=== /gex — DEALER GAMMA REGIME · SPY === QUESTION: Is the market set up to BLAST in a direction, or chop in jagged broken shards? VERDICT: SET UP TO BLAST Dealers are short gamma where price is trading, so hedging pushes WITH moves. BLAST CONFIGURATION, AND A DURABLE ONE. Dealers are short gamma at this price with no centre-of-mass magnet nearby, so their hedging pushes with the tape rather than against it. Little of this gamma expires today, which means the amplifying configuration is not a same-day artefact — it carries into the following sessions unless the book changes. DIRECTIONAL LEAN: NONE NO DIRECTIONAL LEAN. Dealer gamma describes the SHAPE of the path — how big and how jagged — not which way it points. The asymmetries that could carry a lean (which wall is closer, where the centre of mass sits, vanna, charm) do not agree strongly enough here to name a side, and on this stack none of them beat simply counting how often sessions close up. -- DOES THE OPTION BOOK CALM OR AMPLIFY THIS MOVE? (regime curve, forward book) -- at spot the book is AMPLIFYING moves (dealers short gamma — hedging runs WITH the tape, an accelerant). dealer gamma at spot: -$3.25B per 1% move. flip (regime boundary): 769.83, 0.51% above spot 765.90. MODELLED, NOT MEASURED. Each contract is repriced with Black-Scholes gamma while its implied vol and days-to-expiry are held FIXED and spot is hypothetically moved (a sticky-strike assumption); r=0; same-day options are floored to 1 day of time so their gamma stays finite (they are removed entirely on the forward book); the contract universe is identical to the per-strike chart (+/-15% of spot, nearest 6 expiries); OI-weighted, never volume; calls +1, puts -1, netted per contract. -- THE RANGE THE OPTION MARKET IS CHARGING FOR -- daily breakeven 0.71% (front at-the-money implied vol 11.28% (expiry 2026-08-26) / sqrt(252)) gamma-conditional range median 1.01%, middle half 0.95% to 1.31% (quintile 3 of banked history, n=24) NOTE: shown for completeness. Part B found the gamma reading adds nothing to the breakeven above. wall-to-wall corridor 0.65% (put wall 765.00 to call wall 770.00) last session delivered 0.31% (5-min sampled) = 0.43x the breakeven -- OVERNIGHT: ES AGAINST THE FROZEN WALLS -- ES has traded down 0.23% since the cash close, which puts SPY at about 764.14. That sits 0.77% below the call wall at 770.00 and 0.11% below the put wall at 765.00. The wall levels themselves are frozen from the last session's settled open interest and do not move until the next session settles. E-mini S&P 500 (ESU6) 7,677.25 179s old as SPY-equivalent 764.14 (-0.23% since the cash close at 765.90) vs frozen call wall 770.00 price is 0.77% below it vs frozen put wall 765.00 price is 0.11% below it vs frozen gamma centre 762.85 price is 0.17% above it CONVERSION: ES shown as SPY-equivalent: ES now 7,677.25, ES at the Tue close 7,695.00, so 765.90 x (7,677.25 / 7,695.00) = 764.14. The move in the future since the cash close, applied to SPY's price at that close — the basis and the index/ETF ratio cancel out, so no ratio is estimated. SCALE: shared — the overlay is in the same units as the wall levels ! The wall levels and the futures price are deliberately from DIFFERENT CLOCKS. Wall levels come from settled open interest, which the clearing house publishes overnight and does not revise during a session — so a wall is a constant until the next settlement, not a stale reading. Futures trade around the clock. Comparing the two is what shows whether price is drifting toward or away from the level while the cash market is shut. ! Overnight futures liquidity is far thinner than the regular session. A wall approached at 03:00 on light volume is not the same event as one approached at 14:00, and neither this page nor the dealer hedging behind the wall treats it as one. measured: overnight bars average 1,075 contracts per 5 minutes against 10,551 in the regular session (10.2%). ! The dealer hedging that would enforce a wall is largely NOT ACTIVE while the cash market is shut, so an overnight approach to a level is not the same event as a daytime one. The caveat about which side dealers are actually on applies here more strongly, not less. ! This overlay adds no predictive power, and this app measured that: when price reached the call wall it closed back below only 37.2% of the time against 38.0% for an arbitrary level the same distance away. The overlay shows WHERE price is relative to a frozen level, nothing more. -- MAJOR SUPPORT & RESISTANCE (forward book) -- These are where dealer hedging CONCENTRATES. They are not levels price is obliged to respect: a big enough directional move overwhelms even a massive wall of gamma. MEASURED: reaching the call wall, price closed back below it 40.2% of the time (n=256); the put wall held 42.8% of 180. Neither beat a level the same distance away with no gamma in it, so these rows are ranked by concentration, not reliability. 