries the frames' VOLUME and OI books sum over. Absent means 3.	count	since 2026-08-24 (29 sessions)
`lad5` Top level · all (display)	The ladder's display scale floor: median of the 5 heaviest strikes over the closing 30 minutes of the 5 prior sessions, keyed `<frame key>\|<slice>` (keys `g v c d e go vo co gv vv cv ga`, slices `all`, `z0` = 0DTE, `nx` = after 0DTE). Cosmetic, changes no measurement.	frame units	since 2022-04-14 (absent on 2019-06-19)
`spot` Minute series	SPX, solved from put-call parity on the option chain (median of C - P + K e^(-rT) over the 10 strikes nearest the crossing).	index points	all sessions
`vwap` Minute series	Not volume-weighted: the running mean of the minute `spot` since the open (an index has no volume).	index points	all sessions
`b2u` `b2l` Minute series	`vwap` plus / minus 2 running standard deviations of `spot`.	index points	all sessions
`skew` Minute series	25-delta risk reversal on the 0DTE chain: put IV minus call IV. Positive = puts bid over calls. `null` unless both wings bracket 25 delta. Calendar-time IV.	vol points	all sessions
`skewpct` Minute series	The same skew as a ratio: how much richer the 25-delta put is than the call.	percent	all sessions
`straddle` Minute series	ATM 0DTE straddle mid (call mid + put mid at the strike nearest spot). Batch days read the 5-minute chain and carry each value forward, so it steps every 5 minutes.	index points	all sessions
`atmiv` Minute series	ATM implied vol backed out of `straddle`: straddle / (0.7979 x sqrt(T) x spot), T on the trading-time clock.	vol points (percent)	all sessions
`div30` Minute series	Change in `atmiv` over the last 30 minutes.	vol points	all sessions
`rvi15` `rvi30` `rvi60` Minute series	Realised over implied range: the high-low range of the per-second spot over the trailing 15 / 30 / 60 minutes, divided by the straddle-implied range over the same horizon (2 x straddle x sqrt(H / minutes left)). Above 1 = the market is moving more than it priced.	ratio	all sessions
`now.spot` `now`	Last minute's `spot`.	index points	all sessions
`now.rv30` `now`	Realised vol of the last 30 one-minute log returns, annualised on the trading-time clock (sqrt(252 x 390)). The only place `rv30` is published. A 30-observation sigma: its 95% interval is x0.80 to x1.34.	percent	all sessions
`now.sessions_in_pool` `now`	Prior sessions in the `gpct` pool (up to 20).	count	all sessions
`ladder.strikes` `ladder`	Strike axis.	strike	all sessions
`frames[].t` `frames[]`	Frame minute.	`HH:MM`	all sessions
`frames[].idx` `frames[]`	Index of `t` in `minutes`.	integer	all sessions
`frames[].strikes` `frames[]`	Strike axis of this frame.	strike	all sessions
`frames[].ivc` `frames[].ivp` `frames[]`	Mean 0DTE implied vol of the call / put at each strike (the smile). Calendar-time clock. `null` where the strike has no 0DTE quote.	vol points (percent)	all sessions
`vol.strikes` `vol`	Strike axis (fixed for the session).	strike	all sessions
`vol.t` `vol.idx` `vol`	Frame minute and its index in `minutes`.	`HH:MM`, integer	all sessions
`vol.cu` `vol.pu` `vol`	Multi-leg volume no package signing could reach (unsigned).	contracts	all sessions
`sec.secs` `sec`	Second axis.	`HH:MM:SS`	all sessions
`sec.step` `sec`	Seconds between rows (1 on every session).	seconds	all sessions
`sec.spot` `sec`	Parity spot solved per second.	index points	all sessions
`sec.vol` `sec`	0DTE contracts printed in that second, as recorded live. Only on sessions the live engine recorded.	contracts	38 sessions, since 2026-08-03
`day` `sym` `min_premium` order ledger	Session, symbol, premium floor.	date, string, $
`asof` order ledger	Time of the newest order, ET.	`HH:MM:SS`
`built` order ledger	When the ledger was rebuilt, ET.	`HH:MM:SS`
`orders[].oid` order ledger	Order id.	string
`orders[].t` order ledger	When the parent order printed, ET.	`HH:MM`
`orders[].kind` order ledger	Structure from package reconstruction (single, vertical, ...).	string
`orders[].legs` order ledger	Up to three legs as a label, e.g. `7740P/7750P`.	string
`orders[].lg` order ledger	Up to four legs as `[strike, "C"\|"P", customer side]`, side +1 bought, -1 sold, 0 unknown.	array
