How the Hunt grades

A rung is a quality tier, strongest first (Top Shelf → Bottom Shelf). Business quality sets it: a name whose business hasn't been shown sound sits near the bottom whatever its chart, and the top tiers need a sound business and a clean trend. It is a grade of quality, not an order of how likely a move is — the default view groups names by this grade, and you can re-sort them however helps.

A rung is a quality grade — the tier a name earns on our checks, business quality first — not a forecast that it will rise. We group by grade, never by how likely a move is.

A name's 1-year high is a price the market actually traded — a plain fact, not a forecast and not a price target.

The obvious way to rank a screener is “more methods agree → stronger.” We measured that on our own strategies across ten years of data, and we refuse to do it. Here is exactly what the measurement found:

1. The methods aren't independent. Measuring how our strategies fire together, the trend and breakout methods collapse into ONE family — so “N methods agree” counts the same bet many times over, not many separate opinions.
2. More agreement did NOT mark a better result — it leaned negative at every horizon we tested, never positive. At our primary three-month horizon the interval barely included zero; at six months it excluded zero, though that six-month window is also the most thinly-powered. The direction, not any single cell, is what the decision rests on.
3. The more of the trend family firing on a name — the “strongly-extended” signature — the WORSE it tended to do next (about −2.2 percentage points at three months, interval excluding zero). This reproduces an out-of-sample audit from the system we ported.
4. A “how well-tested” sort carried no forward content either — it leaned slightly negative too.

So the Hunt never sorts by agreement. It groups names by the quality grade and lets you re-sort on plain facts. We measured our own “more methods agreeing means stronger” signal, found it doesn't hold, and refuse to rank by it — and we publish that measurement. That refusal is the point.

• Graded only on names in the sandbox universe today — names that failed or delisted aren't here, so the picture flatters the survivors.
• One market, graded on one day's data. A grade is a snapshot, not a track record.
• The measurement behind the no-ranking decision has thin statistical power (about 18 independent long-horizon windows in ten years). The direction — never positive — is robust across several independent checks, but a precise effect size is not pinned down.
• Crypto can't validate findings drawn from stocks, so the agreement signal is suppressed there: a stock-derived posture never orders a crypto row.