Method
Last updated (this deploy) · Shortlist rules b2 · coin counts are from this deploy · What changed
Essentials
Link to Essentials- Returns use rolling windows that end now; a return vs BTC is a ratio, not a difference.
- Activity measures turnover: how busy trading is, not buying or selling pressure.
- Unusualness compares a move with the coin’s own typical volatility.
- Spread is an initial liquidity check, not order-book depth or executable liquidity.
- Groups are equal-weight baskets of sectors and narratives; a coin can sit in more than one.
- Market read describes how closely coins moved with BTC in this window; it is not a trading signal.
- Derivatives figures come from OKX USDT perpetuals, not spot, and are never relative to BTC.
- Shortlist rules are transparent conditions, not predictions, and run on Binance only.
- Sources are public spot USDT market data, measured separately per exchange (Binance and OKX): never mixed across exchanges or filled in.
Returns
Link to ReturnsEvery 15m, 1h and 24h return on the site, in tables, bubbles, headlines, the Shortlist and share cards, is that rolling window’s own return, whichever window you are viewing. Returns vs BTC compare the coin’s growth with BTC’s as a ratio, not a difference.
return = latest candle close ÷ first included candle open − 1
First included candle: the first candle opening at or after now − window length. The latest candle may still be in progress.
Calculation details
A custom range uses the same rule from its own start. Chart lines show the same change at each candle close, so every line starts at zero. The 15m return is not shown for historical custom ranges. Candle sizes for each window are under Sources and freshness.
7d return
The 7d return comes from the hourly baseline: the close of the last completed hour ÷ the close 168 hours earlier − 1.
Return vs BTC
vs BTC = (1 + coin return) ÷ (1 + BTC return) − 1
Coin and BTC must come from the same source and the same times; otherwise the value is shown as unavailable. In the vs BTC chart view the same ratio is taken at each point and rebased to zero at the start.
30 completed UTC days
Link to 30 completed UTC daysSeparate on-demand context in Timeframes for one verified spot USDT asset and same-venue BTC, available on binance and okx when enabled. It uses completed UTC daily candles, not the live rolling chart or a historical Custom range.
Return = 100 × (close(E) ÷ close(S) − 1)
E is the eligible UTC midnight; it advances on demand after 00:05 UTC. S is 30 days earlier. Thirty-day return needs 31 closing observations. vs BTC is 100 × ((closeCoin(E) ÷ closeCoin(S)) ÷ (closeBTC(E) ÷ closeBTC(S)) − 1), with aligned venue, quote and endpoints.
High and low use only the 30 complete OHLC days in [S,E); the preceding anchor candle supplies close(S) but its extrema are excluded. Dates identify the daily candle, not the intraday time of the extreme. Endpoint below the 30-day high is 100 × (close(E) ÷ high30 − 1).
Opening Previous 30 completed UTC days loads a separate window from E − 60 days through S, using its own 31 closing observations and excluding its own anchor OHLC. Its measurement endpoint is S; refresh and retained-history expiry follow the publication cycle E. Adjacent absolute returns appear together only when source, UTC boundaries and the shared closing price match. Closing or hiding the disclosure cancels its waiter; current history remains independent. Fresh aligned own-asset ranges share one linear USDT scale: current is solid, previous is dashed, and the vertical marker is the current completed endpoint close, not a live price. Flat ranges are points. Unmatched or stale histories remain independently labelled; BTC availability and missing return anchors do not suppress complete own-asset OHLC ranges.
Rebased closes independently start coin and BTC at 100. The BTC-relative chart instead starts the ratio return at 0%. Below running high measures each daily close against the highest included daily OHLC high so far; its anchor is a zero baseline and does not seed the running high. With anchor close 100 and first day high 90 / close 85, the first value is −5.56%. These are neither maximum drawdown nor current live drawdown.
Gaps are never filled. Missing benchmarks do not invalidate independently complete coin metrics. Source endpoints and fetch times are separate. Same-source stale history retains original bounds, for up to 48 hours after its expected next publication; failures retry no sooner than 10 minutes or the full rate-limit cooldown. Cache is bounded, process-local and lost on restart. Demo and historical Custom views do not gain current daily history; frozen cards are unchanged.
Activity
Link to ActivityActivity compares turnover (USDT traded) with a reference period. It says how busy trading is, not which way it went or whether buyers or sellers drove it.
activity = average turnover per completed candle ÷ average turnover per candle in the reference period
Two versions appear: Activity for the selected window, against the equal-length period just before it, and Activity 1h, the last hour against the rolling 24 hours.