`orders[].exp` order ledger	Expiry carrying most of the order's gamma.	`Mon D`
`orders[].size` order ledger	Contracts in the parent order.	contracts
`orders[].net` order ledger	Premium.	$M
`orders[].em` order ledger	Each leg's strike from spot in expected moves to the close (2 x ATM straddle); `null` unless every leg expires that day.	EM units
`orders[].sh` order ledger	The order's share of its busiest strike's session volume.	percent
`orders[].le` order ledger	Each leg's own expiry, aligned with `lg` (four at most). `exp` is one expiry for the order; an order on two expiries names each leg's contract with this. Without it the ORDERS panel reads every leg as `exp`.	`YYYY-MM-DD` list	ledgers rebuilt from 2026-10-11
`stats.n` `stats.net` order ledger	Orders above the floor, their total premium.	count, $M
`built` `from` `to` `earn_to` calendar file	Build time (UTC), calendar span, last date with historical earnings.
`days.<date>[]` calendar file	Events on that ET date: `t` clock (`HH:MM`, `""` for all day), `k` kind (`fomc macro earn opex early closed fed auction ff`), `n` name, `i` impact (`high med low`), optional `f` forecast, `p` prior, `u` source link.
`stats_built` `stats_measure` calendar file	When the statistics were built and the measure's definition.
`ledger` calendar file	The pre-registered FOMC forward ledger (`registered`, `built`, `in_sample`, `rows`).
`next_earn.<SYM>` calendar file	Next earnings date `d` and hour `h` per covered stock.
`blackouts` calendar file	Fed blackout windows as [start, end].

LAB columns 54

column	meaning	unit	built from
`min` time	minutes since the 09:30 open (0-389)	minutes since 09:30	`minutes` the row's index on the minute axis
`minutes_to_close` time	minutes left to the session's last minute: 15:59 on a full day, 12:59 on a half day	minutes to the last minute	`minutes` (rows in the session - 1) - min
`spot` price	SPX spot price	index points	`spot` copied
`spot_mean` price	the running mean of spot since the open (an equal-weight mean of minute prices; NOT a VWAP, no volume weights it)	index points	`vwap` copied
`spot_mean_dist` price	spot - spot_mean, points	points, signed	`spot` `vwap` spot - vwap
`spot_sd` price	one running standard deviation of spot since the open, points (the half-width of the band over 2)	points	`b2u` `b2l` (b2u - b2l) / 4
`spot_mean_z` price	spot's distance from its running mean in running SDs (+3 = stretched 3 SD above). The 'sigma stretch' of the chart's band	running SDs, signed	`spot` `vwap` `b2u` `b2l` (spot - vwap) / spot_sd
`spot_mean_up2` price	the band's upper edge, spot_mean + 2 running SDs, price	index points	`b2u` copied
`spot_mean_dn2` price	the band's lower edge, spot_mean - 2 running SDs, price	index points	`b2l` copied
`ret1` price	spot now minus spot one minute ago, points (the last 1-minute candle's move)	points, signed	`spot` spot minus the previous minute's spot, within the session
`ret15` price	spot now minus spot 15 minutes ago, index points (the move just made; positive = drifted up)	points, signed	`spot` spot minus spot 15 minutes earlier, within the session
`ret30` price	spot now minus spot 30 minutes ago, index points	points, signed	`spot` spot minus spot 30 minutes earlier, within the session
`rv_1m` price	intra-minute realised vol from the per-second spot, annualised in % -- comparable to atm_iv; a choppy vs smooth minute	annualised %	`sec.spot` `sec.secs` standard deviation of the per-second log returns inside the minute x sqrt(252 x 6.5 x 3600) x 100; NaN under 5 seconds
`book_width` expiry · volume and OI scope	how many post-0DTE expiries the session's volume and OI books sum over -- a provenance stamp, not a market state: 3 on sessions archived narrow, 21 on sessions archived wide (the ones from 2026-08-24 on, which carry lad_n); read it per session. Use it ONLY to keep a rule on gamma_next_* / gamma_next_share_* (vol or oi) or gross_oi_* inside one definition; never as a market condition	count	`lad_n` lad_n, 3 when the payload has none
