SomerQuant Daily — Options Flow + Dark Pool, 16 September 2026

SomerQuant Daily · Wednesday, 16 September 2026 · EOD options flow & dark-pool research · SomerQuant

This edition is published late: the desk was dark while AZTMM.com moved to SomerQuant, and Wednesday’s session was never written up on the night. Every figure below is the 16 September session exactly as measured, unrevised, with nothing carried back from the sessions that followed.

16 September 2026 — EOD read. The tape gave back a good morning. SPY opened at 759.50, worked up to 761.67 by 11:45, then lost twelve handles into a 749.60 low at 15:25 and closed at 754.05, down 0.44% on the session. The options tape told the same story on the same clock: market-wide net call-minus-put premium peaked at +$191M at 11:45 and fell to a −$169M trough at 15:25, crossing zero exactly once at 14:55.

SPY opened at 759.50. The first fifteen minutes framed a 758.72 to 759.69 range, and the session spent the rest of the morning above it, reaching the 761.67 high in the 11:45 bar. The afternoon undid all of it. The low of 749.60 printed in the 15:25 bar, and the close came back to 754.05, roughly four and a half points off the bottom. Fourteen SPY dark-pool prints cleared the regular session, and every one of them carried the vendor’s contingent or derivative-priced condition, so all fourteen are drawn hollow. Eleven of those fourteen carry the same 757.432 tag, the largest of them 228,601 shares for $173.1M at 15:35; that repeated figure is a price struck off a derivative leg, not a level where stock changed hands, and it should not be read as support. The largest print of the fourteen was 412,000 shares for $312.8M at 759.26, again contingent. Of the five heaviest premium shelves, three sat inside the day’s range: 760 at −$86.7M, 759 at −$32.3M and 750 at −$77.7M. Two sat off the price scale below it, 745 at −$65.5M and 730 at −$29.7M.

The tide opened at +$20M and never looked troubled before lunch. It peaked at +$191M at 11:45, the same five-minute bar that carried the session high, then bled steadily through the early afternoon. Of the 79 five-minute readings, 65 sat above zero and 14 below, and the series crossed zero exactly once, at 14:55. From there it fell hard to a −$169M trough at 15:25, the same bar as the 749.60 price low, and recovered only partly to finish at −$98M. One crossing in a full session is a clean reversal rather than a choppy one, and the clock on it matches the price to the bar.

15 September’s radar — OI confirm

Verdict describes the open-interest update only, never the contract.

Six of the nine contracts we froze on Tuesday came back as new open interest on Wednesday morning, one partially, and two did not. The cleanest of them was the SKHY June 2027 $225 calls, where 851 contracts of Tuesday volume produced 658 contracts of fresh open interest, a ratio of 0.77. The two misses are worth naming as plainly as the hits. SMH October $545 puts traded 20,222 contracts on Tuesday and added only 1,903 of open interest, a ratio of 0.09, which is the shape of volume that closed or rolled rather than opened. AMZN November $290 calls did worse than that: open interest fell by 1,988, so whatever moved on Tuesday was on balance an exit. This is the point of running the loop at all. Volume is not positioning until the next morning’s update says it is.

What happened

Wednesday was a session of two halves joined at 11:45. The morning was orderly. SPY held above its opening range from the first half hour, the tide of net options premium built without interruption to +$191M, and there was nothing in the flagged-events screen that read as defensive. The afternoon reversed the whole thing in about three and a half hours, and it did so without a single false start: one zero crossing in the tide, one low, one close off it.

The flagged-events screen, which we take as the 50 largest premium lines of the session, was dominated by index paper. The four heaviest lines were all SPX, and three of the four were long-dated calls: 600 February 2027 7450 calls for $29.0M, 500 December 2027 8050 calls for $26.5M and 500 September 2027 7950 calls for $23.5M. Against those sat 500 December 2027 7400 puts for $18.4M. Long-dated index paper on both sides of the book is not a directional statement about Wednesday, and we are not going to dress it up as one.

