Five companies accounted for close to $590 million of insider selling in August, and on the surface the filings look interchangeable: venture-backed businesses, most of them recent listings, with founders and funds heading for the exit. Sorted by one field that most screens discard, they split into two groups that mean opposite things.
That field is plan status. A sale executed under a Rule 10b5-1 plan was scheduled months before the seller knew anything about the quarter it landed in. A sale executed outside one was a decision taken with current knowledge. Both appear on the same form, both carry the same transaction code, and only the footnotes tell them apart. This month's scan through FMP's Latest Insider Trading API covers Illumina, Chime Financial, CoreWeave, Septerna and Snowflake, and it uses that distinction, together with the direction of each stock relative to its own range, to work out which of these transactions actually carried information.
Key Takeaways
- The month's two largest discretionary sales, at Illumina and Septerna, were executed outside Rule 10b5-1 plans, while the founder selling at CoreWeave and Chime ran on plans adopted in 2025.
- Direction relative to price separates the same names again: Illumina and Chime were sold into 52-week highs, CoreWeave at roughly half its high.
- Chime shows why ticker-level aggregation matters, with a sponsor, the chief executive, a co-founder and the finance chief all filing inside the same fortnight.
- Residual ownership reframes almost every figure here, most sharply at CoreWeave, where an $83 million sale sits against a multi-class stake of well over forty million shares.
Five Registers That Changed in August
Illumina, Inc. (ILMN)
Keith Meister, a director whose firm advises the funds holding the position, sold 741,127 Illumina shares across fifteen transactions on August 4 and 5 at prices between $196.74 and $203.28, realising approximately $148.2 million. He returned on August 20 for a further 285,238 shares at roughly $213.86, another $61.0 million. Combined, that is close to $209 million and a reduction from 2,830,552 shares to 1,709,187, a cut of about 40% in a fortnight.
Two details make this the month's most informative filing. The first is that neither tranche was executed under a Rule 10b5-1 plan. These were discretionary decisions, taken twice, with whatever the seller knew at the time. The second is the direction: the shares traded around $215 in late August against a 52-week high near $232 and a low of $88, and the second sale was executed at a higher price than the first. This is an activist-adjacent holder reducing into strength after a very large recovery, not liquidating under pressure.
Neither observation makes the sale bearish on its own, and a fund trimming a position that has nearly tripled is doing something entirely ordinary. What it does mean is that this filing belongs in a different bucket from the plan-driven selling elsewhere in this month's screen, and that the next filing from the same source is worth more attention than the last. FMP's Search Insider Trades by Reporting Name API is the practical way to hold that history in one place, since it assembles a single filer's record across companies and makes it obvious when a familiar name breaks from its own established pattern.
Chime Financial, Inc. (CHYM)
Chime produced the broadest insider activity of the month rather than the largest single line. DST Global Advisors, filing as a 10% owner through seven affiliated vehicles, sold 2,837,521 shares between August 6 and 12 for approximately $92.0 million, then a further 856,825 shares on August 20 and 21 at $32.50 to $33.62 for about $28.1 million, retaining roughly 42.5 million shares. Alongside it, chief executive Chris Britt sold 550,000 Class A shares for approximately $16.5 million after converting from Class B, co-founder Ryan King sold 375,000 shares on August 6 at a weighted average of $30.27 for about $11.4 million, and the chief financial officer sold 65,000 shares. The month totals roughly $150 million across four separate insiders.
Four insiders filing inside a fortnight is the kind of pattern that looks damning until the plan field is read. King's sale ran through a family trust under a Rule 10b5-1 plan adopted in August 2025, a year before it executed, and it was paired with an option exercise, which is the signature of a scheduled liquidity event rather than a view. The sponsor's selling is a different matter, but a venture investor distributing a position after a listing is following its own fund lifecycle, and it retains a stake many times what it sold.
The context that ties it together is price. Chime traded around $33.20 in late August against a 52-week high of $34.16 and a low of $15.88, with a 50-day average near $24.58, so the entire cluster executed close to the top of the range after a steep run. That is when scheduled plans naturally trigger and when sponsors naturally distribute. FMP's Insider Trade Statistics API is built for exactly this shape of problem, aggregating every filer at a ticker so that a co-ordinated-looking fortnight can be tested against the company's own baseline rather than read as a single event.
CoreWeave, Inc. (CRWV)
Chief executive and co-founder Michael Intrator sold approximately $83 million of CoreWeave stock across August, in tranches on the 4th, the 11th and the 25th. The final block, 307,692 Class A shares at $88.31 to $89.93, was executed under a Rule 10b5-1 plan adopted on November 20, 2025, and accompanied by a routine conversion of Class B shares into Class A. Chief development officer Brannin McBee sold 481,379 shares for approximately $41.4 million, with the chief operating officer and the product and engineering lead filing smaller amounts. Across the team, August selling exceeded $128 million.
