Data Sources

Most prices, index levels, and futures levels are sourced from Yahoo Finance via the yfinance library. This is a free, unlicensed data feed, not a paid or contractually licensed market-data vendor. Quotes may be delayed and are not suitable for time-sensitive trading decisions. We do not currently subscribe to a real-time or exchange-licensed data feed. Some screens are the exception: the ETF Top 20 and momentum screens instead display last-sale price, 1-year return, and market cap as supplied by Nasdaq's own screener feed — see Section 6.

Sentiment signals are derived from public posts on StockTwits and Reddit, plus trending-coin data from CoinGecko's free trending endpoint, which also powers our crypto coverage. Reddit content is fetched by racing Reddit's own old.reddit.com JSON endpoint against pullpush.io, a third-party Reddit archive — whichever returns a non-empty result first wins, with pullpush.io as the reliable fallback on cloud IPs where Reddit's endpoint is often blocked; we have no direct data relationship with Reddit. We display only aggregated, derived metrics (such as mention counts or sentiment scores) attributed to their source — never raw scraped post content presented as our own analysis. These metrics reflect retail chatter and discovery-ranked interest, not institutional positioning, and can be noisy or manipulated.

Institutional holdings data comes from SEC Form 13F-HR filings via SEC EDGAR, which is public-domain government data. 13F filings disclose long equity positions only and can lag the actual reporting date by up to 45 days. Short positions, cash, foreign equities, and most derivatives are excluded from these filings. CUSIPs on those filings are mapped to tickers using OpenFIGI, Bloomberg's free-tier securities identifier API. Separately, the SEC's own company_tickers.json registry — also public-domain — powers our ticker search.

News headlines come via Google News RSS, which aggregates coverage across many publishers (Reuters, CNBC, Bloomberg, and others). We show only a headline, the originating publisher, and a link back to the original story. We do not reproduce, rehost, or display full article bodies. Click through to the source to read the full story.

Macro series (such as rates, spreads, and other indicators behind our market-condition and fear/greed views) are sourced from FRED, the Federal Reserve Bank of St. Louis's public economic data service. This is free, public-domain government data. FRED also feeds the Key Events economic calendar, but only where we've configured a FRED API key — without one, Key Events falls back to a curated FOMC schedule only. Even with a key, FOMC meeting dates themselves come from that same curated internal schedule, not from FRED; FRED there supplies only the prior fed funds rate. Release-date and actual-value data for the other indicators (CPI, payrolls, retail sales, etc.) is FRED, key-gated as above.

Nasdaq's public API feeds several parts of the Service, keylessly and browser-impersonated (Nasdaq blocks plain datacenter requests). It supplies: (a) the momentum screen's stock universe, including each name's last-sale price and market cap; (b) the ETF Top 20 screen's live price and 1-year return for each ETF; and (c) the primary path for the earnings calendar — report dates, session (pre/post-market), consensus EPS estimate, market cap, and analyst-coverage count. The earnings calendar is used on the public /earnings page, by our chat assistant, and in the free in-app Brief. If the Nasdaq feed is unavailable (for example, our cloud IP gets blocked), the earnings calendar falls back to a second path: a curated list of smart-money-relevant tickers pulled one-by-one from Yahoo Finance via yfinance, with every name shown as prominent and, where available, a revenue estimate that the Nasdaq path doesn't carry. The API response tells you which path served a given request. The covered-call idea screen draws from a separate source: it starts from S&P 500 constituents published as a CSV on GitHub, then applies our own fundamental screen; that candidate list is auto-grown daily with any name traded on the Covered Calls Advisor blog (see Section 7). The Income Engine screen uses yet another universe: a bundled S&P 500 + Nasdaq-100 constituents snapshot sourced from Wikipedia, refreshed manually rather than fetched live.

Mondegar's Pro covered-call tip sheet compares our recommendations against the actual trades posted on coveredcallsadvisor.blogspot.com, an independent third-party blog. We scrape its recent posts daily, parse each into a structured trade, use those trades to auto-grow our own candidate universe (Section 6), and score our recommendations against them. We have no affiliation with that blog's author.

Chat responses and AI-generated summaries are produced by a large language model. Depending on deployment configuration, the model is provided by Anthropic, DeepSeek, or OpenRouter (which itself routes to multiple underlying model providers), or a locally hosted model.

Stock ticker logos (shown on the earnings calendar, research pages, stock pages, watchlist, and the in-app Brief) are hotlinked from two third-party logo CDNs — FinancialModelingPrep's image-stock endpoint, then Parqet's, tried in order — with a typeset monogram fallback if neither has a logo for a ticker. We use these images solely to identify the security or company being discussed; nominative fair use covers our use of the underlying brand marks, but the images themselves come from those two vendors' infrastructure, not from us or from the companies directly. News-publisher icons (Section 4) are similarly fetched live, per publisher domain, from Google's public favicon service.

Fund-manager photos shown on guru profiles are curated portraits — mostly sourced from Wikipedia infoboxes (CC BY-SA) — that we've downloaded and committed to our own codebase; they're served from our own domain at /managers/*.jpg, not fetched live from Wikipedia. 7 of the 12 currently-tracked managers have a committed portrait; the other 5 fall back to the manager's initials, rendered the same way a missing ticker logo falls back to a monogram. A Wikipedia REST API backfill script exists, but every tracked manager already has a photo_url on record — including the 5 whose file is missing — so the script never re-checks those 5 and is dormant in normal operation.

Questions about our data sources should be directed to: info@mondegar.ai.