This article explains a classic trigger and its lineage. It does not claim the trigger is profitable on this market standalone; our own evidence says the surrounding system decides that. Nothing here is investment advice. Decisions are yours.
The 20-day high is the simplest breakout trigger in the canon: buy when price exceeds its highest point of the last twenty days. No indicator, no weighting, one lookback. Its persistence across seventy years of markets is not despite the simplicity; the simplicity is the mechanism.
The lineage
The trigger descends from Richard Donchian, the mid-century pioneer of trend following, whose channel, the running N-day high and low, made "new high ground" a computable event rather than a feeling. Its most famous stress test came in the 1980s Turtle experiment, when Richard Dennis taught a 20-day-breakout-family system to a room of novices to settle a bet that trading could be taught; the rules, later published, are a museum piece of the whole philosophy: mechanical entries on channel breaks, volatility-scaled position sizes, wide exits, and no discretion anywhere. The details aged; the architecture, simple trigger plus strict risk plus regime tolerance, is recognizably the ancestor of every serious trend system since, including the family our own research works in. The BTC trend evidence measures how that family has held up.
Why the dumb trigger resists dying
Three reasons, in ascending importance. It is a clean implementation of time-series momentum: a 20-day high is the asset telling you, in its own prices, that recent demand beat every seller of the last three weeks, an anomaly the momentum evidence finds across asset classes. It is self-calibrating: "highest high of N days" scales automatically with any price level and any volatility, needing no re-tuning as an asset reprices from cents to hundreds. And, decisively for this site's philosophy: it has one parameter, which means almost no surface for overfitting. A trigger with one honest knob cannot be quietly molded around historical accidents the way a six-parameter entry can, which is why simple rules that survive are more trustworthy than complex rules that shine: the complex ones had more ways to cheat on the exam.
The 24/7 recalibration
Crypto keeps the trigger and quietly changes its meaning: twenty daily bars here span twenty calendar days, against the twenty trading days, four full weeks, an equity chart means, the same lookback drift that hits moving averages in a market that never closes. The channel also inherits this market's wick problem: an N-day high can be set by a single cascade spike that no close confirmed, so implementations must decide, in writing, whether the channel is built from highs or closes and whether a break must CLOSE above the level or merely touch it. Those two boring decisions change the trigger's behavior more than moving the 20 to a 30 does, which is exactly the kind of specification detail published methodology exists to pin down.
What the trigger does not do
Our transplant experiment is the cleanest possible caution against trigger worship: the same breakout logic produced a losing ungated system and a surviving regime-gated one on identical data. The trigger contributed identically to both; the regime filter that answers the chop problem, the stops, and the sizing formulas decided which one lived. A breakout trigger is a sentence, not a book: it says when to pay attention, and everything that determines profitability, when to be exposed at all, how much, where the exit lives, is written elsewhere in the system. The 20-day high keeps working as a component because it is honest, cheap, and unoverfittable. Sold alone, as a strategy, it is a museum piece with a price tag. Decisions are yours.