We looked for a strategy you could actually trade. We didn't find one.
Three rounds after the first search (65 markets; the six passes are where the shelf’s morning move comes from), the predictions written down before each one ran. Round 1 has its own page; rounds 2 and 3 are told here. The short version: the strategies that pass beat the coin flips, and you could not have sat through them. This was before the shelf; the setups that have passed since are on the front page.
One strategy, 31 markets
We took the premarket reversion (the morning move on the shelf) as it is and moved it to each market's own opening time. 26 failed, 5 were close, none passed. Crude oil came nearest: only 1 of 5,000 coin flips beat it on the years it never saw, and it still scored 79, because it made nothing in the years we could see, breaks even at two ticks of cost, and keeps its money in a few late years.
Every result →Picking the exit fresh every year
Could a trader beat the Friday opening-range strategy by choosing its exit time again each year from what had worked so far? No. The 3:30 exit won all sixteen years we tried it on, so the adjusting trader made the same trades as the fixed strategy. Choosing from recent years alone would have lost a fifth of the money on the years never seen, and the pass with it.
Cattle and hogs
Livestock keeps its own hours, so we anchored the strategy to the previous day's close. Live cattle and lean hogs lose on the years they never saw. Feeder cattle came close: 31 of 5,000 coin flips beat it, it still worked when its settings were nudged (87%), and it scored 59, a close call. But 2020 alone was 118% of what it made on the years it never saw, it went 56 months without a new high, and its typical year made $704.
What it means
Every strategy that passed looks the same. It's real, and you couldn't live on it. The money lands in two or three years out of eighteen. Between them you wait 34 to 139 months for a new high, and a typical year makes hundreds of dollars on a strategy whose best year makes tens of thousands. Changing the time didn't fix that. Adjusting the strategy didn't. A new market didn't. Adding a condition usually made it worse, and when it helped, it helped by squeezing the money into even fewer, later years. That isn't a flaw in these strategies. It's what a real edge looks like.
So the strategy you could live on isn't a different strategy. It's the same real strategy, traded small enough to survive its own dry spell, by someone who can tell a dry spell from a dead strategy. Telling those apart was to be the job of a monthly Ledger, planned when this was written in September 2026 and not built as of November 6, 2026: each month it would check the strategy against 5,000 coin flips on the same days and say whether it is still real or has become a coin. The third-Friday strategy read as a coin for seventeen months, then made all of its money. A Ledger would describe. It would never tell you what to do.
We wrote the predictions down before each run. Search No. 2 scores its own line by line; Search No. 1 scores its own in prose, under “Methods honesty”. The predictions for rounds 2 and 3 are not scored on any page.
A fourth round of the same. A new search needs a new question, and the record points at one: can a person hold a real strategy through the years the record says are coming, if it's sized to the dry spell and a Ledger told him the truth each month?
That question, tested forward in public if the Ledger is built: the shelf's setups, one contract each, sized from the record, marked every month. Published either way. As of November 6, 2026 it has not started.
The Shelf →