Calculation details
- Activity (window)
- Average turnover of the window’s completed candles ÷ average turnover of the equal-length period just before it. Unavailable when that earlier period is missing more than 10% of its candles or had no turnover, or when no candle in the window has closed yet. On OKX, if the older page of a coin’s 24h data fails to load, the coin keeps its full window and only its window activity is unavailable, with a warning. The tables mark a value shown as 2.0× or more as busy.
- Activity 1h
- One value per coin, the same on every window, from the rolling 24-hour window’s 5-minute candles: average turnover of the last 12 completed candles ÷ average of all completed candles in that 24-hour window.
- Words used for activity
- Below usual under 0.8×, usual under 1.5×, elevated under 2×, busy from 2×. Unusualness uses its own words on a different scale.
Unusualness
Link to UnusualnessHow large a move is for this coin, measured in its own typical moves: 1× is a typical-sized move for a window this long, 3× is three times that. It describes the move; it is not a signal.
unusualness = |ln(1 + return)| ÷ (σ × √hours)
where:
- return
- the window’s return as a fraction (+2% is 0.02)
- ln
- the natural logarithm
- | |
- the absolute value: size only, not direction
- σ
- the coin’s typical hourly move, from its hourly returns over the prior week
- hours
- the window length in hours
Unusualness 1h, a Shortlist column, always uses a fixed 1-hour window: the rolling 1h return with the same σ and hours = 1, whichever window the dashboard shows.
Calculation details
σ is 1.4826 × the median absolute deviation of the coin’s hourly log returns over the week before the most recent one, with a floor of 0.05% per hour so very quiet coins do not produce huge multiples. In the vs BTC view, σ is taken from the coin’s hourly returns in excess of BTC. Values above 10 are shown as >10×.
It is not measured when fewer than 90% of that week’s hours are available, more than half of the hourly returns are zero, or more than 10% of hours had no turnover. It needs a live window, because it uses volatility from before the range. Words used for unusualness: ordinary under 1×, elevated 1–2×, high 2–3×, very high 3× and above. Activity uses its own words on a different scale.
Spread
Link to SpreadThe gap between the best bid and the best ask, in basis points (bps): 1 bps is 0.01%. It is an initial liquidity check from the best quotes only; order-book depth and executable liquidity are not measured.
spread (bps) = (ask − bid) ÷ midpoint × 10,000
Calculation details
Best bid and ask come from each exchange’s bulk ticker, refreshed every 15 seconds. A missing, crossed or non-positive quote is unavailable, never zero. Words used: tight up to 5 bps, moderate up to 20 bps, wide above that. In the Shortlist, a spread quote older than 60 seconds is stale, and a wide chip marks a fresh spread wider than 20 bps, whatever the Max spread setting.
Groups
Link to GroupsCoins are grouped two ways: sectors, which follow CoinGecko categories, and narratives, which are proposed editorial themes. Groups can overlap, and a coin appears once in market totals.
group line = equal-weight average of members’ return lines
Calculation details
- Sectors
- Follow CoinGecko categories: L1 / Smart contract platforms; L2 & Scaling; DeFi; Infrastructure & Oracles; AI & Compute; DePIN & Storage; Gaming & Metaverse; Memes; RWA; Privacy.
- Narratives
- Proposed editorial themes, each with a tier: Core, Active or Watch.
- Tier
- Editorial status of the theme, not investment quality.
Group figures
- Rising, falling
- A coin is rising when its return is positive after rounding to the two decimals shown, and falling when negative; the card shows how many are rising out of all loaded members, and how many are falling. The vs BTC view leaves this line out; Beat BTC covers it.
- Equal-weight
- The group line: an equal-weight average of the members’ return lines, using only members with data at both the start and end of the window. It needs at least 3 such members, covering at least 80% of the group’s coins on the exchange, with identical candle times; otherwise the reason is shown instead.
- Median
- The middle return of the members with data at both the start and end of the window (the same members as the group line); with an even number, the average of the two middle returns.
- Ex-largest
- The equal-weight average return of those same members, leaving out the one with the largest move in either direction. Needs at least 3 members.
- Beat BTC
- How many of those same members have a positive return vs BTC, shown out of all loaded members.
- Top mover
- The largest single move’s share of all loaded members’ moves added up regardless of direction: 60% means one coin made 60% of the group’s total movement. It is not that coin’s return.
- Coverage
- How many of the group’s coins are listed on this exchange, out of all its coins.
- Moving
- A coin is moving when the size of its return reaches 0.25% for 15 minutes, 1% for 1 hour and 3% for 24 hours; custom ranges use thresholds in between on a log scale, and 3% beyond 24 hours.
- Breadth
- The share of coins (BTC excluded) above their window start at each point, or beating BTC in the vs BTC view. Coins that had not started trading yet are not counted.