The nearer-dated paper leaned the other way. SPY October 2 747 puts took 10,980 contracts for $5.8M against open interest of just 266, a volume-to-open-interest ratio of 45.5, and QQQ October 16 713 puts took 4,410 contracts for $6.8M on a descending fill. On the volatility side, VIX December 22 calls printed 98,600 contracts for $20.9M across three floor prints against open interest of 8,938. Those are the lines that match what the price did after lunch.

The dark pool told a different story from the options tape, as it often does on a reversal day. The complete $100M-and-above universe came to $28.818B across 86 prints in 48 distinct names, and 84.3% of that notional was ETF. The three largest names in the universe were IVV at $8.254B, VOO at $5.484B and SPY at $5.136B, and almost all of that came in the closing window. This is index rebalancing plumbing, not conviction.

Why it matters

The alignment between the price and the premium tide is the thing worth keeping. Two independent measurements, one from the tape and one from the options book, put their turn and their extreme in the same five-minute bars. When those two agree that precisely, the reversal is more likely to be a real change of hands than a liquidity accident, and the levels it left behind carry more weight into the next session.

The strike ladder is the second thing. Every one of the ten strikes carrying the largest net premium on Wednesday was put-led, from 760 down to 730, with no call-led shelf in the top ten at all. That is a one-sided book. The nearest strike where call premium led was 762, at +$13.6M, which is small next to the −$86.7M sitting at 760. A book stacked that heavily on one side does not predict direction, but it does mean the hedging flows around those strikes are asymmetric.

The third thing is what the dark pool did not do. At $28.818B the $100M universe was well below Tuesday’s $39.81B, and its ETF share jumped from 62.9% to 84.3%. Fewer single names cleared the threshold on a day the market fell. That is the opposite of a distribution tape, and it is worth saying so rather than reading a large closing-auction number as demand.

What to watch into Thursday’s session

• 749.60 — Wednesday’s low, and the level to watch. A close below it would mark the reversal as extending rather than exhausting.
• 761.67, the session high, and 759.69, the top of the opening range. Reclaiming the second and failing the first would leave the day’s whole span intact.
• The 760 strike, which carries −$86.7M of net premium across all expiries, the heaviest shelf on the board.
• The 750 and 745 shelves at −$77.7M and −$65.5M, which sit at and below the low.
• Whether the tide holds above zero in the first hour. It opened positive on Wednesday and stayed there for 65 of 79 readings before it broke.
• The QQQ October 16 713 puts and the SPY October 2 747 puts, both of which are checked against Thursday’s open-interest update in the next edition.

Names on our radar

Every row below is frozen at publication and scored mechanically in the Accountability Ledger at +5 and +21 sessions — misses stay on the page. Options prints are checked against the next morning’s open-interest update in the following session’s post.

Positioning into Thursday

Our positioning gauge reads zero across all three baskets. No leveraged-long, inverse or volatility ETF from the tracked lists appeared in either the $100M dark-pool universe or the flagged-events screen. Zero is information rather than an absence of it: on a session that fell 0.44% and traded twelve handles peak to trough, nobody reached for the leveraged wrappers in size. The hedging that did happen went through index options and through VIX December calls, which is a more deliberate way to do it.

The book itself is stacked on the put side and has been all day. That is not a forecast. It means the market maker inventory around 760, 750 and 745 is one-directional, and moves through those strikes are more likely to be amplified than dampened.

The boards — data appendix

Wednesday’s $100M-and-above universe of $28.82B sits between the two extremes of the week. It is 27.6% below Tuesday’s $39.81B and 55.6% above Monday’s $18.52B. The print count fell from 127 to 86 and the number of distinct names from 85 to 48, so the money that did clear the threshold was concentrated in fewer places. The ETF share is the number that moved most: 84.3% against 62.9% on Tuesday and 46.8% on Monday, the highest of the five sessions on record here. A board that is five-sixths ETF on a down day is index machinery, not single-name accumulation.

The heaviest shelf on the board is 760, which carries $136.6M of call premium against $223.3M of put premium for a net of −$86.7M, and it sits 0.79% above the close. Every one of the ten strikes shown is put-led. The nearest shelf on the other side, where call premium leads, is 762 at +$13.6M, and the largest call-led shelf anywhere in the band is 780 at +$14.1M. Both are small beside the 760 and 750 figures. These totals aggregate every expiry trading at each strike, so they describe where the book is stacked, not what was opened on Wednesday.

