Data sources
All data presented on Polaris comes from public sources or established API providers. Every figure is traceable to its origin.
Twelve Data
Equity, FX, crypto and bond quotesQuotes and historical series for equities (including extended hours), currencies, cryptocurrencies and bonds. This is the primary source for the prices shown in watchlists and instrument pages.
https://twelvedata.com →Financial Modeling Prep (FMP)
Fundamentals, indices, earnings, 13F filingsQuarterly and annual financial statements, ratios, market indices, earnings calendar, analyst estimates and institutional holdings (13F filings). Also serves as a fallback when the primary price source is silent.
https://site.financialmodelingprep.com →Polygon.io / Massive
US market closes, futures, newsGrouped closing prices for the entire US market (used by the screener and the Score’s performance tracking), copper and natural-gas futures, news and dividends. Polygon.io and Massive are the same provider under two domain names.
https://polygon.io →Finnhub
Analyst consensus, insider transactionsAnalyst consensus, twelve-month price targets, earnings surprise history and insider transactions (SEC Form 4).
https://finnhub.io →Alpha Vantage
Put/call ratioRatio of put to call options, used by the Smart Money section as an indicator of market positioning.
https://www.alphavantage.co →FRED (Federal Reserve Bank of St. Louis)
US macroOfficial macroeconomic series: Treasury yield curves, CPI/PCE, unemployment rate, VIX, high-yield credit spreads, Global Economic Policy Uncertainty (GEPU). Public reference source.
https://fred.stlouisfed.org →Bank of Canada (Valet API)
Canada macro, exchange ratesPolicy rate, Canadian yield curve (2, 5, 10 and 30 years), Canadian CPI and official exchange rates. Official public source, no API key required.
https://www.bankofcanada.ca/valet/docs →multpl.com
Shiller CAPE, market valuationTen-year cyclically adjusted price-to-earnings ratio (Robert Shiller’s series), trailing S&P 500 price-to-earnings ratio and dividend yield.
https://www.multpl.com →NBER
Recession datingOfficial dates of US recessions, used to shade historical charts (yield curve, inflation).
https://www.nber.org/research/business-cycle-dating →CoinGecko
Cryptocurrency catalogue and quotesCryptocurrency catalogue, quotes and historical series for the crypto section. Also used as a fallback when the primary quote is unavailable.
https://www.coingecko.com →Anthropic Claude
Synthesis, analysis, editorial readingAI models used to generate editorial summaries, contextualized analyses and morning briefs. Prompts integrate macro and technical context to produce situated readings, never predictions.
https://www.anthropic.com →OpenAI
Text-to-speech, model failoverTwo uses: text-to-speech for the audio briefing (the briefing text is sent to be converted into audio) and failover for AI features — if Anthropic is temporarily unavailable, the same request is replayed there so the service stays up. Processed in the United States.
https://openai.com →News sources
Polaris reads the public feeds published by these outlets and institutions. It never republishes a full article: a headline, a summary and a link back to the source. These feeds are read by our servers — your browser never contacts them.
Service providers
In addition to data sources, Polaris relies on the following providers to operate the service. None of these providers receive your financial data for resale purposes.
Market data is refreshed at variable intervals depending on the source: intraday prices (~15-minute delay outside premium feeds), end-of-day closings, weekly or monthly macro series. Polaris is not a real-time trading platform.
Despite our verifications, errors or omissions may occur in the data we relay. The data is provided "as is", without warranty of accuracy or completeness. Always cross-check with the original source before any investment decision.
Polaris does not generate its own financial data. Our added value is in aggregation, cross-referencing, editorial synthesis and contextual reading — not in creating new data.
Last updated: May 22, 2026