- Coverage notes
- Partial coverage, a small basket (fewer than 6 coins), or one coin accounting for more than half of the group’s absolute movement. These describe the data, not a forecast.
In the vs BTC view, Median, Ex-largest and Top mover use each member’s return vs BTC. Beat BTC already compares each member with BTC, so it shows the same count in both views.
Market read
Link to Market readThe line above the chart says whether coins moved with BTC in this window or on their own, and how far apart their returns were. It describes the window; it is not a trading signal.
co-movement = Σ (turnover × correlation with BTC) ÷ Σ turnover
Calculation details
- Correlation
- How closely a coin’s candle-to-candle moves track BTC’s over the window’s candles (a Pearson correlation, from −1 to 1), using the candle times both have. It always uses each coin’s own return, also in the vs BTC view: it measures moving together, not beating BTC.
- Turnover weighting
- Each coin’s correlation counts in proportion to its USDT turnover in the window, so the read reflects where trading is. With no turnover at all, it is a plain average.
- Tick-bound
- A coin whose price did not change on more than 25% of its candle steps and whose moves are all about one size: it never moved, or its smallest move is at least half its median move. That is a price sitting on its exchange’s smallest price step (one tick). A quiet coin whose moves vary in size is still measured. A tick-bound coin’s correlation cannot be measured, so it is left out and counted as “tick-bound left out”. Coins with fewer than 3 shared candle steps or no measurable correlation are left out too; both still count in “of N coins”.
- Words used for the market read
- 0.6 or more reads as “High co-movement with BTC”, 0.3 or less as “Low co-movement with BTC”, anything between as “Mixed”.
- Middle half
- The range from the 25th to the 75th percentile of the coins’ window returns (BTC excluded), so half of the coins sit inside it and one outlier cannot stretch it. In the vs BTC view it uses returns vs BTC.
- Confidence
- Marked low confidence when BTC has fewer than 20 candle steps in the window, or when fewer than half of the coins considered could be measured; the line then states the coverage, such as “measured 97 of 205 coins”. Unavailable when BTC has fewer than 3 points or fewer than 5 coins can be measured.
Derivatives
Link to DerivativesThe Leverage table and each coin’s OKX perp details come from OKX USDT perpetual futures (perps), not spot. They describe positioning in those contracts and are never relative to BTC, also in the vs BTC view. A coin without an OKX USDT perp is marked “not an OKX USDT perp”.
funding (8h eq) = current funding rate × 8 ÷ hours between settlements
Calculation details
- Funding (8h eq)
- OKX’s current estimate of the next funding payment, in percent, scaled to 8 hours so coins that settle at different intervals compare directly. Positive means longs pay shorts. The line under it says how often the coin settles and notes when the rate is at OKX’s cap or floor; an extreme chip marks a rate of 0.1% or more per 8 hours, either way.
- Funding (settled 72h)
- The average of the funding actually paid in each of the last nine complete 8-hour UTC blocks (72 hours), in percent per 8 hours; the block in progress does not count. Each block adds up the settlements inside it, using OKX’s realised rate where it reports one. Unavailable unless the settlements cover all nine blocks in full.
- OI
- Open interest: the size of all open positions in the coin’s OKX USDT perp, in coins and in US dollars, as OKX reports it.
- OI 24h
- The change in open interest counted in coins, so price moves do not count, from OKX’s hourly reading nearest to 24 hours before the latest one, within an hour either way; unavailable without such a reading. A small base chip marks a change from less than $10M.
- Refresh
- The server reloads funding, open interest and premium from OKX once they are a minute old, and the 24h OI change once it is 15 minutes old. When a reload fails, the last good figures are kept for up to 15 minutes and 1 hour respectively; the table footer shows their age.
Shortlist
Link to ShortlistThe Shortlist applies fixed, transparent conditions (rules version b2) to the live 24h window together with the rolling 15m and 1h returns. The conditions describe the data now; they are not predictions, and the thresholds are percentage returns, turnover multiples and basis points, not a score. The Shortlist runs on Binance only; it is not offered for OKX.
For every preset: On live data, each of 15m, 1h and 24h is that rolling window’s return. A fresh spread above the Max spread setting excludes a coin; a stale or unavailable spread does not: the coin is listed and marked unverified.
Strength · Relative strength + activity
Outperforming BTC by more than 0.5% on 15m and 1.0% on 1h, with activity 1h at least 1.5×.
Ordered by 1h outperformance vs BTC, then activity.
Agreement · Timeframe agreement
Positive 15m, 1h and 24h returns, positive 24h return versus BTC, and 1h unusualness at least 1.0× its hourly baseline. These horizons overlap; they are not independent confirmations.
