Koryu
← Research library
Research · 研究 · 03 · Formulas

Every number on the board, recomputable by you.

13 Jul 20266 min readThe systemKoryu Research

The monitor ranks and describes markets using published formulas. It does not recommend buying or selling anything, and a high or low rank is not advice. If a number on the board cannot be reproduced from this page and public data, that is a bug, and we want to hear about it. Decisions are yours.

Most crypto dashboards ask you to trust a proprietary score. This board is the opposite bet: every column is a formula you can recompute in a spreadsheet, published here in full, frozen by an attestation chain, and chosen for what it measures rather than what it sells. This page is the board's single source of truth, and it is deliberately detailed enough to be criticized.

Universe rules

The monitor covers the top 100 assets by 30-day median dollar volume on vetted spot markets, recomputed monthly, with four published exclusions. Stablecoins are out, because a momentum rank on an asset designed not to move is noise. Wrapped and pegged assets are out, because they duplicate their underlying. Exchange leveraged tokens are out, because their decay mechanics make daily formulas misleading. And an asset must have traded for at least 90 days, because the first weeks after a listing are dominated by allocation mechanics and airdrop flows that momentum arithmetic misreads. Additions and removals are announced on the board, with the effective date.

One honest boundary, stated twice on this site because it matters twice: a CURRENT top-100 is the right universe for a live monitor and the wrong one for a backtest. Research at this house runs on a separate, point-in-time universe that includes dead assets, per the survivorship problem. The board describes today; it is not, and must never be retrofitted into, a historical sample.

The daily close convention

Crypto never closes, so every daily formula needs a declared measuring stick. Ours is the 00:00 UTC daily bar: each day's open, high, low, and close are taken from the UTC calendar day on the reference market, and the board's daily recompute runs at 00:05 UTC once those bars are final. The choice is convention, not physics; UTC midnight is simply the least ambiguous line in a market with no bell, and using one fixed convention everywhere makes every number on the site comparable and every historical claim checkable. All annualized figures use 365 days, because this market trades weekends, a detail that quietly changes every Sharpe-style statistic imported from equities.

The columns, exactly

Price and 24h change. Last traded price from the reference market, refreshed roughly every 15 minutes around the clock, displayed with its as-of timestamp. The 24h change compares the live price to the price 24 hours earlier, not to the last UTC close, so weekend readers are never looking at a two-day-old comparison dressed as fresh.

30-day and 90-day momentum. The close-to-close percentage change over the trailing 30 and 90 UTC days: today's close divided by the close N days ago, minus one. A worked example with round numbers: an asset closing at $130 today that closed at $100 thirty days ago shows +30.0%. No smoothing, no weighting, no exclusions of bad days. These horizons sit inside the range the academic momentum literature has studied most, long enough to be slow to manipulate, short enough to matter.

Trend state. A three-state pill from two simple moving averages of the daily close. UPTREND requires price above the 50-day average and the 50-day above the 200-day. DOWNTREND requires price below the 50-day and the 50-day below the 200-day. Everything else is NEUTRAL. This is deliberately the oldest trend definition in the book; its value is not novelty but incontestability, and its known weaknesses, lag and whipsaw in ranging markets, are documented rather than hidden.

Volatility, shown as ATR14%. The 14-day Average True Range divided by the last close. True range for a day is the greatest of: high minus low, the absolute gap from yesterday's close to today's high, and the absolute gap from yesterday's close to today's low; ATR14 is the 14-day simple average of that. An asset showing 9% here moves about 9% on a normal day, which reframes every other number on its row. Two assets with identical momentum and ATRs of 3% and 11% are different instruments; this column exists so the board can never hide that.

Distance from 52-week high. The percentage gap between the last close and the highest daily close of the trailing 365 days. Zero means at the high; -80% means one fifth of the peak remains. The column separates trending leaders from recovering wrecks at a glance, two populations that momentum columns alone can conflate. The evidence behind the anchor, and its crypto caveats, are treated in the 52-week-high piece.

Liquidity, 30-day median dollar volume. The median of daily quote-currency volume over 30 days, from vetted venues only. The median, not the mean, because volume is the most manipulated number in crypto: one wash-traded spike can inflate an average for a month, while a median requires sustained real trading to move. This is a mitigation, not a cure; volume data in this market deserves permanent suspicion, and the board's vetted-venue list is published alongside the universe.

Ranking, refresh, and proof

The default sort is 30-day momentum, descending, because the board's job is to show where recent strength is concentrated. Sorting is presentation, not endorsement; re-sorting by any column is one click, and no composite score exists anywhere in the pipeline. The full board recomputes at 00:05 UTC daily; prices refresh continuously. At publication, each daily snapshot, the whole board plus the regime dial state, is serialized, hashed, and committed to a public attestation chain, with the previous day's hash included so history chains. What that proves, what it cannot, and how to verify a day yourself in seconds are documented in the attestation mechanics.

Known limitations

Stated plainly, because a methodology page that lists no weaknesses is an advertisement. The universe is current-listings by design, so the board says nothing about history. Daily formulas on a 24/7 market compress information; a coin can round-trip 20% inside one UTC day and the daily columns will shrug. Volume, even median-filtered from vetted venues, remains partially fake market-wide. Moving-average trend states lag turns. And a 100-asset cut excludes the long tail where both the best and worst of crypto happens. Every one of these is the price of publishing formulas simple enough to verify, a trade we make consciously and explain in measurement vs advice.

Change policy

Formulas on this page change rarely, loudly, and forward-only: any modification is announced, versioned, dated, and applied from that date onward, never retroactively, and the attestation chain makes silent revision detectable by anyone. If you recompute a board number from this page and get a different answer, one of us is wrong, and we have made it easy to prove which. That is the entire product. Decisions are yours.

Related reading
The systemThe Regime Dial methodology: a published risk-on/risk-off formula5 min readThe systemThe fifth guardian: why this site exists, and its relation to Shishin5 min readMethodologyThe transplant experiment: running a stock engine on crypto, in full7 min read
Frequently asked

What is the Momentum Monitor?

A daily board covering the top 100 crypto assets by 30-day median dollar volume, showing price, 30- and 90-day momentum, a moving-average trend state, ATR volatility, distance from the 52-week high, and liquidity. Every column is a published formula; there is no proprietary score.

When does the board update?

The full board recomputes once per day at 00:05 UTC from closed daily bars, and displayed prices refresh roughly every 15 minutes around the clock. Each daily snapshot is hashed and committed to a public attestation chain at publication time.

Which coins are included in the universe?

The top 100 by 30-day median dollar volume on vetted spot markets, excluding stablecoins, wrapped or pegged assets, and leveraged tokens, with a minimum listing age of 90 days. The current list, additions, and removals are published.

Is a high momentum rank a buy signal?

No. The board describes the market's present state with fixed formulas; ranking by 30-day momentum is presentation, not endorsement. The site's non-advisory line is absolute: measurements are published, decisions are yours.

Why use median dollar volume instead of average?

Medians resist manipulation. A single wash-traded spike can inflate an average and promote an illiquid asset onto the board; the 30-day median requires sustained real liquidity, which is harder to fake.