Nine downgrades landed on The Trade Desk inside a single session, and none of them were marginal trims. Targets that had been sitting in the twenties and thirties were reset into the low teens and single digits, which is what firms do when they are rebuilding a model rather than adjusting one. That cluster set the register for the week: sentiment moved in blocs across advertising technology, application software and telecom, with one name running hard in the opposite direction.
Using FMP's Stock News API alongside the Stock Grades API, this article works through five companies where multiple firms revised at once, separates what each cluster was actually repricing, and shows how revision density can be built into a repeatable screen rather than read as a headline count.
Key Takeaways
- Revision density identifies where consensus is moving quickly, but the content of the revisions determines whether the change is about growth, margin, capital allocation or a transaction.
- The Trade Desk, HubSpot and TELUS drew downgrades for structurally unrelated reasons: pricing and mix pressure, a terminal-value reset alongside improving profitability, and a rebuilt capital returns framework.
- DoubleVerify's seven downgrades carry almost no analytical view, since ratings and targets converged mechanically on an announced acquisition price.
- Unity's four upgrades were the week's only positive cluster, and the reasoning centred on data coverage expansion rather than a broad recovery in advertising demand.
The Names Analysts Rewrote This Week
The Trade Desk Inc. (TTD) - 9 Downgrades
Nine firms cut The Trade Desk after second-quarter revenue of $715 million grew 3% year over year and landed roughly 5% below consensus, with EBITDA of $241 million at a 34% margin, down about five points. The forward guide did the real damage: third-quarter revenue of $650 million or better implies a 12% decline and sits 19% below the street, while EBITDA guidance near $160 million at a 25% margin came in more than 50% below consensus. Evercore ISI moved to In Line and cut its target to $13 from $27, Susquehanna went to Neutral at $14 from $34, BMO Capital shifted to Market Perform at $15 from $38, Guggenheim moved to Neutral at $12 from $25, Truist Securities went to Hold at $16 from $35, Baird cut to Neutral at $9 from $27, and RBC Capital, Cannonball Research and Raymond James all stepped down, the last of those to Underperform.
What makes the cluster analytically interesting is the asymmetry between the two lines. Revenue is guided down roughly 12% while EBITDA is guided down by more than half, and that spread is the actual subject of the reset. Management pointed to macro pressure on consumer packaged goods and automotive advertisers, which together account for around a quarter of the business, and to those advertisers shifting budgets toward programmatic guaranteed and fixed-price formats sold by lower-priced competitors. Those formats involve far less decisioning, which is precisely where the platform earns its take rate. A mix shift of that kind compresses margin faster than it compresses revenue, and several firms folded execution missteps and share loss into the same estimate cut, with two-year revenue reductions in the 14% to 27% range.
The signal is that this is an estimate problem rather than a quarter problem, and one firm was candid about having very limited conviction in its own revised forecast. That combination, wide target dispersion sitting on top of low stated confidence, is worth monitoring rather than acting on. Take rate, the share of spend running through guaranteed and fixed-price channels, and the trajectory of Kokai and OpenPath adoption are the variables that would confirm or contradict the current framework. FMP's Analyst Estimates API is the natural place to watch this develop, since it makes visible whether the forward revenue and EBITDA lines keep drifting after the initial cut or stabilise around the new base.
DoubleVerify Holdings Inc. (DV) - 7 Downgrades
Seven downgrades on DoubleVerify look identical to a fundamentals cluster in a count-based screen, and they are nothing of the sort. Nielsen agreed to acquire the company for $13.60 per share in cash, valuing it at roughly $2.15 billion and representing a 30% premium to the sixty-day volume weighted average price. Needham moved to Hold and suspended its target outright, RBC Capital, Canaccord, BMO Capital and Truist Securities all reset to exactly $13.60, and Raymond James and Citizens moved to their respective neutral equivalents. The company cancelled its earnings call in light of the transaction.
The tell is the convergence. Seven independent firms arriving at the same price to the cent is not seven independent judgments; it is the deal price being written into the models. The genuine information sits in what the notes say around it, namely that no competing bidder is anticipated, partly because a private equity holder of roughly 12% of the shares has already committed to vote in favour. The transaction is expected to close by the end of the fourth quarter subject to shareholder and regulatory approval, and the combined business is framed as generating over $4 billion in pro forma revenue by pairing verification and media quality data with cross-screen audience measurement.
For a research process, this is the most useful case in the screen, because it exposes a false positive that revision-count screens produce reliably. Two of the notes did reference slowing revenue growth and elevated quarterly volatility as context for why an all-cash exit was a reasonable outcome, so there is a fundamental thread, but it is not what moved the ratings. Cross-referencing revision clusters against FMP's Mergers and Acquisitions API filters this category out cleanly, and reading target dispersion afterwards is instructive: a cluster where dispersion collapses to zero is a deal, while a cluster where dispersion widens is a disagreement.