Universe. Every board below is computed on the complete set of dark-pool prints of $100M or more on 16 September 2026, retrieved by pagination until a page returned fewer than 50 rows, with canceled prints dropped and exact duplicates removed. That universe is 86 prints, $28,817,678,275 of notional, across 48 distinct names.

Relative size. Relative size is each print’s share count divided by that name’s 30-day average daily volume. The board above is a $100M screen and the table below is a $10M screen, so a name can appear in one and not the other: PNTG’s $23.6M print is the largest relative-size reading of the session and never comes close to the $100M universe, while IVV’s $1.93B print is only 0.30× its own 30-day average. The biggest-print figure in the tile row comes from the $100M board; the unusual-by-size figure comes from this $10M screen.

Cluster watch. Prints grouped by ticker at the same price within 0.1%, two prints or more. Clustered prints are not re-narrated as standalone blocks anywhere above.

Notable data points

1. SPY closed at 754.05, down 0.44% from 757.39, on a session range of 749.60 to 761.67.
2. QQQ closed at 704.72 against 704.54, up 0.03%, essentially unchanged while SPY fell.
3. Market-wide net options premium spent 65 of 79 five-minute readings above zero and crossed zero once, at 14:55.
4. All ten of the SPY strikes carrying the largest net premium were put-led, across all expiries.
5. The $100M dark-pool universe was 84.3% ETF, the highest of the five sessions on record here.
6. Eleven of the fourteen SPY dark-pool prints in the regular session carried the same 757.432 contingent price.
7. VIX December 22 calls traded 98,600 contracts against open interest of 8,938.
8. Six of the nine contracts frozen on 15 September were confirmed by the 16 September open-interest update.

Method note

Flow and dark-pool data sourced from Unusual Whales. MPI score and regime classifier are our internal composite; daily synthesis is AI-assisted from those inputs.

This edition was published after the session it covers. Every pull behind it was scoped to the 16 September session by date or by an explicit 09:30 to 16:00 Eastern window, and every returned row was checked to carry a 16 September timestamp before it was used. No figure from 17 or 18 September appears anywhere in this post.

The internal index feed was not updated for this session. Its most recent reading is stamped 17 September, a later session, so it fails the within-two-sessions test for a 16 September edition and the MPI and regime cells are left as em-dashes rather than filled. No MPI value has been estimated, interpolated or back-computed.

The vendor returned an integer-truncated VIX close of 17 for both 16 and 15 September while both sessions carry decimal highs and lows, so no percentage change is computed from those closes. The session range of 16.40 to 18.94 is printed instead.

Volume-over-OI activity is unconfirmed as new positioning until the next morning’s open-interest update; confirmations are published in the following session’s post. The confirm loop above was run from per-contract daily open-interest history so that the 16 September update is compared with 15 September volume, rather than from the current open-interest snapshot, which would have carried a later session into this edition.

Relative size is a print’s share count divided by that name’s 30-day average volume as reported with the print, computed by SomerQuant; it is not any vendor’s relative-size or relative-strength score.

Option premium by strike is aggregated across all expiries.

The session map is drawn from 5-minute bars; the opening range is the first fifteen minutes; dark-pool prints marked contingent carry the vendor’s contingent or derivative-priced sale condition.

Board baselines are trailing-session averages computed from our own archived pulls; derived indicators are computed by SomerQuant from licensed market data. The flagged-events screen is the 50 most recent lines of $1M or more in premium within the 16 September regular session, and every statement about it is bounded by that depth.

This is research, not advice. Nothing here is a recommendation to buy, sell, or hold any security.

AZTMM HLDGS LLC is not a registered broker-dealer, investment adviser, or FINRA member. All content is retrospective research published for general circulation — not personalized advice, not trade signals. Options involve substantial risk, including losses that may exceed the initial investment. Full disclaimer.

New here? Start Here · Pulse Lab · Accountability Ledger · Trading Academy · Subscribe for daily updates