Koryu · 黄龍
Koryu
Dial · 針

Which weather is it?
Fixed rules, one answer a day.

Market breadth · 市場90% above MA50 · 10% below
RegimeRisk-onBTC is stacked above its 50-day; the long board is live.

What Strategy v2.3 reads

Candidate · in shadow · holds no capital
RISK-ONthe long book would be live · ensemble 4/5 voting risk-on

The candidate’s two gates are mutually exclusive by construction, so they collapse to three states: a five-model hidden Markov ensemble decides RISK-ON; the v2.3 short gate decides RISK-OFF (a downtrend held five straight days, BTC below its 50-day with the 20-day below too, plus two stand-down floors: Fear & Greed at 15 and a DVOL capitulation cap); and neither means NEUTRAL. Nothing here is traded; the record that decides whether it earns the job lives on the shadow page, with the full reasoning in strategy v2.0 in shadow.

v2.3 regime · last 20 days2026-08-15 to 2026-09-03
RISK-ON · long book liveNEUTRAL · both legs in cashRISK-OFF · short book live
Long gate · ensemble votes (risk-on at 3/5)2026-08-15 to 2026-09-03
2026-08-15: 0/5 voting risk-on2026-08-16: 0/5 voting risk-on2026-08-17: 0/5 voting risk-on2026-08-18: 0/5 voting risk-on2026-08-19: 0/5 voting risk-on2026-08-20: 1/5 voting risk-on2026-08-21: 4/5 voting risk-on2026-08-22: 4/5 voting risk-on2026-08-23: 4/5 voting risk-on2026-08-24: 4/5 voting risk-on2026-08-25: 4/5 voting risk-on2026-08-26: 4/5 voting risk-on2026-08-27: 4/5 voting risk-on2026-08-28: 4/5 voting risk-on2026-08-29: 4/5 voting risk-on2026-08-30: 4/5 voting risk-on2026-08-31: 4/5 voting risk-on2026-09-01: 4/5 voting risk-on2026-09-02: 4/5 voting risk-on2026-09-03: 4/5 voting risk-on
dashed: 3/5 votes opens the long book · a 2px stub is a real 0
Short gate · BTC vs its MA50 (confirmed short at 5 days below)2026-08-15 to 2026-09-03
2026-08-15: BTC -0.8% vs MA502026-08-16: BTC -1.1% vs MA502026-08-17: BTC +1.3% vs MA502026-08-18: BTC +1.4% vs MA502026-08-19: BTC +8.3% vs MA502026-08-20: BTC +13.6% vs MA502026-08-21: BTC +21.2% vs MA502026-08-22: BTC +18.7% vs MA502026-08-23: BTC +19.2% vs MA502026-08-24: BTC +20.6% vs MA502026-08-25: BTC +19.4% vs MA502026-08-26: BTC +19.5% vs MA502026-08-27: BTC +20.7% vs MA502026-08-28: BTC +16.6% vs MA502026-08-29: BTC +16.7% vs MA502026-08-30: BTC +15.4% vs MA502026-08-31: BTC +16.2% vs MA502026-09-01: BTC +14.0% vs MA502026-09-02: BTC +13.5% vs MA502026-09-03: BTC +18.6% vs MA50
dashed: the MA50 itself · dark bars below zero accrue the 5-day confirmation
Short gate · four conditions, all must hold2 of 4 true · 2026-09-03
Gate 1 · Downtrend structureFALSEBTC +18.65% (+5.19) vs MA50 · MA20 +9.20% (+0.81) vs MA50 · needs both below
Gate 2 · Held 5 straight daysFALSEstreak 0 of 5 (flat) · accrues only while gate 1 holds
Gate 3 · Fear floorTRUEFear & Greed 65 (+2) · stands down under 15
Gate 4 · Volatility capTRUEDVOL percentile 0.136 (+0.074) · stands down at 0.36+
all four true = short book live (RISK-OFF) · a floor going false means the panic is too deep to short · bracketed value = change since the prior bar
Latest completed bar 2026-09-03: ensemble 4/5 votes on the long gate (opens at 3) · the short gate reads 2 of 4 true above. Distances, not signals.

