The journey → Season 1
Season 1 · breakout · book mechanical
This is a research log, not a product.
A backtest is the claim being tested; the live paper book is a conformance test of the implementation, not evidence about the strategy.
At this size the running total is statistically meaningless and always will be — it is published so the code can be checked, not so the result can be copied.
No capital is at risk. Nothing here is a recommendation.
When a session's range is the narrowest of the last seven, the following session often expands out of it. Entry is a stop above the narrow day's high, the stop is that day's low, and the target is twice its range — so all three levels are fixed before the order is placed.
Source: Day Trading with Short Term Price Patterns and Opening Range Breakout — Toby Crabel
Restated in full every time, so a reader arriving mid-season can check the result against the claim rather than take it on trust.
Crabel published this as an intraday opening-range breakout on futures. We run it on daily bars as a swing bracket, long-only, on US equities, with a hard stop at the prior bar's low. The stop is ours, not the author's - the source pattern carries no published stop rule. Universe trimmed from 16 to 12 at plan time: CSX, HAL, BMY and KMI measured below the INV-09 liquidity floor on the IEX feed on 2026-08-03.
This matters more than the running total. Several of these methods were published with no stop loss at all; a bracket order cannot express that, so the stop is ours, not the author's — and a stop changes a mean-reversion system materially.
Hypothesis. Narrow-range-7 bars occurring above the 200-day SMA precede directional expansion often enough that a stop-entry above the narrow bar's high, risking that bar's low and targeting twice its range, yields positive expectancy in R across 12 liquid US large-caps at 2 concurrent slots. Crabel's original claim rests on pre-1990 futures data; independent re-tests put NR7 alone near a coin flip and attribute what edge exists to the trend filter. This season tests the filtered version.
Falsifier. Conclude retired on whichever comes first: 20 closed trades at cumulative expectancy <= 0R, or book drawdown >= 10 percent ($200). If fewer than 10 trades close in 84 days, conclude inconclusive - too few signals to judge, not evidence of failure.
Both were written before the season started. Deciding afterwards what would have counted as failure is how a research log turns into a highlight reel.
Modern testing is more sober: positive expectancy in trending regimes, weak to negative in chop, and highly sensitive to whether a trend filter is applied. Treat NR7 alone as close to a coin flip; the 200-day filter above is the actual candidate.
0 signals evaluated, 0 refused by the risk rules before reaching the broker, 0 trades placed, 0 settled.
Machine-readable: /api/public/journey.