One rule, every market we have, nothing changed.

The question

Does the premarket reversion exist outside index futures, with the rule unchanged? Not "does something like it work on gold" with the clock moved and the threshold retuned; that is a different question and a different search, and it is counted as one. This is the narrowest honest version: the rule as locked in Teardown No. 3, run everywhere, once.

The rule, as run on every market

Family A, the anchored move. Reference price at 08:00 ET; at 09:30 ET, if price has moved at least 0.40% from the reference, trade against the move at the next bar's open; market out at 10:15 ET. Both directions, one contract, $4.20 round-turn commission, one tick per side, seen era declared from 2023, version #1. Engine v0.16, 5,000 coin-flip twins per market, the same seed base for all.

What this page does not do

It selects nothing. A table of every result with nothing left out carries no selection bias; the selection happens afterward, by a rule fixed before the first run, and that rule is quoted below. The declaration was filed to the corpus before the batch started.

Every market, sorted by score

Seen net is 2023 onward, the years anyone would have developed on. Unseen net is 2008 through 2022. Unseen p is the share of 5,000 coin-flip twins, same days, same costs, random direction, that did as well or better on the unseen years. Breakeven is the friction in ticks per side at which the full-window result reaches zero. Records for every row are in the corpus folder for this search.

The shortlist, by the declared rule

On seen-era columns only: seen net above zero, at least 60 seen trades, seen profit factor at least 1.20, and seen net still above zero at two ticks per side. 6 of 65 scored markets clear it: EW, RTY, NQ, ES, MNQ, MES. The unseen columns were printed for every market and consulted for nothing in this step.

The book, as the first answer

The four shortlisted indices, one contract each, summed on the calendar: every trade of every part, ordered by day, with the coin-flip twins run across the whole book and livability measured on the combined curve under the five numbers the declaration fixed. Each part was admitted only after its own 5,000 twins rebuilt exactly from its trade log, so the book is built from the records above and nothing was re-run.

Correlation of daily profit and loss, measured before anything was summed. A day without a trade counts as zero on the union; the shared column is the days both parts traded.

The lowest pair on shared days is 0.730, the highest 0.928. The book: 2,690 trades, net +$286,064, 0 of 5,000 twins as good or better on the full window and 0 of 5,000 on the unseen years. Then the five numbers.

Not livable under the declared numbers, on the no-new-high stretch and the years to recoup the drawdown. Four indices at one clock are one bet taken four times. Every pair moves together on the days both trade, and the book inherits the years every part slept through: its longest stretch without a new equity high is 34 months, ending 2014-08, and the drawdown takes 3.85 median years to recoup. Adding a fifth index cannot fix that. Livable needs something that earns in those years, which is a different clock, a different class or a different rule, and that is what the second batch looks for.

Known limits, stated before the run

1. The clock is an index clock. On energy, grains, rates and currencies the session opens somewhere else, and a rule that reads 08:00 to 09:30 ET is reading the wrong hour on most of them. A per-class clock is the second batch, and it is counted as a second look per class.
2. One tick a side is honest on the big four and generous on a contract that trades two thousand a day. The shortlist rule requires the seen-era result to survive two ticks; the three-tick line is in every record.
3. Data quality is whatever the integrity review of September 3 found for each file. The engine prints exclusions and a density caution per file; a market with either is marked in the records and its number is not used for anything but this table.
4. Sixty-five results printed means sixty-five looks taken. The unseen-era p of any shortlisted market is reported at lineage #1, the honest count for a selection that never consulted the unseen columns, and at #65, the conservative count for having printed them all.

Methods honesty — the correction ledger

1. The declaration's cost clause was worded loosely ("survives the three-tick line"). The reading applied, fixed before the table was built and stated in the builder script, is: seen-era net at two ticks per side above zero. The records are on file for anyone who wants to apply a different one.
2. Four treasury records (TN, US, ZN, ZT) failed the builder's first reconciliation gate by one to eight cents: the record writes each trade rounded to the cent and the net unrounded, and on a contract whose tick is 1/64 or 1/128 of a point the half-cents add up. The gate was widened to half a cent per trade, the four rows entered the table, and the engine's trade log is to be written unrounded so this never needs a tolerance again.
3. The examiner predicted before the run: the equity-linked contracts following the big four; the rest splitting between marginal rates and currencies, rejected energy, metals and grains, and a few markets with no trades at all. The next entry scores it.
4. Tally, scored against the table above. The equity-linked contracts followed the big four: MidCap 92, EAFE 92, Russell 92, Nasdaq 90, S&P 89, Dow 89, every index with a full history validated, and the Nikkei came in at 72 with zero of 5,000 twins on the unseen years and still failed, because it lives inside 1.7 ticks of spread. Hit. Energy, metals, grains, softs and livestock rejected: hit. Rates and currencies marginal: miss, they were rejected almost to a name, the E-mini yen at 50 the one exception. A few markets with no trades: miss, all sixty-five traded. The examiner's own test for the clock caution, "more than ten markets outside equities validate," came out at zero, so the caution stands and the per-class clock is the next batch, not a formality.
5. Five markets show zero seen-era trades while their files run to 2026 (HE, LE, GF, LBS, ZR: hogs, cattle, feeders, lumber, rice). Read September 6, before SEARCH 02: the engine sets its anchor only on a bar within ten minutes after the anchor minute and otherwise skips the day, and since 2021 these five files have no bar before 09:30 (10:00 for lumber), so no day anchored and the seen era is empty; their unseen trades are pre-2021 days that happened to carry a stray overnight bar near 08:00. Livestock has therefore not been run at its own clock, and cannot be by Family A as written, whose anchor is a clock time and not the prior close; that is an engine item on the board (D-10), and the five rows stand as printed.
6. The shortlist is four US equity indices and two micros of the same indices. Micros are the same evidence at a tenth the size, not new evidence; the four indices are correlated with each other, and measuring that correlation is the first job of the book, before any equity curve is summed.
7. The YM probe was run first and predicted to validate in the 80s. It scored 89, 475 trades. That one was a hit, and it was the easy one.
8. Book scoring, Sep 6: the examiner predicted before the run that the four-index book would pass breakeven and the positive-windows share, fail the eighteen-month stretch and the two-median-year recoup, with the largest-year line a coin flip, and that the daily profit of NQ and ES would correlate between 0.5 and 0.8 on shared days. The verdict lines were hits; the correlation came in at 0.918, above the range, a miss in the direction that makes the book worse.

What comes next

The second batch ran on September 6, and its page is up: Search No. 2. As declared here, the second batch is the per-class clock: the same rule with the anchor, signal and exit set to each asset class's own session, declared as one look per class before any run, published either way. The book scoring path now exists, so whatever that batch validates outside the indices goes into the book beside them and is measured the same way.

The standing challenge

If you trade a version of this rule on a market in the table with a different clock, write the clock down in numbers and submit it at thinridge.io. A changed clock is a changed rule, it is scored as one more look, and it will be published beside this table either way.