Ordered by recent hours beating BTC, then 24h vs BTC.
Weakness · Relative weakness + activity
Underperforming BTC by more than 0.5% on 15m and 1.0% on 1h, with activity 1h at least 1.5×.
Ordered by 1h underperformance vs BTC, then activity.
Calculation details
- Max spread: 10, 25, 50 (default) or 100 bps, or off; the Max spread control counts the coins listed as
unverified. Turn on “Require a fresh quote” to exclude coins without a fresh quote instead; they then appear under Why not listed as “No fresh quote”. - BTC is the benchmark and never listed. Coins that started trading after the window start, or that lack one of the figures a preset needs, are left out and counted as data exclusions.
- “Recent hours beating BTC” counts the last 6 completed UTC hours in which the coin’s return beat BTC’s.
- A near miss failed exactly one condition, by no more than that condition’s set tolerance.
Rules changes
- b2: 15m and 1h inputs use the one-number-per-period rolling-window returns (1-minute candles).
- b1: Initial rules.
Share cards
Link to Share cardsA share card is built from the same figures at the moment it is captured, with its times in UTC. Every card carries the line “Spread is an initial liquidity check; depth not measured.” and “Descriptive market data · not investment advice”. A saved snapshot is frozen at capture: it never updates, and its “Explore the live view” link opens the same coin and settings with current data. An unavailable figure on a card shows its own reason.
Back to ContentsSources and freshness
Link to Sources and freshnessPrices and turnover come from public spot market data for USDT pairs on Binance and OKX, one exchange at a time, fetched by the site’s server. Nothing is mixed across exchanges, and when an exchange cannot be reached the site never fills in from another exchange or from sample data.
The current list has 262 coins: 206 on Binance and 173 on OKX. When a figure has no honest value it is shown as unavailable, with the reason; when a refresh fails, the last good data is shown as stale, with its time.
Calculation details
Refresh
- Market data
- Your browser asks for fresh figures about every 30 seconds while the tab is visible. The server reuses one load for 15 seconds (15m and 1h windows) or 30 seconds (24h and custom ranges that run to now). On OKX, rolling-window data may be served up to 30 seconds past its refresh interval while a single background refresh runs, always labelled with its own update time; Binance always waits for fresh data.
- Spreads
- Best bid and ask for every coin from the exchange’s bulk ticker, refreshed every 15 seconds.
- Hourly baseline
- Two weeks of hourly candles per coin, reloaded every 10 minutes. It provides the 7d return and the typical move used for unusualness.
- Market context
- BTC dominance and total market value from CoinGecko, stablecoin supply from DefiLlama, the Fear & Greed Index from alternative.me, and BTC perpetual funding and open interest from OKX. The Context page uses OKX USDT perpetuals.
Which coins
The list is built from CoinGecko: the top 100 coins by market capitalisation, a reviewed list of additions (large and established coins that would otherwise fall between the cuts), and up to 30 of the highest-ranked coins in each sector. A coin must have a live spot USDT pair on a supported exchange. Stablecoins, wrapped and liquid-staking tokens, and tokenised gold and stocks are left out; when two coins share a ticker, the larger by market capitalisation is kept. Each coin’s exchange pairs are checked against CoinGecko’s exchange listings.
Windows and candles
The rolling windows are 15m, 1h and 24h, each ending now. A custom range can run from 5 minutes to 31 days, starting no earlier than 1 January 2017.
Rolling windows use 1-minute candles for 15m and 1h, 5-minute candles for 24h. A custom range uses the smallest candle (1, 3, 5, 15 or 30 minutes, or 1, 2 or 4 hours) that keeps it to at most 300 candles; older OKX history is limited to 100 candles. The window starts at the first candle boundary at or after the requested start, so no earlier trades are counted. Moves inside a candle are not shown.
Unavailable and stale
- Stale
- A refresh failed and the last good data is shown with its time instead. Market data is served this way for up to 15 minutes and the hourly baseline for up to 60 minutes. The header shows Stale when the figures are older than 90 seconds.
Limitations
Link to Limitations- Candles hide moves inside them; a 24h window moves in 5-minute steps.
- Each exchange is measured on its own; prices and turnover differ between exchanges, and only spot USDT pairs are covered.
- The 15m, 1h and 24h windows overlap, so agreement across them is not independent confirmation.
- Spread is not depth: a narrow spread can sit on a thin order book.
- Unusualness uses typical moves from one recent week of hourly data, so it changes as that week rolls on.
- Sector and narrative memberships are editorial or follow CoinGecko categories, and can change.
- Everything here describes market data. It is not investment advice.