📈 Earnings Estimate Revisions Tracker — Upward / Downward Analyst EPS Revisions Ahead of Earnings

The single highest-alpha factor in earnings-momentum quant strategies — sell-side analyst EPS estimate revisions over the trailing 30 / 60 / 90 days. Pulled fresh per run from Yahoo Finance's earningsTrend surface, computed the way a long-short equity PM actually wants to see it: signed % revision per window, positive-revision cluster boolean, strength bucket, and the next earnings date so you can time the entry.

📊 Sample Output

When a fund's covering analysts revise their next-quarter EPS estimate 4% higher in the trailing 30 days, the stock has — empirically — about a 2.3% statistically significant excess return in the subsequent month (the foundational Estimize / I/B/E/S earnings-momentum literature, replicated across every major decade since the 1980s). The signal is so robust that it survives transaction costs, sector neutralisation, market-cap controls, and post-publication arbitrage. It is the core factor in Two Sigma, AQR, and Renaissance's earnings books. Bloomberg charges $24,000/year/seat to pipe these revisions into your terminal. This actor does it for $0.15 per symbol on demand, no contract, no setup fee.

What This Actor Does

For each US-listed stock symbol you pass in, the actor:

1. Acquires a fresh Yahoo Finance crumb + cookie jar (Yahoo blocks anonymous JSON, requires a one-time auth token per session).
2. Calls the quoteSummary endpoint requesting the earningsTrend, earningsHistory, recommendationTrend, calendarEvents, price, and assetProfile modules in a single round-trip.
3. Parses the per-period EPS trend trajectory — current, 7-day-ago, 30-day-ago, 60-day-ago, 90-day-ago consensus EPS — for the current quarter (0q), next quarter (+1q), current year (0y), and next year (+1y).
4. Computes signed revision percentages for each rolling window: (current - past_value) / abs(past_value) * 100. EPS sign-flips (positive to negative, etc.) are handled correctly so a worsening loss shows up as a DOWN revision.
5. Classifies direction (UP / DOWN / STABLE) and strength (large / medium / small / negligible) based on the user-selected primary window.
6. Flags positive-revision clusters — is_positive_revision_cluster = true when 3+ analysts revised UP in the trailing 30 days (or, when Yahoo doesn't expose the per-analyst granularity, when the 30d aggregate revision is ≥ +1% and 3+ analysts cover the name).
7. Returns the next scheduled earnings date and the most recent earnings surprise (actual EPS vs. estimate) for cross-signal context.

Routed through Apify's RESIDENTIAL proxy group by default — Yahoo Finance throttles datacenter IPs aggressively, and a residential rotation reliably stays under the per-IP threshold.

---

Output Schema

Each dataset item is one stock-symbol record:

Field	Type	Example
symbol	str	NVDA
company_name	str	NVIDIA Corporation
sector	str	Technology
industry	str	Semiconductors
market_cap_usd	float	3450000000000
current_price	float	135.42
Current-quarter trajectory
current_quarter_period	str	2026-04-30
current_quarter_estimate	float	0.92
current_quarter_estimate_7d_ago	float	0.91
current_quarter_estimate_30d_ago	float	0.88
current_quarter_estimate_60d_ago	float	0.84
current_quarter_estimate_90d_ago	float	0.79
revision_7d_pct	float	+1.10
revision_30d_pct	float	+4.55
revision_60d_pct	float	+9.52
revision_90d_pct	float	+16.46
Coverage breadth
n_analysts_current	int	45
current_quarter_growth_estimate	float	0.234
Revision counts (when exposed)
n_analysts_revised_up_7d	int | null	3
n_analysts_revised_up_30d	int | null	8
n_analysts_revised_down_30d	int | null	1
Direction / strength
primary_window	str	30d
revision_direction	enum	UP / DOWN / STABLE
revision_strength	enum	large / medium / small / negligible
is_positive_revision_cluster	bool	true (3+ analysts UP in last 30d)
Next quarter (+1q)
next_quarter_estimate	float	1.05
next_quarter_revision_30d_pct	float	+3.21
Current year (0y)
current_year_estimate	float	3.92
current_year_revision_30d_pct	float	+2.84
Earnings context
earnings_date_next	str	2026-05-28
last_actual_eps	float	0.81
last_estimate_eps	float	0.75
last_surprise_pct	float	+8.0
Analyst rating distribution
rating_strong_buy_count	int	28
rating_buy_count	int	12
rating_hold_count	int	4
Provenance
yahoo_url	str	https://finance.yahoo.com/quote/NVDA/analysis/
data_source	str	yahoo_finance_earnings_trend
fetched_at_utc	str	2026-05-14T17:45:32Z