`atm_iv` vol	0DTE ATM implied vol, % annualised on the trading clock, backed out of the ATM straddle (pair with top_pctl for an 'IV rank'). It climbs ~3x from open to close by construction, so compare at a fixed clock	IV, % (trading clock)	`atmiv` copied
`atm_iv_chg1` vol	1-minute change of atm_iv, vol points	vol points per minute	`atmiv` 1-minute difference, within the session
`atm_iv_chg30` vol	atm_iv now minus atm_iv 30 minutes ago, vol points -- the IV leg that gives back a correct realised-vol call. Mostly the clock: prefer it inside a min window	vol points	`div30` copied
`straddle` vol	the 0DTE ATM straddle price, index points: the market's expected absolute move to the close	index points	`straddle` copied
`straddle_pct` vol	the ATM straddle as a % of spot -- the expected move, comparable across days	% of spot	`straddle` `spot` straddle / spot x 100
`em_used` vol	\|spot - the session's first spot\| / the opening expected move (2 x the first ATM straddle): the EM drawer's 'used so far', the same number. Above 1 = the opening move is spent, the chart's CHOP	ratio	`spot` `straddle` \|spot - the session's first spot\| / (2 x the session's first straddle)
`range_ratio_15` vol	realised RANGE over the last 15 min (per-second high minus low) / the range the straddle implies for 15 min; > 1 = moving more than priced	ratio	`rvi15` copied
`range_ratio_30` vol	the same realised/implied RANGE ratio over 30 min -- the strongest single forecast of the next 30 min's range this programme found	ratio	`rvi30` copied
`range_ratio_60` vol	the same realised/implied RANGE ratio over 60 min	ratio	`rvi60` copied
`rv30_over_atm_iv` vol	realised vol over the last 30 one-minute returns / atm_iv, both annualised on the trading clock, plain ratio: 1 = the tape moves as fast as the straddle prices; > 1 = the straddle is cheap -- 'VRP / realised vs implied' as a ratio	ratio	`spot` `atmiv` standard deviation of the last 30 one-minute log returns x sqrt(252 x 390) x 100, over atmiv
`skew_25d` vol	25-delta put IV minus 25-delta call IV, vol points (calendar clock)	vol points (calendar clock)	`skew` copied
`skew_25d_ratio` vol	25-delta skew as a RATIO: put IV over call IV minus one, in PERCENT (median ~10, 95th pctl ~25) -- NOT a 0-1 percentile; 'rich skew' is skew_25d_ratio >= 15 or top_pctl on it	percent	`skewpct` copied
`skew_25d_chg1` vol	1-minute change of skew_25d, vol points	vol points per minute	`skew` 1-minute difference, within the session
`volume` tape	contracts traded per minute across the 0DTE chain: from the raw tape where the archive has it; otherwise the snapshot's, which on older sessions is a 5-minute increment spread evenly over its minutes	contracts per minute	`raw 0DTE tape` `sec.vol` `vol.cb` `vol.cs` `vol.cu` `vol.pb` `vol.ps` `vol.pu` contracts per minute from the raw tape; without it, volume_snap
`volume_pctl` tape	traded-VOLUME percentile of this minute vs the same minute of the prior 20 sessions (0-1) -- 'heavy for this time of day'; volume, NOT implied vol	percentile	`volume` share of the same minute in the prior 20 sessions at or below this one, 0..1; NaN under 10 comparands
`volume_snap` tape	volume as the snapshot carries it: per-second prints summed on newer sessions, a 5-minute increment spread per minute on older ones	contracts per minute	`sec.vol` `vol.cb` `vol.cs` `vol.cu` `vol.pb` `vol.ps` `vol.pu` sec.vol summed per minute; without it, each 5-minute increment of the cumulative vol block spread evenly over its minutes
`ev_high_impact` calendar	1 on a day with a high-impact release: FOMC, CPI, payrolls or PCE; else 0	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_fomc` calendar	1 on an FOMC decision day (statement 14:00, press conference 14:30)	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_cpi` calendar	1 on a CPI day (08:30, before the open)	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_payrolls` calendar	1 on a payrolls day (08:30, before the open)	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_pce` calendar	1 on a PCE inflation day (08:30, before the open)	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_ism` calendar	1 on an ISM day, manufacturing