HubSpot (HUBS) - 8 Downgrades
HubSpot drew eight downgrades in a quarter where the reported numbers were, on their face, solid. Revenue of $911.7 million grew 20% as reported and 17% in constant currency, non-GAAP operating margin expanded to 20.3% from 17.0%, the customer base reached 306,446 and grew 14%, the company repurchased $531.9 million of stock during the quarter and the board authorised a further $1.0 billion programme. Bernstein SocGen Group cut its target to $220 from $381, Stifel moved to Hold at $200 from $275, Piper Sandler went to Neutral at $220 from $250, BMO Capital reset to Market Perform at $215 from $230, and Oppenheimer, Wolfe Research, Stephens and CapitalOne all stepped to the sidelines.
The disconnect resolves once you look past the income statement. Net new customer additions came in around 7,000 against an expectation of 9,000 to 10,000, management guided toward 5,000 to 6,000 ahead, and net revenue retention was flat rather than improving. Those are the inputs that determine the shape of the revenue curve two years out, and several firms now model constant-currency growth decelerating toward the low teens in 2027 from the mid-teens in the back half of this year. The Bernstein target cut, better than 40% from its prior level, is the clearest expression of what happened: this was a terminal-value reset, not a reaction to the printed quarter.
The AI dimension cuts both ways here and deserves care. Agent adoption grew sharply, and management projected two to three points of operating margin expansion next year, ahead of prior investor day targets. At the same time, heightened customer budget scrutiny is pushing net new annual recurring revenue growth below reported revenue growth, which is the mechanical precursor to slower reported growth later. One firm anchored its case on valuation near ten times forward enterprise value to free cash flow, which is an argument about where the multiple sits rather than about the business improving. FMP's Financial Growth API is the dataset that makes this legible over time, since it puts revenue, operating income and cash flow growth rates side by side and shows whether margin expansion is outrunning the top line or quietly substituting for it.
Telus (TU) - 4 Downgrades
TELUS produced the week's sharpest single move, with one firm going from Buy directly to Underperform and cutting its objective to C$13 from C$22, skipping the neutral rung entirely. National Bank Financial and RBC Capital both moved to Sector Perform at C$15, and Desjardins went to Hold at C$13.50 from C$19. The trigger was a wholesale reset of the 2026 framework: service revenue guided flat to down 2% against prior growth of 2% to 4%, adjusted EBITDA down 2% to 4% against prior growth of the same magnitude, free cash flow to roughly C$1.8 billion from about C$2.45 billion, and capital expenditure up to around C$2.6 billion from C$2.3 billion.
Revenue guided down while capital spending is guided up is an unusual configuration for a mature telecom, and it is the crux of the downgrades. Management attributed the capex increase to inflation, supply chain dynamics and investment in sovereign AI data centres, with an additional C$100 million of transformation restructuring costs on top. The dividend was reduced by 55% to C$0.1875 per quarter effective in October, with the payout target moved to 45% to 60% of trailing free cash flow from 60% to 75%, freeing roughly C$2.7 billion of cumulative cash through 2028 for debt reduction against a target of three times net leverage or lower by the end of 2028. One firm explicitly noted that its constructive thesis had rested on free cash flow growth and asset monetisation, neither of which materialised.
Underneath the guidance, the operating picture is more nuanced than the headline suggests. Mobile network revenue grew 1% to C$1.7 billion for a third consecutive quarter of growth, with 17,000 mobile and 20,000 internet net additions, though blended churn rose to 1.08%. The more striking item is a C$2.1 billion impairment at the digital services unit, attributed to accelerated customer churn as clients automate legacy services using AI, with the unit repositioning toward higher-value AI datasets. That is a rare instance of AI showing up as a measurable revenue headwind on a reported line rather than as a narrative, and it belongs on the watch list alongside the pace of non-core asset sales. FMP's Cash Flow Statement API is where this thesis is either confirmed or not, because the entire debate reduces to whether operating cash flow holds while capital intensity runs above plan.
Unity Software Inc. (U) - 4 Upgrades
Unity was the week's only upgrade cluster, and the reasoning behind it was unusually specific. BofA Securities moved to Buy with a $50 target from $30, Deutsche Bank matched that target from $31, and HSBC and Benchmark both stepped up from Hold. The quarter supported it: total revenue of $546 million grew 24%, strategic revenue of $486 million grew 38%, Grow Solutions rose 63% to $329 million, adjusted EBITDA of $160 million expanded 77% to a record 29% margin, and free cash flow reached $202 million, up 59%. Management pulled its GAAP profitability target forward to the third quarter, two quarters ahead of prior expectations.
