=== /gex — DEALER GAMMA REGIME · NVDA === QUESTION: Is the market set up to BLAST in a direction, or chop in jagged broken shards? VERDICT: SET UP TO CHOP IN JAGGED BROKEN SHARDS Dealers are long gamma where price is trading, so hedging pushes AGAINST moves. JAGGED, BROKEN-SHARD CONFIGURATION, SHORT-DATED. Dealers are long gamma at this price, so hedging leans against moves — selling strength, buying weakness — which is what compresses ranges and turns breakouts back. Price is away from the gamma centre of mass, so the damping comes from the local book rather than from a magnet level. Most of this gamma expires today: the stabiliser stops existing at the close. 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 CALMING moves (dealers long gamma — hedging leans AGAINST the tape, a shock-absorber). dealer gamma at spot: +$340.7M per 1% move. flip (regime boundary): 200.74, 5.83% below spot 213.16. 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 5.02% (front at-the-money implied vol 79.63% (expiry 2026-08-28) / sqrt(252)) gamma-conditional range median 2.92%, middle half 2.26% to 3.64% (quintile 2 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 18.77% (put wall 190.00 to call wall 230.00) last session delivered 1.86% (5-min sampled) = 0.37x the breakeven -- 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. 240.00 +26.84 pts +12.59% RESISTANCE secondary call concentration (2nd largest) 10.1% of book 87,232 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. 230.00 +16.84 pts +7.90% RESISTANCE call wall (largest) 12.4% of book 106,877 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. 225.00 +11.84 pts +5.55% RESISTANCE secondary call concentration (3rd largest) 7.3% of book 63,424 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. 213.16 <<<< PRICE IS HERE 200.00 -13.16 pts -6.17% SUPPORT secondary put concentration (2nd largest) 4.1% of book 35,709 net contracts A further band of put hedging below price, lighter than the wall. The same mechanics apply, at smaller size. 195.00 -18.16 pts -8.52% SUPPORT secondary put concentration (3rd largest) 2.9% of book 25,418 net contracts A further band of put hedging below price, lighter than the wall. The same mechanics apply, at smaller size. 190.00 -23.16 pts -10.87% SUPPORT put wall (largest) 10.7% of book 91,979 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. 220.00 +3.21% REGIME BOUNDARY — not support, not resistance The strike where the running total of dealer gamma changes sign. Above it dealer hedging leans against moves and damps them; below it hedging leans with moves and amplifies them. It is a boundary between two behaviours, NOT a level price is repelled by — and this page reads the live regime from the gamma right around spot rather than from this level, because spot-versus-flip is much weaker than it is usually treated as. 220.48 +3.44% 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, 67.6% 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 213.16 as of 2026-08-25T20:23:01+00:00 (post_close) net gamma in +/-1% band +$66.3M across 5 strikes [THE REGIME MEASURE] = 92% of the gross gamma in that band, netting LONG (flat below 20%) net dealer gamma at the single nearest strike +$3.0M at 212.50 whole-book gamma +$350.4M call wall 230.00 (7.90% above) put wall 190.00 (10.87% below) walls by CONTRACT COUNT call 230.00 put 190.00 [the stable read] gamma centre of mass 220.48 (3.43% from spot) zero-gamma cross 220.00 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 -- 67.6% — share of the FORWARD book's gross gamma sitting at its front expiry (2026-08-28) [drives the 'fast decay' in the verdict; fast at 50%] 0.0% of the gross gamma on the SESSION book expired at 2026-08-25's close 0% of the at-spot gamma disappeared with that expiry (+$3.0M on the session book -> +$3.0M on the forward book) -- WHERE THIS SITS VERSUS ITS OWN BANKED HISTORY -- net dealer gamma at spot 33.6 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 67.6 % 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 100.0 percentile The walls are about as far apart as they get. The corridor of concentrated dealer hedging is unusually wide, which leaves a large span around price where the option book exerts little mechanical pull either way. distance to the zero-gamma cross 51.4 percentile Spot sits about its usual distance from the zero-gamma cross. The crossing strike is neither unusually near nor unusually far. 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 2026-08-25T21:50:47Z 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] server clock now 2026-08-26T01:56:15Z (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 2026-08-25T21:50:47Z 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 276 known, 0 MISSING (excluded, never zeroed), 30 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 2026-08-26T01:56:13Z · full JSON: /apps2/gex/api.php?sym=NVDA&full=1