Financial News Sentiment Analyzer

What does Financial News Sentiment Analyzer do?

Financial News Sentiment Analyzer scrapes financial news from multiple sources, runs AI-powered sentiment analysis on every headline, and outputs structured, machine-readable data enriched with live market data, confidence scores, and company classifications.

It is an affordable, pay-per-use alternative to Bloomberg Terminal ($24,240/year), Refinitiv Eikon ($22,000/year), and other institutional data providers. You pay $0.005 per article ($5 per 1,000) with the built-in AI sentiment model, with no subscriptions and no minimums. Run it once, or schedule it as a watchlist monitor that only charges for new articles and alerts you in Slack or Discord.

Key features:

• Multi-source aggregation — Yahoo Finance, Google News, SEC EDGAR filings, and AI-powered search in a single run
• AI sentiment by default — a financial sentiment model runs inside the Actor (~78% accurate on real 2026 headlines vs ~52% for word-list tools like VADER)
• LLM mode with reasons — Gemini AI (~91% accurate) adds sentiment per company, the news event type and a one-line reason, and skips quote pages without charging
• Live market data enrichment — Current price, price change %, market cap, P/E ratio, short interest, earnings dates per ticker
• Proprietary composite scoring — Sentiment weighted by source credibility and model confidence
• Deep Search — Full-text article extraction from any URL for superior sentiment accuracy
• Company & sector enrichment — Automatic ticker-to-company mapping with industry classification
• Language detection — 55+ languages detected, filter non-English noise automatically
• Watchlist monitor & alerts — Scheduled runs skip articles you've already paid for and post new headlines to Slack, Discord, n8n, Make or Zapier
• Budget protection — Automatically stops when your spending limit is reached, never overcharges
• Pay only for what you use — No monthly fees, no setup costs

Data sources

Source	Type	Data	Cost
Yahoo Finance	RSS feed	Headlines, summaries, per-ticker news	Free
Google News	RSS feed	Aggregated headlines from 80,000+ sources	Free
SEC EDGAR	API	8-K, 10-K, 10-Q, Form 4 filings	Free
AI Search	API	AI-curated financial articles missed by RSS	$0.04/search
Deep Search	URL scraper	Full article text from any URL (CNBC, Reuters, etc.)	$0.08/page

What data do you get?

Each article in the output dataset contains:

Field	Description	Example
title	Article headline	"Top Analysts See Solid Upside in Nvidia Stock"
source	Source identifier	google_news, sec_edgar, ai_search
sentiment_label	Bullish / Bearish / Neutral	"bullish"
sentiment_score	Continuous score (-1.0 to +1.0)	0.850
sentiment_confidence	Model confidence (0.0 to 1.0)	0.95
composite_score	Score × credibility × confidence	0.547
sentiment_model	Which engine produced the score	"local", "llm" or "vader"
sentiment_reason	One-line explanation (LLM mode)	"CEO ouster report hits the stock"
event_type	News event type (LLM mode)	"analyst_rating", "earnings", "management"
ticker_sentiment	Sentiment per company named (LLM mode)	{"NVDA": "bearish", "AMD": "bullish"}
tickers	Mentioned stock tickers	["NVDA"]
companies	Resolved company names	["Nvidia"]
sectors	Sector classification	["Technology"]
industry	Industry classification	"Information Technology"
source_credibility	Source reliability score	0.70
published_at	Publication timestamp (UTC)	"2026-03-21T07:10:39Z"
language	Detected language	"en"
article_type	Category	"news", "filing", "press_release"
mentioned_tickers	Additional tickers found in text	["AMD", "INTC"]

With market data enrichment enabled ($0.01/ticker):

Field	Description	Example
current_price	Live stock price	172.70
price_change_pct	Change from previous close	-3.28
market_cap	Market capitalization (USD)	4197473583104
pe_ratio	Trailing P/E ratio	35.24
short_ratio	Short interest ratio	1.32
earnings_date	Next earnings date	"2026-05-20"

Pricing

Event	Price	When charged
Article analyzed	$0.005 ($5 / 1,000)	Per article scored by the built-in AI model (or VADER)
Article analyzed (LLM)	$0.03 ($30 / 1,000)	Per article scored by Gemini, with reasons, events and per-company sentiment. If the LLM is unavailable, the article is scored by the built-in model and charged $0.005 instead
AI Search query	$0.04	Per search query when AI Search source is enabled
Deep Search page	$0.08	Per URL scraped for full-text extraction
Ticker enrichment	$0.01	Per unique ticker enriched with live market data

Paid Apify plans get automatic discounts on article events: Bronze $4.50, Silver $4.00, Gold $3.50 per 1,000 articles (LLM: $27 / $24 / $21). In watchlist monitor mode, articles seen in earlier runs are never charged again.