The mechanism matters more than the magnitude. Vector, the machine learning advertising system, grew 23% sequentially at close to double internal forecasts and crossed a $1 billion annualised run rate ahead of plan, but it currently draws runtime data only from titles built on the newer engine version, which represents roughly 30% of the base. One firm framed the quarter as the first financial validation that runtime data can translate into targeting efficacy, and the upgrade thesis rests on extending that data coverage to older titles. In other words, the growth vector being underwritten is data supply, not advertising demand, which is a materially different claim and a more testable one.
That specificity is what makes the cluster worth attention, and also where the caution sits. Third-quarter guidance implies strategic revenue of $540 million to $550 million with Grow accelerating to 68% to 70% growth and EBITDA margin near 33%, so a great deal is already embedded in the near term. Create Solutions, growing 14% on an adjusted basis, remains the slower half of the business, and adoption metrics such as the quarter's tripling of spend on newer return-on-ad-spend campaigns are early rather than established. FMP's Revenue Product Segmentation API is the cleanest way to track this, since the entire thesis depends on the Grow and Create split behaving differently as engine coverage widens.
Five Clusters, Four Different Mechanisms
Read as a group, these five names make a case against treating revision density as a single signal. The counts look comparable, seven to nine downgrades on three of them, but the underlying mechanics have almost nothing in common. The Trade Desk is a pricing and mix problem expressed as a margin reset. HubSpot is a terminal-value reset arriving alongside genuinely improving profitability. TELUS is a capital allocation reset in which the revenue line and the spending line moved in opposite directions. DoubleVerify is not a view at all, but seven models absorbing a fixed transaction price. Unity, the sole upgrade cluster, is a bet on data coverage expansion rather than on end-market recovery.
That heterogeneity is the practical finding. A screen that ranks by revision count alone would have surfaced all five with roughly equal weight and would have been most confident about the one carrying the least analytical content. The discriminating step is cheap: pass the cluster through FMP's Mergers and Acquisitions API first to remove transaction-driven convergence, then read dispersion through the Price Target Summary API. Collapsing dispersion indicates a mechanical reset, while widening dispersion, as with The Trade Desk's targets running from $9 to $16, indicates genuine disagreement about the forward model and is usually the more informative case.
From there the question becomes which part of the model was rewritten, and that is answerable with financial data rather than ratings. Pairing the Analyst Estimates API against the Income Statement API separates a revenue-driven reset from a margin-driven one, which is what distinguishes The Trade Desk's situation from HubSpot's despite similar downgrade counts. The Cash Flow Statement API does the equivalent work for capital-intensive names, where the relevant question is whether the reset touched operating performance or only the allocation of cash, as TELUS demonstrates. Key Metrics TTM then normalises across business models with very different asset bases, which matters when a screen surfaces an advertising platform, a subscription software company and a telecom operator in the same week.
The Earnings Transcript API adds the layer that ratings data cannot supply, namely the language management used to describe the change, which frequently signals whether an issue is being framed as cyclical or structural well before estimates fully reflect it. The wider dataset coverage available through FMP is what makes this kind of sequencing practical, because the ratings feed, estimate history, financial statements and transaction record sit inside one reference frame rather than four disconnected ones. The takeaway for a research desk is that a revision cluster is a detection tool with a known false positive rate, and the analytical work begins after the detection, not with it.
Turning Analyst Revisions Into a Structured Research Workflow
Analyst ratings only become actionable when you stop treating them as isolated headlines and start handling them as structured data. The goal is straightforward: capture changes as they happen, standardize them, and then connect those changes back to real-world events. Before anything else, make sure your API key is active and ready to run.
1. Pull Latest Analyst Ratings
Begin with a fresh pull from the Stock News API. This endpoint aggregates recent market-moving headlines, including upgrades and downgrades. It's the fastest way to establish a baseline of who changed their view and when, without needing to scrape multiple sources.