The published dial

Attested · recomputable by hand
RISK-ONas of 2026-09-04 · on the 2026-09-03 UTC close
BTC close vs MA50$81,270 vs $68,498 (above)
BTC MA20 vs MA50$74,797 vs $68,498 (stacked)
Breadth (board coins above own MA50)90.0% (>= 40%)
RISK-ON: BTC close > MA50 and MA20 > MA50 and breadth >= 40%
RISK-OFF: stack false or breadth < 25%
NEUTRAL: otherwise
Breadth: % of the board's coins above their own MA50
Dial verdict · last 20 days2026-08-15 to 2026-09-03
Breadth · % of the board above its own MA5023% to 90%
2026-08-15: 23.0%2026-08-16: 18.0%2026-08-17: 25.0%2026-08-18: 25.0%2026-08-19: 53.0%2026-08-20: 74.0%2026-08-21: 92.0%2026-08-22: 92.0%2026-08-23: 92.0%2026-08-24: 91.0%2026-08-25: 86.0%2026-08-26: 88.0%2026-08-27: 89.0%2026-08-28: 81.0%2026-08-29: 81.0%2026-08-30: 72.0%2026-08-31: 80.0%2026-09-01: 85.0%2026-09-02: 86.0%2026-09-03: 90.0%
dashed: 40% risk-on threshold · 25% risk-off threshold

Data: Binance USDT spot daily klines (completed UTC days) + CoinGecko market-cap ranking. Measured through 2026-09-03 (completed UTC days). Formulas: the dial methodology. The 20-day history is the published formula RECOMPUTED over public bars, not a record of what was printed each day: only 2026-07-13 onward was published and attested, and breadth is measured against today’s board, so earlier days are indicative. The BTC stack is exact throughout. A regime reading is a description, not a forecast. Decisions are yours.

Questions

What is the Koryu regime dial?

A rules-based risk-on / risk-off instrument for the crypto market. It combines BTC's trend stack (close above the 50-day average, 20-day above 50-day) with market breadth (the share of the board's coins above their own 50-day average) and prints one of three states: RISK-ON, NEUTRAL, or RISK-OFF.

What are the exact thresholds?

RISK-ON requires the BTC stack to be true and breadth at or above 40%. RISK-OFF prints when the stack is false or breadth is below 25%. Everything between is NEUTRAL. The thresholds are fixed and published; they do not get tuned quietly.

Why are two regimes shown, and which one is real?

Both are real; they answer different questions. The dial is the published instrument: fixed thresholds, breadth included, recomputable by hand from public data. The v2.3 regime is what our candidate strategy runs on: a five-model hidden Markov ensemble for the long side, and for the short side a downtrend held five straight days with two stand-down floors (extreme fear, capitulation volatility). The candidate is in shadow and holds no capital, so the dial is what the site publishes and attests. Where they disagree, the disagreement itself is the information.

Why do the two ever disagree?

Mostly over the difference between not-bullish and bearish. The dial prints RISK-OFF whenever BTC's stack is false, which folds a flat market in with a falling one. The v2.3 regime splits them: NEUTRAL means neither leg fires and everything sits in cash, RISK-OFF means the downtrend is confirmed and the short book is live. A market that has bounced above its 50-day average but has not repaired its 20-day reads RISK-OFF on the dial and NEUTRAL on v2.3.

Is the 20-day history a record of what you published?

Only the most recent days. Koryu has published and attested the dial since 2026-07-13; everything before that is the published formula recomputed over public bars, which anyone can reproduce but which we never printed at the time. Breadth on those earlier days is also measured against today's board, which drifts, so it is indicative rather than exact. The BTC stack is exact for every day shown.

Is a regime reading a prediction?

No. It is a description of current conditions under fixed rules. Regimes can flip the day after any reading, and the instruments will simply describe the new state. They carry no instruction and no forecast. Decisions are yours.