Methodology

Every number on this site is either measured or modelled, and the difference matters. Measured numbers come straight from an exchange or a chain and we simply display them. Modelled numbers are our calculation, and those are the ones that can be wrong in ways you should understand. This page documents both.

Where the data comes from

Everything is fetched by your browser from public APIs. We run no backend, store nothing about you, and proxy nothing — the requests go from your machine to the source.

• Binance spot API — prices, 24-hour statistics, candles, order book depth.
• Binance USDⓈ-M futures API — open interest, funding rates, long/short account ratios, top-trader positioning, taker buy/sell volume.
• CoinGecko — market capitalisation, dominance, the top-100 list, circulating supply. CoinPaprika is the automatic fallback when CoinGecko rate-limits.
• alternative.me — the Fear & Greed Index, daily since February 2018.
• DefiLlama — stablecoin circulating supply and chain TVL.
• mempool.space — Bitcoin chain tip, difficulty adjustment and recommended fee rates.
• Owlracle — Ethereum gas price percentiles.
• blockchain.info — Bitcoin price history back to 2010, captured at build time because that endpoint does not permit browser requests.

When Binance is unreachable

Binance answers requests from a number of jurisdictions — the United States most prominently — with HTTP 451, "unavailable from a restricted location". Because most of the derivatives data here originates there, we test once per visit and route around it automatically when it is blocked:

• Spot prices and candles fall back to Kraken, then to Coinbase. Nothing on the site loses accuracy — only the venue behind the quote changes.
• Open interest and funding rates fall back to Hyperliquid, an on-chain perpetual venue with a public, keyless API and no geographic blocking. Hyperliquid settles funding hourly, so its rates are multiplied by eight before display to match the eight-hour convention used everywhere else on this site. Where a figure comes from Hyperliquid, the page says so next to the number.
• Historical open interest, long/short account ratios and taker buy/sell flow have no free equivalent anywhere. Rather than substitute something that is not the same measurement, those panels show an explicit "not available from your location" notice.

Liquidation heatmap — a model, not a ledger

This is the most important entry on this page. No public data source publishes the price at which individual positions would be liquidated. That information exists only inside each exchange's risk engine. Every liquidation heatmap you have ever seen, including the paid ones, is therefore an estimate.

Ours works like this:

1. Each candle in the window is treated as a cohort of positions opened at that candle's typical price, (high + low + close) / 3, sized in proportion to the volume traded in it.
2. That cohort is divided across four leverage tiers with fixed shares: 10× (40%), 25× (30%), 50× (20%), 100× (10%).
3. For each tier we solve for the liquidation price, using a 0.4% maintenance margin — the smallest Binance USDⓈ-M bracket:
   long liquidation = entry × (1 − (1/leverage − 0.004))
short liquidation = entry × (1 + (1/leverage − 0.004))
4. The level is carried forward on the map from the candle that created it until a later candle's range contains that price — at which point the cluster is treated as consumed and removed.

Known limitations. Candle volume is a proxy for position size and includes closing and spot-style flow. The leverage tier shares are an assumption nobody can verify. Cross-margin and manually added collateral move real liquidation prices invisibly. And the model sees one venue. Read it as a map of where liquidity probably sits.

Altcoin Season Index

The share of the top 50 coins by market capitalisation that outperformed Bitcoin over the selected window, expressed 0–100. Stablecoins are excluded because a pegged asset cannot outperform; wrapped and liquid-staked derivatives (WBTC, WETH, stETH, cbBTC and similar) are excluded because they duplicate an asset already counted. Daily closes come from Binance USDT pairs, so a coin without a Binance pair drops out of the sample.

Known limitation. Today's top-50 membership is applied across the whole window, so a coin that recently entered the ranking after a large rally is counted for the full period. This gives the historical series a mild upward bias.

Bitcoin dominance history

The live figure is CoinGecko's published dominance, unmodified. The history is reconstructed, because no free API publishes a historical total-market-cap series:

1. Circulating supply for each of the top 30 assets is taken from CoinGecko.
2. Each is multiplied by that asset's daily closing price from Binance to produce a market-cap series.
3. Bitcoin's share of the summed total is computed for each day.
4. The whole series is multiplied by one constant so its final point equals today's published dominance.

Known limitations. Supply is held at today's value across the window — accurate to well under a percent for large caps over a year. The long tail beyond the top 30 is absorbed by the calibration constant rather than measured. The shape of the series is reliable; individual historical points are close approximations.

Bitcoin rainbow chart

An ordinary least squares fit of log10(price) = a × ln(days since 2009-01-03) + b across every weekly close since August 2010. Bands are drawn at fixed offsets in log10 space either side of the fitted line, so each band is a constant percentage width. The fit is recalculated as new data arrives, which means band boundaries drift over time.

Known limitation. The curve is fitted to the price history it is drawn over, so of course it describes it well. Its only input is time. It is a perspective tool, not a model of value, and it cannot tell you anything about the future.

Everything else

• Open interest, funding rates, long/short ratios, taker flow — displayed as published by Binance. The only calculation we add is annualising funding (rate × 3 × 365) and converting open interest to notional dollars (OI × mark price).
• Fear & Greed Index — displayed exactly as published by alternative.me. We add the historical band shading, nothing else.
• Market heatmap — a squarified treemap of the top 100 by market capitalisation. Tile area uses market cap ^ 0.82 rather than raw market cap, which keeps Bitcoin from swallowing the canvas while preserving the ordering. Colour saturates at ±10% for the 1h and 24h views and ±30% for 7d and 30d.
• Halving countdown — block height from the live chain tip; the date projected using the average block time of the current difficulty epoch rather than a flat ten minutes.
• Gas tracker — Ethereum costs computed as gas limit × gas price ÷ 10⁹ × ETH price, using standard gas limits (21,000 transfer, 65,000 token transfer, 150,000 swap, 85,000 mint). Bitcoin costs assume a 140 vB single-input, two-output SegWit transaction.
• Stablecoin dominance — DefiLlama aggregate supply divided by CoinGecko total market capitalisation.

Refresh intervals and caching

Prices refresh every 20–60 seconds, derivatives data every 45 seconds to 2 minutes, market-cap data every 3 minutes, and daily series once a day. Responses are cached in your browser's session storage so moving between dashboards does not re-hit rate limits. If an upstream API fails, the last successful response is shown rather than a blank panel — which means a value can occasionally be a few minutes stale rather than missing.

Exchange data

Fee schedules, country restrictions and product availability are compiled by hand from each exchange's own published pages, and every review and table carries the date it was checked. These change frequently and sometimes without notice. We link to the primary source on each page so you can confirm before you deposit; please do.

Corrections

If a number here is wrong, we want to know. Email hello@cryptodata.today with the page and what you believe the correct figure is, and we will check it and publish a fix.