800.00 +34.10 pts +4.45% RESISTANCE secondary call concentration (3rd largest) 2.0% of book 22,609 net contracts A further band of call hedging above price, lighter than the wall but closer to the tape than the numbers alone suggest. The same mechanics apply, at smaller size. 785.00 +19.10 pts +2.49% RESISTANCE call wall (largest) 4.0% of book 45,215 net contracts The heaviest call hedging overhead. On the usual assumption that dealers are LONG these calls, staying hedged means selling into rallies as price approaches — which is what would slow a move here. 780.00 +14.10 pts +1.84% RESISTANCE secondary call concentration (2nd largest) 2.6% of book 29,145 net contracts A further band of call hedging above price, lighter than the wall but closer to the tape than the numbers alone suggest. The same mechanics apply, at smaller size. 765.90 <<<< PRICE IS HERE 765.00 -0.90 pts -0.12% SUPPORT secondary put concentration (2nd largest) 3.6% of book 40,709 net contracts A further band of put hedging below price, lighter than the wall. The same mechanics apply, at smaller size. 760.00 -5.90 pts -0.77% SUPPORT put wall (largest) 4.5% of book 50,472 net contracts The heaviest put hedging below. On the usual assumption that dealers are SHORT these puts, staying hedged means buying as price falls toward it — which is what would cushion a move here. 750.00 -15.90 pts -2.08% SUPPORT secondary put concentration (3rd largest) 3.4% of book 37,910 net contracts A further band of put hedging below price, lighter than the wall. The same mechanics apply, at smaller size. 762.85 -0.40% MAGNET — the opposite of a barrier The gamma-weighted centre of the book: the level hedging flow pulls price TOWARD, rather than a level that pushes price away. Price sitting on it is the pinned configuration. ! WHICH SIDE ARE DEALERS ACTUALLY ON? This is the one thing gamma exposure cannot see, and it decides whether a wall brakes or accelerates. Every number here INFERS it from a convention — that customers sell calls and buy puts, so dealers end up long calls and short puts, which is what makes a call wall behave like a ceiling. When that convention is wrong the sign flips with it: if customers are BUYING calls, dealers are SHORT them, and staying hedged means buying INTO a rally through the same strike — the level becomes an accelerant instead of a brake. On the book these levels are drawn from, 28% of the gross gamma sits at the front expiry — the short-dated, retail-facing end of the chain, where that assumption is least safe. Nothing on this page can confirm the direction, so read the labels as where hedging is concentrated, not as which way it will push. ranked by: netted contract count at the strike (call open interest minus put open interest), as a share of the netted contracts across the whole band -- LEVELS (forward book) -- The session has closed, so the same-day contracts in the final snapshot are already dead. This verdict is built on the FORWARD book — the gamma that still exists for the next session. spot 765.90 as of 2026-08-25T20:23:01+00:00 (post_close) net gamma in +/-1% band -$2.21B across 15 strikes [THE REGIME MEASURE] = 49% of the gross gamma in that band, netting SHORT (flat below 20%) net dealer gamma at the single nearest strike +$33.8M at 766.00 whole-book gamma -$2.91B call wall 770.00 (0.53% above) put wall 765.00 (0.12% below) walls by CONTRACT COUNT call 785.00 put 760.00 [the stable read] gamma centre of mass 762.85 (-0.40% from spot) zero-gamma cross — NOTE: the zero-gamma cross is a LEVEL, not the regime boundary. The regime above is read from the NET GAMMA IN THE +/-1% BAND around spot. -- HOW FAST THIS REGIME DECAYS -- 28.0% — share of the FORWARD book's gross gamma sitting at its front expiry (2026-08-26) [drives the 'slow decay' in the verdict; fast at 50%] 22.7% of the gross gamma on the SESSION book expired at 2026-08-25's close 105% of the at-spot gamma disappeared with that expiry (-$668.9M on the session book -> +$33.8M on the forward book) -- WHERE THIS SITS VERSUS ITS OWN BANKED HISTORY -- net dealer gamma at spot 47.5 percentile Dealer gamma at spot is unremarkable — it sits about where it usually does. Hedging flow is neither strongly absorbing moves nor strongly adding to them, so the option book is not the dominant force in the tape at this price. share of the book at the front expiry 28.0 % of the book The share of gamma at the front expiry is unremarkable for this book. A meaningful slice resets when it expires and the rest carries forward. room between the walls 15.6 percentile The walls are closer together than usual. The corridor of concentrated hedging around spot is tight. last session delivered vs what options charged 0.6 percentile The tape delivered far less movement than the option market charged for. Whoever paid for that volatility got a session that never came close to justifying it, and whoever sold it collected the difference. -- PART B: THIS APP MEASURING ITS OWN CLAIM -- THESIS HELD: NO NOT ESTABLISHED. Dealer gamma tracks the volatility regime, but it did not beat what the option market had already priced. - RAW, the thesis looks right: the most negative-gamma quintile was followed by a 1.82% median range and the most positive by 1.48% (pooled rho -0.098, p=0.0023). - CONTROLLED, it disappears. Measure the same range in units of what options