SomerQuant · The Tally
91 of 212 resolved directional reads aligned with what followed
92 were invalidated and 29 closed without a clean read, while 17 are still inside their window. Regime reads alone aligned in 33 of 48 resolved cases (69%), 21 sessions. 565 observations that stated no direction are recorded, not scored.
The Tally · scored after every close · as of the 2 October 2026 close
By the numbers
Two tiers, each with its full denominator
Directional reads regime, radar and watch lines that stated a direction
91 of 212 resolved directional reads (43%), with 92 invalidated and 29 unresolved beside them. Windows: 21 sessions for regime and radar reads, the next session for watch conditions, each against the band set out in the method below.
A directional read is a published regime label, radar line or watch condition that stated up, down or flat. It is scored once its window closes, and hit, invalidated and unresolved always add up to the resolved total.
Observations and confirmation counted, never scored as hits or misses
An observation is a published line that stated no direction, such as a level, a print or a flagged contract; it stays in the table, linked to its edition, and is recorded, not scored. Radar confirmation is a separate published measure of whether a flagged contract came back as new open interest the next session, and it says nothing about price.
The record
Every read, and how it closed
Each row is a dated line from a published edition, frozen at publication and never edited. The table opens on the 229 directional reads; the Observation tier holds the other 565 lines, each linked to the edition that published it. Statements and notes are shown as the ledger records them.
The table fills from the data carried in this page when scripts run.
| Published | Type | Tier | Ticker | Statement | Status | Resolved | Move after | Note |
|---|---|---|---|---|---|---|---|---|
| With scripts off, every row is in the edition that published it; the scored record is published on this site. | ||||||||
Misses first puts invalidated reads at the top, then unresolved, then hits, then open. Each published date links to the edition the read was frozen from.
By month
Resolved directional reads by month
Hit, invalidated and unresolved directional reads, stacked by the month each window closed. The figure above each bar is that month’s resolved total. Observations are not in this chart.
Source: SomerQuant Research, The Tally (794 rows from editions published 5 June to 2 October 2026); scored after the close with licensed, delayed market data; as of the 2 October 2026 close.
| Month resolved | Hit | Invalidated | Unresolved | Resolved |
|---|---|---|---|---|
| June 2026 | 11 | 18 | 5 | 34 |
| July 2026 | 40 | 43 | 18 | 101 |
| August 2026 | 30 | 17 | 2 | 49 |
| September 2026 | 8 | 14 | 4 | 26 |
| October 2026 | 2 | 0 | 0 | 2 |
Radar confirmation
Radar confirmation, session by session
Across the last 20 Dailies that published an open-interest check, 3 September 2026 to 2 October 2026, 132 of 191 flagged contracts confirmed, 27 were partial and 32 did not confirm.
Each Daily checks the contracts flagged in the previous session against the next morning’s open-interest update. Ratio: next-session open-interest change divided by prior-session volume; 0.40 or above confirmed, 0.10 to 0.40 partial, below 0.10 or a fall in open interest not confirmed. The verdict describes the open-interest update only, never the price.
Source: SomerQuant Research, the open-interest check printed in each Daily, newest first; 17 September 2026 published no check, so the 20 sessions run from 3 September 2026. Licensed options data, delayed.
The 20 sessions as a table
| Edition | Contracts flagged | Confirmed | Partial | Not confirmed | Checked |
|---|---|---|---|---|---|
| 2 October 2026 | 1 October | 5 | 0 | 5 | 10 |
| 1 October 2026 | 30 September | 9 | 1 | 0 | 10 |
| 30 September 2026 | 29 September | 6 | 3 | 0 | 9 |
| 29 September 2026 | 28 September | 5 | 3 | 2 | 10 |
| 28 September 2026 | 25 September | 9 | 1 | 0 | 10 |
| 25 September 2026 | 24 September | 7 | 1 | 1 | 9 |
| 24 September 2026 | 23 September | 5 | 3 | 1 | 9 |
| 23 September 2026 | 22 September | 6 | 1 | 3 | 10 |
| 22 September 2026 | 21 September | 6 | 2 | 2 | 10 |
| 21 September 2026 | 18 September | 7 | 2 | 0 | 9 |
| 18 September 2026 | 17 September | 6 | 0 | 3 | 9 |
| 16 September 2026 | 15 September | 6 | 1 | 2 | 9 |
| 15 September 2026 | 14 September | 7 | 1 | 1 | 9 |
| 14 September 2026 | 11 September | 7 | 2 | 1 | 10 |
| 11 September 2026 | 10 September | 4 | 2 | 4 | 10 |
| 10 September 2026 | 9 September | 6 | 1 | 3 | 10 |
| 9 September 2026 | 8 September | 8 | 1 | 1 | 10 |
| 8 September 2026 | 4 September | 9 | 1 | 0 | 10 |
| 4 September 2026 | 3 September | 5 | 1 | 1 | 7 |
| 3 September 2026 | 2 September | 9 | 0 | 2 | 11 |
Method
How scoring works, and what is not claimed
How scoring works
Regime reads are scored on SPY over 21 sessions: a Bull read aligns when SPY is higher 21 sessions later, a Bear read when it is lower, and a Neutral read when the move stays inside plus or minus 2%. Radar lines that state a direction align at a move of 1% or more that way over 21 sessions, are invalidated at 1% or more the other way, and stay unresolved inside that band. Watch conditions that state a direction are scored on the next session with a plus or minus 0.25% band. Returns use dividend-adjusted closes and count trading sessions, not calendar days, and scoring runs after the close.
What is not claimed
These figures are not returns, not a record of trades, not a forecast and not a performance record of any strategy. They count how written observations lined up with the market afterwards, out of the resolved totals printed beside them. There is no backtest and no audited study on this site. Invalidated and unresolved rows stay on the page beside the hits, each linked to the edition that published it.
Rows from editions published 5 June to 23 September 2026 come from the public ledger file in the aztmm-mpi-data repository, with their statuses as recorded there. Rows from 24 September onward were taken from the editions as published and are carried in this page. The Method page sets out the inputs.
SomerQuant is a publication of AZTMM HLDGS LLC. This edition is impersonal market commentary and education, published on a fixed schedule to the general public. It is not tailored to any reader, it is not investment, legal or tax advice, and it does not offer or solicit any security or strategy. No trade signals. SomerQuant does not know your circumstances and cannot tell you what to do. Market data is sourced from licensed third-party providers and is delayed. AZTMM HLDGS LLC and its principal may hold positions, long or short, in instruments discussed, and may change them at any time without notice. No compensation has been or will be received from any issuer, broker or third party for the coverage of any security. Full terms, disclaimer and editorial policy at somerquant.com.
© 2026 AZTMM HLDGS LLC. AZTMM HLDGS LLC is not a registered broker-dealer, investment adviser or FINRA member. Full disclaimer and terms.