---

Input Filters

Filter	Type	Description
symbols	array	List of US ticker symbols. Defaults to 5-stock smoke test.
revision_direction	enum	ALL / UP / DOWN / STABLE — filter by direction of trailing-30d revision.
min_revision_pct	int	Absolute threshold — skip symbols whose primary-window revision % is below this.
time_window	enum	30d (default) / 60d / 90d — which window drives the direction + strength classification.
limit	int	Max symbols to push. 0 = no cap.
proxyConfiguration	object	Apify proxy. Defaults to RESIDENTIAL group (recommended for Yahoo).

---

Use Cases

1. Hedge Funds — Long-Short Earnings-Momentum Books

The single most-traded factor in long-short equity. Long the top decile of positive-revision-cluster names, short the bottom decile of negative-revision names — sector-neutralised, market-cap-controlled, rebalanced weekly. This actor is the data-feed input.

2. Earnings-Momentum Traders — Single-Name Plays

Run with revision_direction=UP, min_revision_pct=3, limit=20 two weeks before earnings season. The names that come out are the textbook "post-earnings drift" setups — 3-4% mean reversion to consensus over the 2-month window leading into the print.

3. Sell-Side Research — Competitive Intel

Track when your competitors' notes move the consensus. If 4 firms cut numbers in the last 7 days on a coverage name and you haven't, you're behind the curve. The revision_7d_pct field flags exactly this.

4. Retail Smart-Money — "Pre-Earnings Bullish Setup" Screens

The combination is_positive_revision_cluster = true AND earnings_date_next ≤ 30 days away is the textbook "analysts are raising numbers into the print" setup. Statistically high-probability long.

5. IR Teams — Self-Coverage Tracking

Run weekly on your own ticker plus your 5 closest peers. Spots the moment your name's consensus diverges from the peer-group trajectory.

6. Algorithmic Re-Rating Detectors

Combined with our analyst-price-targets actor, the joint signal (price targets up + EPS estimates up in the same 30-day window) is the strongest sell-side conviction signal that exists outside of paid I/B/E/S data.

---

Comparison vs Bloomberg / FactSet / Refinitiv I/B/E/S / Zacks Premium

Capability	This Actor	Bloomberg Terminal	FactSet	Refinitiv I/B/E/S	Zacks Premium
EPS revisions 30 / 60 / 90 days	✅ Native, signed %	✅ EE function	✅ Estimates db	✅ ANL_REV	✅ Style Scores
Revision direction enum (UP/DOWN/STABLE)	✅ Built-in	⚠️ Build it yourself	⚠️ Build it yourself	⚠️ Build it yourself	✅ Zacks Rank
Revision strength bucket (large / medium / small)	✅ Built-in	❌ DIY	❌ DIY	❌ DIY	⚠️ Coarse
Positive-revision-cluster boolean (3+ UP in 30d)	✅ Built-in	⚠️ EE Sweep DIY	⚠️ DIY	⚠️ DIY	⚠️ Partial
Next-quarter (+1q) revisions	✅	✅	✅	✅	✅
Current-year (0y) revisions	✅	✅	✅	✅	✅
Most recent EPS surprise	✅	✅	✅	✅	✅
Earnings date next	✅	✅	✅	✅	✅
REST API, no SDK lock-in	✅ JSON dataset	❌ DDE / BLPAPI	❌ FactSet Connect SDK	❌ DataStream/Eikon	❌ Proprietary
No contract, no minimum	✅ Pay-per-symbol	❌ $24K/yr minimum	❌ ~$12K/yr/seat	❌ $22K+/yr/seat	❌ $249/mo subscription
5-symbol smoke test cost	$0.76	$24,000/yr seat	$12,000/yr seat	$22,000/yr seat	$2,988/yr
1,000-symbol monthly refresh	$150	Already paid	Already paid	Already paid	Already paid
Latency	~1 min (cold start)	Realtime	Realtime	Realtime	Realtime
Trade-execution allowed	⚠️ Personal research / SaaS	✅	✅	✅	✅

Where this actor wins: ad-hoc revision sweeps, R&D / backtesting on 50-500 names, SaaS products that need a revisions data source without negotiating a $20K+ data contract, retail-facing dashboards, single-PM family offices, sell-side associates pricing what their seniors get from Bloomberg. Where the paid feeds win: realtime tick-by-tick streaming revisions, regulated trade execution, broad universe coverage (every global ADR), and audit-trail-required compliance shops.