Endpoint:
|
https://financialmodelingprep.com/stable/news/stock-latest?page=0&limit=20&apikey=YOUR_API_KEY |
Sample Response:
|
[ { "symbol": "INSG", "publishedDate": "2025-02-03 23:53:40", "publisher": "Seeking Alpha", "title": "Q4 Earnings Release Looms For Inseego, But Don't Expect Miracles", "image": "...", "site": "seekingalpha.com", "text": "Inseego's Q3 beat was largely due to a one-time debt restructuring gain, not sustainable earnings growth, raising concerns about future performance. The sale of its telematics business for $52 million allows INSG to focus on North America, but it remains to be seen if this was wise. Despite improved margins and reduced debt, Inseego's revenue growth is insufficient, and its high stock price remains unjustifiable for new investors.", "url": "https://seekingalpha.com/article/4754485-inseego-stock-q4-earnings-preview-monitor-growth-margins-closely" } ] |
2. Isolate Rating Changes
From that initial feed, filter for entries that explicitly reference upgrades or downgrades. This step is about narrowing the universe—identifying which tickers are actually seeing analyst activity versus general news flow. Once you have that subset, you're ready to move from raw headlines to structured tracking.
3. Quantify the Activity
For each ticker identified, call the Stock Grades API to retrieve the latest analyst actions tied specifically to that name. This is where the workflow shifts from collection to measurement.
Group the results by ticker and separate upgrades from downgrades. Single mentions tend to be incidental; repeated actions across multiple firms indicate something more deliberate. This aggregation step is what surfaces clusters—names where sentiment is actively being recalibrated.
4. Map the Catalyst
Once the high-activity names are clear, the next step is attribution. Use the Search Stock News API to pull company-specific headlines and align rating changes with underlying developments—earnings releases, M&A announcements, regulatory updates, or sector-wide shifts.
Endpoint:
|
https://financialmodelingprep.com/stable/news/stock?symbols=AAPL&apikey=YOUR_API_KEY |
Example Workflow: Finding the “Most Active” Stocks
- Pull a rolling seven-day window of headlines from the Stock News API and filter for upgrades/downgrades.
- Extract tickers that appear at least once with a rating change.
- For each ticker, query the Stock Grades API to retrieve the full set of recent analyst actions.
- Count total upgrades and downgrades per name.
- Prioritize tickers with three or more revisions (or another threshold aligned with your coverage).
- Run those tickers through the Search Stock News API to line up rating shifts with the underlying catalyst.
Scaling Analyst Workflows Across the Organization
Workflows built around analyst revisions rarely remain isolated for long once they begin producing repeatable signal quality. Inside most firms, the transition usually starts when a handful of analysts consistently surface actionable sentiment shifts earlier than the broader desk — not because they have access to different information, but because their process for structuring and contextualizing that information is more systematic. Over time, those workflows tend to move from individual notebooks and spreadsheets into shared operating infrastructure.
That evolution changes the role of the data itself. Analyst revisions stop functioning as scattered research notes and become part of a governed internal framework that can be referenced across research, portfolio management, and risk teams simultaneously. Centralized dashboards replace duplicated manual tracking, standardized query logic reduces interpretation drift between teams, and structured datasets create an audit trail around how sentiment changes were identified and interpreted. The operational benefit is not simply speed — it is consistency. When multiple desks are evaluating the same revision clusters against the same underlying financial and market data, internal discussions become materially less fragmented.
In practice, analysts often become the internal champions driving that standardization effort. After pressure-testing workflows across multiple earnings cycles, they begin codifying the inputs that proved most reliable: revision velocity, estimate changes, target dispersion, earnings transcript language, sector-relative performance, and catalyst alignment. Once standardized, those inputs become reusable across broader coverage universes rather than dependent on individual institutional knowledge. That continuity becomes increasingly important as coverage rotates, teams expand, or macro conditions force firms to reevaluate exposure across sectors quickly.
Infrastructure becomes increasingly relevant at that stage because scaling a workflow across teams requires more than simply increasing API usage. The underlying data pipelines, timestamps, outputs, and revision histories need to remain consistent and traceable across users and mandates. The Financial Modeling Prep Enterprise plan tend to fit into this layer less as standalone data products and more as centralized coordination systems — enabling firms to maintain shared reference frameworks as analyst sentiment, earnings revisions, and positioning data evolve over time.
Reading Consensus Before It Settles
The value in a week like this one is not the direction of the ratings but the variety of things they turned out to be measuring. Running revision clusters through FMP's Stock News API and Stock Grades API keeps the detection systematic, which is what leaves room for the harder work of establishing what each cluster is actually about.
For additional trading ideas backed by data, explore: Weekly Signals Desk | Five Insider Trades That Matter — Tracked via the FMP API
Disclosure: Signals Desk content is provided for informational and analytical purposes only and does not constitute investment advice or trade recommendations. The analysis reflects interpretation of market data and publicly disclosed or third-party information, including data accessed via Financial Modeling Prep APIs, at the time of publication. Signals discussed are probabilistic, can be wrong, and may change as market conditions and consensus data evolve. This content should be considered alongside broader research, individual objectives, and risk assessment.
