Strategies — October 2026

Strategies and rule sets, whether coded or read by hand. A rule appearing here does not mean it works. Each entry says how far it has been tested.

October 4 — A 90-minute bias with a one-minute trigger: the bias carried no information

The idea, as it was shown to us: split the day into 90-minute blocks, take a bias from the previous block, and enter on the first strong one-minute displacement bar in that direction. We wrote the rules down before running anything: four New York blocks, continuation and fade versions, a bar body of at least 1.5 times the recent average range closing near its extreme, entry on the next bar, a stop beyond the displacement bar, a 1.5R target and an exit at the end of the block.

Over about 4,100 trades on NQ, the result before costs was zero in all four versions, and after costs every version lost slightly. None of them beat entries in a random direction on the same bars. Following the bias and fading it came out the same, which says the bias carries no information at this horizon.

The first run showed two trades instead of four thousand. A unit error (seconds read as minutes) flagged almost every trade as a data gap. It was caught by counting how many trades survived each step, and fixed before anything was reported. The news-driven part of the idea was not tested: there are too few scheduled releases a year to measure it. This tests one mechanical reading of the idea, not the judgement of the person who trades it.