---

Cross-Signal: Combine With Our Sister Actors

The Earnings Revisions feed is the first leg of every serious sell-side-signal quant strategy. Layer it with:

• Analyst Price Targets — when price targets go up in lockstep with EPS estimates over the same 30-day window, it's a "conviction stack" signal. Cross-tabulate revision_30d_pct > 3 AND upside_pct > 15 for the textbook bullish-pre-earnings setup.
• Earnings Calendar — without the next-print date you can't time the entry. Use earnings_date_next ≤ 21 days to filter for "imminent print" setups.
• Earnings Transcripts — read management's guidance commentary from the last call, then check if analysts revised UP in the trailing 30 days. Tight feedback loop.
• Short Interest Tracker — names with high short interest AND positive revision clusters are the textbook squeeze setup (Tesla 2020, GameStop 2021, NVIDIA Q3 2023). Layer the two feeds.
• 13F Holdings Delta Tracker — when a top whale (Tiger, Coatue, Lone Pine) increases a position AND analysts revise UP in the same quarter, that's the "smart-money-plus-sell-side" double signal.
• Finance MCP Server — expose this actor (and all the sister feeds) directly to Claude / GPT / Gemini agents over MCP. Build an LLM-driven earnings-revision research workflow in 10 minutes.

---

Pricing

• $0.01 flat per actor start
• $0.15 per symbol record returned

Examples:

• 5-symbol smoke test (AAPL, NVDA, TSLA, MSFT, AMZN) → $0.01 + 5 × $0.15 = $0.76
• 50-symbol institutional screen → $0.01 + 50 × $0.15 = $7.51
• 200-symbol Russell 1000 sweep → $0.01 + 200 × $0.15 = $30.01
• 1000-symbol full universe → $0.01 + 1000 × $0.15 = $150.01

Premium tier — hedge-fund alpha source. The first 1000 symbols of an I/B/E/S subscription cost ≈ $22,000/year fully loaded. This actor delivers the same primary signal (revisions, strength, direction, clusters) for 150x less.

---

Anti-Bot / Rate Limit Notes

• Yahoo Finance throttles datacenter IPs aggressively from a single egress. The actor defaults to Apify RESIDENTIAL proxy group, which rotates residential IPs per session and reliably stays under the throttle. Free-tier accounts may downgrade to auto but expect dropped symbols on bursts > 5.
• The actor uses a single-in-flight semaphore + 1.2s per-symbol stagger with 6-attempt exponential-backoff retries on HTTP 429. Crumb refresh is automatic if the session token expires mid-run.
• Stockanalysis.com and Zacks URLs are emitted as stockanalysis_url / zacks_url in every record — useful for click-through verification. We do not scrape them in the current build (they would require a separate Playwright pass with deeper anti-bot evasion); they're reserved for a future "data_source=stockanalysis" fallback when Yahoo is completely down.

---

Roadmap

• v0.1 — Add Stockanalysis.com as fallback when Yahoo returns empty earningsTrend (most common for small-caps and non-US-domiciled ADRs).
• v0.2 — Add Zacks Rank cross-reference (proprietary 1-5 earnings revision rank — paid Zacks Premium feed, but we can scrape the free preview).
• v0.3 — Optional Estimize crowd-sourced consensus comparison.

Built and maintained by NexGenData — Apify's most-downloaded finance fleet. Sister actors share schema conventions for trivial cross-actor joins.

---

About NexGenData

NexGenData publishes 220+ buyer-intent actors covering SEC filings, YC alumni, Delaware DOC, global stock screeners across 30+ exchanges, IPO calendars, IP and patent intelligence, FDA approvals, B2B lead generation, and more. Every actor is pay-per-result with no seat licensing.

Apify affiliate program — free credits + 30% off

Sign up to Apify via our referral link and you'll get:

• Free starter credits to test this actor and the rest of our 220+ actor fleet
• 30% off platform fees for the life of your account

Browse the full NexGenData catalog and sign up here — same Apify, same actors, just cheaper for you.

Built and maintained by NexGenData.