charged for it (range / IV breakeven) and the relationship is rho 0.034, p=0.30 — nothing. Negative-gamma sessions ARE high-implied-vol sessions: median breakeven fell from 1.70% in the short-gamma quintile to 1.38% in the long-gamma quintile. Gamma was reading the thermometer, not adding to it. - The BLAST-vs-SHARDS part specifically shows nothing: path efficiency (how much of the day's travel became net progress) against gamma is rho -0.010, p=0.75. - The FAILED-BREAKOUT prediction runs the wrong way. Long dealer gamma is supposed to produce MORE failed breakouts; measured, long gamma faded 32.6% of breakout attempts and short gamma faded 35.5%. - No directional signal beat the base rate: 52.6% of next sessions closed up, and not one of the four asymmetries did better. + ONE reading did keep a signal after the calendar was removed, and it is the DECAY RATE rather than the level: the 0DTE share of gamma. Ranked inside each (root x weekday) cell — which removes the weekday effect completely, not just its linear part — a higher 0DTE share went with the next session realising MORE range than options had charged for: pooled rho 0.098 (n=847 across 35 cells), and POSITIVE on all 7 roots tested. It is called SUGGESTIVE and not established, for three reasons: only 1 of 7 roots reach significance on their own, the pooled p-value is overstated because the roots share market days, and the sample is five months of one year. QUINTILES OF DEALER GAMMA AT SPOT -> WHAT THE NEXT SESSION DID (pooled, within-root cuts) Q n next range middle half breakeven ratio path eff failed breakouts 1 200 1.82% 1.19-2.94% 1.70% 1.07 0.469 37.1% of 178 2 192 1.75% 1.23-2.56% 1.81% 1.00 0.480 30.9% of 165 3 192 1.89% 1.20-2.76% 1.77% 1.07 0.407 35.3% of 184 4 192 1.70% 1.01-2.72% 1.54% 1.08 0.509 32.0% of 178 5 192 1.48% 0.92-2.35% 1.38% 1.09 0.438 33.3% of 168 Q1 = most NEGATIVE dealer gamma (blast end) Q5 = most POSITIVE (jagged/pinned end) ! POOLED SIGNIFICANCE IS OVERSTATED. The roots share the same market days and SPY/SPX/QQQ are close to the same trade, so a pooled n is not that many independent observations — the effective n is nearer the per-root session count. ! SPXW is excluded from every pooled set: same underlying as SPX, so its outcome rows are duplicates. ! THE WEEKEND ARTEFACT. breakeven divides by sqrt(252) TRADING days, so a Friday reading is compared against a Monday session carrying three calendar days of risk. Measured below as friday_artefact. Friday rows carry weekend_span=true. ! 2026 only. The OI-weighted gamma archive starts 2026-03-02 (the OCC open-interest purchase window), so there is no multi-year sample and no other market regime in it. ! This measures ASSOCIATION on banked sessions. It is not a trading result and carries no costs, slippage or capacity. ! DIFFERENT CONSTRUCTION FROM THE LIVE READ. These hit rates are measured on the DAILY DOLLAR-GEX AGGREGATE, because that is the only per-session gamma that exists back through 2026-02. The live panel on this page is built on per-strike NETTED CONTRACT COUNTS from the 5-minute chain recorder, which only starts 2026-08-05. Dollar gamma at a strike explodes as spot approaches it and collapses as it leaves, so the aggregate is exactly the weaker construction — this table does not validate the live measure exactly, and it is not claimed to. -- HOW FRESH, AND WHAT IS STALE BY DESIGN -- options chain 2026-08-25T20:23:01+00:00 expected every 5m [held at the last regular-hours reading — correct once the market closes, not a failure] profile build 2026-08-25T21:50:14Z expected every 10m [held at the last regular-hours reading — correct once the market closes, not a failure] open interest 2026-08-25 expected every 1d [settled once overnight — a day behind by market structure, not by neglect] daily history 2026-08-24 expected every 1d [settled once overnight — a day behind by market structure, not by neglect] ES futures 2026-08-26T01:15:00+00:00 expected every 5m server clock now 2026-08-26T01:17:59Z (skew baseline only — NEVER a data timestamp) chain_day 2026-08-25 chain_asof 2026-08-25T20:23:01+00:00 chain_phase post_close chain_built 2026-08-25T21:50:14Z chain_cadence every 5 minutes in regular trading hours, plus a pre-open and a post-close bracket open_interest PREVIOUS session, settled overnight by the clearing house. It does not move intraday — a market-structure fact, not a defect. history_last 2026-08-24 history_gex_window from 2026-03-02 (the OCC open-interest purchase); continues past 2026-08-04 from the live chain recorder rather than ending there validation_built 2026-08-25T09:20:01Z open-interest census 1674 known, 0 MISSING (excluded, never zeroed), 390 genuine zeros -- STANDING CAUTION -- Gamma is not destiny. A large enough directional move overwhelms even a very big wall of dealer gamma — the hedging flow described here is one force among several, and it is the one that gets run over when something bigger arrives. Everything on this page describes how the option book is positioned; none of it is a statement about what price does next. generated 2026-08-26T01:17:56Z · full JSON: /apps2/gex/api.php?sym=SPY&full=1