or services (10:00, inside the session)	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_earnings` calendar	1 when a mega-cap (NVDA, AAPL, MSFT, ...) reports today, before the open or after the close	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_opex` calendar	1 on a monthly or quarterly options expiration day	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_pre_open` calendar	1 if a high/medium release came out before 09:30, so the session already trades AFTER the number	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_minutes_to` calendar	minutes until the next scheduled event inside the session (FOMC 14:00, ISM 10:00, FOMC minutes); NaN when none is ahead, so a condition on it selects event days by itself	minutes	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_minutes_since` calendar	minutes since the last scheduled event inside the session; NaN before the first one	minutes	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_opex_quarterly` calendar	1 on the quarterly third Friday (March, June, September, December): options, the ES and NQ futures and the index rebalances expire or settle together. About 4 days a year	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_expiry_stack` calendar	1 when two or more expiry-class events share the date: every quarterly third Friday, the January LEAPS expiry, a month-end that meets an index rebalance or the monthly VIX settlement	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_quarter_end` calendar	1 on the last session of a quarter, the SPX quarter-end expiry. About 4 days a year	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_month_end` calendar	1 on the last session of a month, quarter-ends included	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_vix_week` calendar	1 on every session of the week that holds the MONTHLY VIX settlement (the Wednesday thirty days before the next month's third Friday), about one week in four	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_futures_roll` calendar	1 on the Thursday eight days before a quarterly expiry, when ES and NQ volume moves to the next contract. About 4 days a year	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`ev_leaps` calendar	1 on the January third Friday, when the long-dated LEAPS cycle expires. One day a year	flag	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`days_to_opex` calendar	weekdays left to the next third-Friday options expiration, 0 on the day itself, up to about 24; holidays count as weekdays	weekdays	`/snapshots/events.json` from the day's calendar rows; a day the calendar does not know reads 0, clocks NaN
`pulse_minutes_since` pulse	minutes since the last minute the tape marked as market-moving (a one-minute index move of 4 sigma of the trailing half hour, or 3 with the dealer book moving too); NaN before the first such minute, so 'pulse_minutes_since <= 5' means 'within five minutes after a tape shock'	minutes	`pulse.firmtape.com marks` from the tape-marked minutes of the session (not a payload field); a day without marks reads 0, clocks NaN
`pulse_z` pulse	the spot z-score of that last marked minute, in sigmas of the trailing half hour; NaN before the first	sigma	`pulse.firmtape.com marks` from the tape-marked minutes of the session (not a payload field); a day without marks reads 0, clocks NaN
`pulse_flash` pulse	1 if that marked minute had two or more public news desks flashing beside it (a headline-driven move), else 0	flag	`pulse.firmtape.com marks` from the tape-marked minutes of the session (not a payload field); a day without marks reads 0, clocks NaN
`pulse_book` pulse	1 if that marked minute had no headline and no calendar event, the dealer book moved on its own; else 0	flag	`pulse.firmtape.com marks` from the tape-marked minutes of the session (not a payload field); a day without marks reads 0, clocks NaN
`pulse_count` pulse	how many tape-marked minutes the session has had so far; 0 before the first	count	`pulse.firmtape.com marks` from the tape-marked minutes of the session (not a payload field); a day without marks reads 0, clocks NaN