Cost examples:

Use case	Configuration	Est. cost
Quick check, 1 ticker, 10 articles	Built-in AI model, Google News	~$0.05
Daily snapshot, 5 tickers, 50 articles	Built-in AI model, all free sources	~$0.25
Hourly watchlist monitor, 5 tickers (~30 new articles/day)	Built-in AI model, monitor mode, Slack alerts	~$4.50 / month
Deep analysis with market data	LLM, AI Search, enrichment, 3 tickers, 30 articles	~$1.05
Full stack, 5 tickers, 50 articles	LLM, AI Search, Deep Search, enrichment	~$2.15

Compare to alternatives:

Provider	Annual cost	Sentiment	Market data	API access
Bloomberg Terminal	$24,240	Limited	Yes	No
Refinitiv Eikon	$22,000	Yes	Yes	Yes
Finnhub Premium	$6,000	Basic	Yes	Yes
This Actor (built-in model)	~$30-150	Yes	Optional	Yes
This Actor (LLM)	~$180-900	Advanced	Optional	Yes

How it works

Quick start

Step 1. Click Try for free and configure your run:

• Enter stock tickers (e.g., AAPL, TSLA, NVDA)
• Choose news sources (default: all free sources)
• Pick the sentiment engine: local (default) for speed and price, llm for the most accurate labels with reasons
• Optionally enable market data enrichment

Step 2. Run the Actor.

Step 3. Download results from the Dataset tab as JSON, CSV, or Excel.

🔔 Watchlist monitor & alerts

Turn the Actor into a news alert system for your portfolio:

1. Enter your tickers (e.g. NVDA, AAPL, TSLA) and enable Watchlist monitor: only new articles.
2. Paste a Slack incoming webhook, Discord webhook, or any URL that accepts JSON (n8n, Make, Zapier) into Alert webhook URL.
3. Click Schedule in Apify Console and run it every hour or every morning.

Each run remembers what it has already analyzed (per watchlist), so you only pay for new articles. When new headlines arrive, you get a message like:

📰 6 new articles for NVDA: 🟢 2 bullish · ⚪ 3 neutral · 🔴 1 bearish
🔴 -0.59 [NVDA] Nvidia board weighs CEO Huang's future as leadership crisis deepens <link>
🟢 +0.57 [NVDA] Nvidia shares rebound ahead of earnings <link>

Use Watchlist ID to run several independent watchlists (e.g. tech-portfolio, energy), and alertMinScore to control which headlines are highlighted.

Sentiment engines compared

We tested each engine on 91 real headlines from 12 November 2026, which we labelled ourselves, and on 400 headlines labelled by independent annotators in the public Twitter Financial News dataset.

	Built-in AI model (default)	LLM (Gemini)	VADER (legacy)
Accuracy, Nov 2026 headlines	~78%	~91%	~52%
Accuracy, human-labelled set	~80%	~76%	~45%
Price	$0.005 / article	$0.03 / article	$0.005 / article
Extra fields	—	Reason, event type, sentiment per company, quote pages skipped	—
Best for	High-volume monitoring	Research, alerts you act on, multi-company stories	Old workflows that expect VADER scores

Example where it matters:

"Nvidia Stock Sinks as Jensen Huang Ouster Report Sends Investors Toward AMD, Broadcom"

• VADER: one label for the whole headline, driven by individual words
• LLM: {"NVDA": "bearish", "AMD": "bullish", "AVGO": "bullish"}, event management, reason "CEO ouster report causes stock to sink"

Output example

{
    "article_id": "cbb7559567eb174a",
    "source": "google_news",
    "source_url": "https://news.google.com/rss/articles/...",
    "source_credibility": 0.7,
    "title": "Apple Inc. Stock Position Raised by Groupama Asset Management",
    "summary": "Apple Inc. Stock Position Raised by Groupama Asset Management",
    "authors": ["MarketBeat"],
    "published_at": "2026-03-22T11:17:21Z",
    "scraped_at": "2026-03-22T14:30:12Z",
    "tickers": ["AAPL"],
    "companies": ["Apple Inc."],
    "sectors": ["Technology"],
    "sentiment_label": "bullish",
    "sentiment_score": 0.850,
    "sentiment_confidence": 0.95,
    "sentiment_model": "llm",
    "composite_score": 0.547,
    "language": "en",
    "language_confidence": 1.0,
    "article_type": "news",
    "topics": [],
    "topic_confidence": 0.0,
    "mentioned_tickers": [],
    "current_price": 247.99,
    "price_change_pct": -0.39,
    "market_cap": 3644938780672,
    "pe_ratio": 31.35,
    "industry": "Computers, Phones & Household Electronics",
    "short_ratio": 2.41,
    "earnings_date": "2026-04-30"
}

Use cases

Algorithmic trading — Feed real-time sentiment signals into your trading system. The composite score combines sentiment, source credibility, and confidence into a single actionable number. Market data enrichment adds price context alongside sentiment.

Portfolio monitoring — Schedule daily runs across your holdings. Track sentiment trends over time. Detect negative news before it hits the price. Get notified when short interest or P/E ratios change.

Research & due diligence — Analyze sentiment around a company before investing. Compare sentiment across competitors in the same sector. Use Deep Search to extract full articles from any URL.

AI agents & automation — Structured JSON output works directly with AI agents, LangChain, and automation platforms. Schedule runs and pipe results into your workflow.

