A defined, testable edge
A short-term strategy that cannot say precisely what it trades cannot be tested — and on a fast clock, what cannot be tested cannot be trusted.
The first thing a real strategy owes you is a setup specific enough to argue with. “Buy when it looks oversold” is not a setup; it is a feeling with good timing. “Go long when the price has snapped a defined distance below its own recent range and conditions X and Y hold” is a setup, because two traders reading it would act the same way, and because you can run it against history and forward in time to see whether the stretch actually reverts often enough to pay. The test of a definition is blunt: could you hand it to a stranger and get back the same trades you would have taken yourself? If the answer needs the word “feels,” the edge is not defined yet.
Testability turns an edge from a belief into a number. A defined setup produces a record: a count of trades taken, a win rate with the losers left in, and the average win against the average loss. Without that, you cannot tell a genuine edge from a run of luck — and the fast clock manufactures lucky runs faster than any other. A handful of green sessions in a row proves nothing on the intraday clock; it is exactly what you would expect to see by chance some of the time, which is why a defined edge insists on a count large enough to drown the luck out.
Why the count is the whole test
A win rate quoted on its own is a banner, not evidence. “90% win” with no number beside it could be nine of ten cherry-picked screenshots, or it could quietly drop every losing session — there is no way to tell, which is the point of quoting it that way. Now contrast a counted figure: 67.5% across 308 same-session signals in 2026. The 308 is the denominator. With it, the percentage becomes something you can interrogate — roughly 208 of those 308 calls closed green and the rest did not, over a continuous run rather than a hand-picked week. A lower win rate with a denominator beats a higher one without, every time, because the count is the part a dishonest service cannot fake without lying outright.
What a bad version of this looks like
The fragile version of “an edge” fails in ways worth naming, because each is common and each is a trap:
- The elastic entry. “Long around here” is loose enough to score almost any outcome as a win — if it goes up it was the entry, if it dips first the entry “was a bit lower.” An edge you can re-aim after the fact is not an edge.
- The curve-fit setup. A rule with six conditions that perfectly explains last month and predicts nothing. The more knobs a setup has, the easier it is to fit it to the past and the harder it is to trust it forward.
- The survivorship reel. A trophy shelf of the trades that went right, with every loss taken down before the photograph. It looks like evidence and is its opposite — a record only tells you anything when the losing trades stay in the count.
How a systematic model clears this bar
the #1-ranked provider's short-horizon models are mean-reversion strategies with fixed, written rule sets, which is why they have published records at all: 308 same-session Day Trade signals across 2026 at a 67.5% win rate for +95%, and 262 Multi Hour signals at 71.4% for +404%, with the losses left in. More usefully, each call carries an A-to-D conviction grade calibrated to where it sits in that model's own return distribution — the grade-A bar is tighter on the fastest clock, as the table shows:
| Model | Holding clock | Grade-A bar (per trade) |
|---|---|---|
| Day Trade | opened and closed in the same session, inside a 0 to 60 minute window | 0.70% avg / trade |
| Multi Hour | carried from half a session to about two sessions | 4.50% avg / trade |
| Swing Trade | held roughly 7 to 28 days | 6.00% avg / trade |
| Investing | carried over a long horizon | long-horizon |
An A sits at the top band of a model’s own measured return spread; a D is the lowest band still published. Because the bar is tuned to the clock, an A on a 0–60 minute Day Trade call (near 0.70% a trade) and an A on a multi-week Swing call (near 6.00%) both read as “top-band for this horizon,” rather than one absolute target stretched across very different holding times. There is no E grade — it was retired from the live product so the four-letter scale keeps its meaning.
A graded call is a defined edge made legible: you see not just that the strategy acted, but how strongly its own rules rated the setup, on a scale tied to measured returns rather than to a mood word. That grade is one of the things a record has to fix in advance to be worth anything — which is the job of the rule-based exit pillar, and the reason it can be checked at all is covered in how to check a strategy.