Methodology
This page explains how CoinRithm collects third-party market data, publishes editorial content, and presents educational tools such as watchlists, mock trading, and prediction market comparisons. Coverage can vary by page, asset, source, jurisdiction, and data availability.
We aim to label sources clearly, separate educational content from investment advice, and update pages when fresher market or editorial data becomes available.
This methodology page covers cryptocurrency market data, news aggregation, prediction market source pages, editorial workflows, and how users can request corrections or clarification.
CoinRithm aggregates cryptocurrency prices, market capitalization, volume, historical chart points, exchange coverage, and related metrics from third-party market data providers and exchange-linked sources. Refresh timing can differ by endpoint, asset, and provider response quality.
Prediction market pages aggregate questions, probabilities, liquidity, volume, statuses, and related metadata from supported platforms such as Polymarket, Kalshi, Metaculus, PredictIt, Limitless, Manifold Markets, Smarkets, Robinhood (Rothera), Futuur, Myriad, ForecastEx, and Gemini. We normalize fields where possible, but platform methodology, settlement rules, and access limits remain source-specific.
Some datasets update in near real time, while editorial pages, legal content, and long-form guides update on a slower review cycle. Sitemap timestamps, page metadata, and source profiles are intended to reflect real freshness rather than future event dates.
CoinRithm Editorial is the organisational byline for CoinRithm’s educational guides and market explainers. Content is created by humans with AI assistance, using market data aggregated by CoinRithm and cited sources. CoinRithm is responsible for the published content.
CoinRithm does not guarantee completeness, uninterrupted availability, or error-free third-party data. Market data, news, and prediction market probabilities can change quickly, and some products may be restricted in specific jurisdictions. Mock trading and AI-generated feedback are educational tools, not personalized financial advice.
If you spot a factual issue, stale source detail, broken page, or misleading classification, contact us so we can review it. For source-specific disputes, we also encourage checking the underlying platform or publisher directly.
CoinRithm does not create market prices, probabilities or news. It aggregates them from the providers below. Each provider is credited here with a link to its own site, and next to the data on our pages where possible.
Aggregated from each exchange's public market data, about once a minute.
Data provided by CoinGecko
Supply refreshes every 30 minutes. All-time high and low start from CoinGecko's record; the high moves up when our live price sets a new one.
Funding rates refresh every 5 minutes.
Refreshed hourly.
Checked every 6 hours. The index itself changes once a day.
Refreshed every 4 minutes.
Refreshed hourly, for price conversion.
Public filings for US-listed market references.
Weekly Commitments of Traders reports for gold, silver, oil, major currencies and US index futures. Public domain.
Catalogs re-ingest about every 5 minutes, and a busy cycle can take longer. Every event page links to the original market.
Polymarket and Kalshi are checked every minute, Limitless every 2 minutes, Futuur every 5 and Myriad every 10. ForecastEx trades come from its delayed file, published about every 10 minutes. Each trade links to its transaction or source record when the venue publishes one.
Feeds are checked every 3 minutes. Every story links to the original article on the publisher's site.
Names and logos belong to their owners. Listing a provider does not imply endorsement or partnership.
These links point to public records and professional profiles that help verify CoinRithm's legal entity and named editors.
Bees-x Limited is registered in England and Wales. Use the official Companies House entry to verify the company number and registered office.
View Companies House recordCoinRithm Editorial identifies the organisation responsible for this content. Its profile lists the published guides and explains how human input, AI assistance and aggregated data are used. Individually credited contributions keep their own bylines.
Every event page shows five CoinRithm-computed signals. Each is defined here from the exact rules the platform runs; when the computation changes, this section changes with it.
A 0-100 tradeability score built from five components: liquidity (up to 25 points, log-scaled, full credit near $2.5M), bid-ask spread (up to 25 points, full credit only for a genuinely executable book under half a point), 24h volume (up to 20 points, log-scaled to $1M), data freshness, and probability-book structure. Honesty caps then apply: a thin, inactive, incomplete or stale market can never score above 49 (always "low"), a market whose raw quotes look structurally suspect is capped at 74, any market with an outcome pinned at 99%+ or 1%- is capped at 89, and nothing scores above 98. Tiers: 75+ is High, 50-74 Medium, below 50 Low.
When the same question trades on several real-money venues, the reference probability is a weighted median across those venues. Each venue is weighted by its open interest (the money at stake) when every venue reports it, otherwise by its reported liquidity, and each weight is capped so no single book dominates. The band shown beside it is the weighted central 80% interval (weighted 90th minus 10th percentile) across venue prices, so one thin outlier cannot widen or manufacture the range; when weight is too concentrated for a stable percentile the full min-to-max range is shown instead. The band is always displayed with the venue count.
For each venue we compare pre-resolution prices against actual outcomes on resolved markets and compute the expected calibration error (ECE). A venue reads "Well calibrated" at ECE of 0.03 or less, "Fairly calibrated" up to 0.06, and "Roughly calibrated" above that. "Limited verified data" means the venue’s resolved, verifiable sample is still too small for the label to be meaningful - it is an honesty state, not a grade.
Raw venue quotes across all outcomes of one market often sum to more than 100% - the difference is the venue’s margin (the vig, or overround). Where the platform can identify a complete book, it also computes a normalized (no-vig) probability by removing that margin, and comparison surfaces use the normalized number. Where the book is incomplete, the raw quote is shown and labeled as such rather than estimating.
Venues do not report on one basis, and we do not force them onto one: Polymarket and Limitless report on-chain traded value, Kalshi and Rothera settlement notional (contracts times $1), Smarkets lifetime matched volume converted from GBP, ForecastEx matched-contract notional on a daily cadence, Futuur and Myriad real-money stablecoin volume only. Play-money and forecast venues (Manifold, Metaculus) contribute market counts and probabilities, never dollars. That is why liquidity figures are comparable within a venue over time but not directly across venues - the same caveat printed on event pages.
Different surfaces refresh on different clocks, and each states its own. Prediction-market catalogs re-ingest about every 5 minutes, with hot, high-volume markets refreshed on a faster lane (typically within 1-2 minutes on the largest venues); pages are edge-cached for up to 5 minutes on top of that. Crypto prices update near real time, typically well under a minute. ForecastEx prices update during the trading session, while its volume and open-interest figures come from end-of-day files and move once per day. The Market Overview aggregates the latest completed-day reports and says so in its caption, and the homepage Data quality strip shows per-venue live/lagging state at a glance. When a number is older than its surface's stated cadence, the freshness caption beside it is the authority.
For company background, visit About. For contact channels, partnerships, or correction requests, use the Contact page. Legal terms, privacy, cookies, and disclaimers remain separate documents and should be read alongside this methodology page.