Portfolio news alerts — Use watchlist monitor mode to get Slack or Discord messages only when new bullish or bearish headlines hit your tickers.

Fintech products — Build sentiment dashboards, alerts, and newsletters. The standardized schema makes integration straightforward across all sources.

Input parameters

Parameter	Type	Default	Description
tickers	string[]	["AAPL", "TSLA"]	Stock tickers to analyze
query	string	—	Free-text search query
sources	string[]	All free sources	Which sources to scrape
maxArticles	integer	20	Max articles per source, shared evenly across tickers
sentimentModel	string	"local"	"local", "llm" or "vader"
includeRawText	boolean	true	Include article summaries
enrichWithMarketData	boolean	false	Add live price, market cap, P/E per ticker ($0.01/ticker)
deepSearchUrls	string[]	—	URLs for full-text extraction ($0.08/page)
onlyNewArticles	boolean	false	Watchlist monitor: skip (and never charge) articles seen in earlier runs
watchlistId	string	—	Name for the watchlist's memory; defaults to tickers + query
alertWebhookUrl	string (secret)	—	Slack, Discord or any JSON webhook for new-article alerts
alertMinScore	number	0.3	Highlight articles with absolute composite score above this

Integration examples

Python

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")

run = client.actor("scionic_dev/financial-news-sentiment").call(
    run_input={
        "tickers": ["AAPL", "NVDA", "TSLA"],
        "sources": ["google_news", "sec_edgar"],
        "maxArticles": 20,
        "sentimentModel": "llm",
        "enrichWithMarketData": True,
    }
)

for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    price = item.get("current_price", "N/A")
    print(f"{item['sentiment_label']:>7s} {item['sentiment_score']:+.2f} ${price} | {item['title'][:50]}")

JavaScript

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: 'YOUR_API_TOKEN' });

const run = await client.actor('scionic_dev/financial-news-sentiment').call({
    tickers: ['AAPL', 'NVDA', 'TSLA'],
    sources: ['google_news', 'sec_edgar'],
    maxArticles: 20,
    sentimentModel: 'llm',
    enrichWithMarketData: true,
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
items.forEach(item => {
    console.log(`${item.sentiment_label} ${item.sentiment_score} $${item.current_price} | ${item.title}`);
});

n8n, Make and Zapier

Use the official Apify node/app: choose Run Actor, select scionic_dev/financial-news-sentiment, pass your tickers, then read the dataset items in the next step. For push-style workflows, set alertWebhookUrl to your n8n/Make/Zapier webhook URL and enable onlyNewArticles — the Actor will call your workflow only when there is new news.

AI agents (MCP) — Claude, ChatGPT, Cursor

This Actor works as a tool for AI agents through the Apify MCP server. Add scionic_dev/financial-news-sentiment to your MCP client and ask things like:

"What's the news sentiment on $TSLA this week, and which headlines are most bearish?"

"Compare news sentiment for NVDA, AMD and INTC and include their P/E ratios."

The agent runs the Actor and gets back structured, scored articles it can reason over.

FAQ

Is this financial advice? No. Sentiment scores are AI-generated and may be inaccurate. This tool is for informational and research purposes only. Always do your own due diligence.

How accurate is the sentiment analysis? On 91 headlines from 12 November 2026 that we labelled ourselves, the built-in AI model was right ~78% of the time, the LLM ~91% and VADER ~52%. On 400 independently labelled headlines from the public Twitter Financial News dataset: built-in model ~80%, LLM ~76%, VADER ~45%.

What about paywalled sources (Bloomberg, WSJ)? We do not scrape paywalled content. All sources are publicly accessible or use official APIs. Use Deep Search URLs to extract articles from sites you have access to.

How fresh is the data? Articles are scraped in real-time when the actor runs. Schedule it hourly or daily for continuous monitoring. Market data enrichment provides live prices at the time of the run.

Can I use this with my AI agent or LLM? Yes. The structured JSON output is designed for programmatic consumption. It works directly with LangChain, AutoGPT, Claude, and any tool that accepts JSON data.

How do I get alerts only for new articles? Enable onlyNewArticles, add an alertWebhookUrl (Slack, Discord, n8n…), and schedule the Actor. Articles analyzed in previous runs of the same watchlist are skipped and not charged.

What happens if my budget runs out? The actor automatically stops processing and returns whatever articles it has already analyzed. You are never overcharged. Set a spending limit in the Apify Console to control costs.

What data do you keep? Besides your run's dataset, the analyzed article records (headline, link, short summary, tickers, sentiment and timestamps) are added to our own archive of public financial news sentiment. Your input, account details and webhook URLs are never stored there.

What happens if a source is down? Each source is fetched independently. If one fails, the others still return data. Errors are logged in the run stats.

Legal disclaimer

This actor only scrapes publicly available data. Article summaries are limited to 200 characters for legal compliance — full article text is never stored or distributed. SEC EDGAR data is explicitly public domain. All other data is derived analysis (sentiment scores) which is protected as transformative use.

Not affiliated with Yahoo Finance, Google, SEC, or any news publisher.