What distinguishes CoreWeave from the rest of this month's screen is direction. The shares traded near $84 in late August against a 52-week high of $153.20, so this is selling into a stock at roughly half its peak, with the executions landing close to the 50-day average rather than near the top of the range. Plan-based selling behaves this way by design, because the schedule does not care where the price sits, which is precisely what makes it uninformative about management's view and why reading it as a signal of lost confidence misunderstands the instrument.
Residual ownership is the number that reframes the rest. After the August sales Intrator continued to report roughly 1.5 million Class A shares directly, more than 22 million Class B shares through a holding vehicle, a further 21.9 million Class B held directly, and several million more across family trusts. An $83 million disposal against a position of that scale is a rounding adjustment in personal balance-sheet terms. The more useful question for this name concerns the debt-funded capital intensity of the business rather than who sold what, and FMP's Balance Sheet Statement API is where that is examined, quarter by quarter, against the obligations already committed.
Septerna, Inc. (SEPN)
Funds managed by RA Capital Management sold 1,493,000 Septerna shares at $42.50 on August 12, realising approximately $63.45 million and leaving 725,443 shares, a reduction of roughly two thirds. Third Rock Ventures sold a further 147,200 shares for about $7.4 million the same day. The RA Capital filing was not made under a Rule 10b5-1 plan, and a partner at the firm holds a seat on the board.
The timing places the transaction two days after the company reported its second quarter, and the disclosures in that report matter for how the sale is read. Septerna held $516.5 million in cash and marketable securities with a runway stated as extending at least into 2029, recognised $26.7 million of collaboration revenue in the quarter under its partnership with a large pharmaceutical developer, and narrowed its net loss year over year while research spending rose. Its lead programme completed dosing work supporting once-daily administration, with fuller data guided to early 2027. This is not a company under funding pressure, which removes the most common benign explanation for a large holder reducing.
That leaves position management as the likeliest reading, and for a specialist fund it is a sufficient one: the shares had roughly tripled from their 52-week low before the block traded, and concentration limits bind regardless of conviction. The distinction worth holding onto is that this was discretionary, executed by a holder with a board seat, days after results, in a name where the next meaningful catalyst is more than a year away. FMP's Cash Flow Statement API is the appropriate cross-check, because operating burn against the disclosed cash position is what determines whether the runway claim survives contact with rising research expense.
Snowflake Inc. (SNOW)
Snowflake's August activity was the smallest of the five and the most straightforwardly mechanical. Director Michael Speiser, whose venture firm has held the position since well before the listing, sold roughly 50,000 shares in early August at prices near $324, worth approximately $16 million, through trust entities and under a trading plan. Christian Kleinerman, the executive vice president for product management, sold 40,000 shares across August 3 and 7 at prices between $293.99 and $325.00, approximately $12.7 million. Together the month came to around $29 million.
Both sellers fit the profile that produces the least informative filings in any screen: long-tenured insiders with wealth heavily concentrated in one stock, selling predetermined quantities on predetermined dates as the price moves higher. The shares traded around $328 in late August against a 52-week high of $341.95 and a 50-day average near $285, so the executions cluster in the upper part of a rising range, which is what a calendar-driven plan produces mechanically in an appreciating stock.
The reason to include it here is as a control case. Set against Illumina, the two look similar in form, insiders selling into strength, and are opposite in substance, one discretionary and one scheduled. Without the plan field there is no way to tell them apart, and a screen ranking purely on value would rate Snowflake's activity as the weaker signal only because the number is smaller, which is the wrong reason to reach the right answer. FMP's Price Target Summary API adds the useful counterweight here, showing whether analyst expectations moved at all around the filings, since routine plan selling into rising consensus is background noise rather than an event.
What the Plan Field Tells You That the Total Does Not
Rank August's five by dollar value and the ordering is Illumina, Chime, CoreWeave, Septerna, Snowflake. Rank them by how much a reader should actually weigh them and the ordering changes materially, because two of these transactions were decisions and three were schedules. Meister's Illumina sales and RA Capital's Septerna block were both executed outside Rule 10b5-1 plans, by holders with board-level proximity, in the days immediately around company disclosures. Intrator's CoreWeave selling, King's Chime sale and Speiser's Snowflake disposals ran on plans adopted between August and November of 2025, when none of the sellers could have known what the intervening year would produce.
Layering direction on top sharpens it further. Illumina and Chime were sold into 52-week highs after very large recoveries, which is where discretionary holders and scheduled plans both naturally transact and therefore where the two are hardest to separate without the footnote. CoreWeave was sold at roughly half its high, which a discretionary seller would find an odd moment to choose and a calendar-driven plan would execute without hesitation. The combination of plan status and price position does more classification work than transaction value ever does on its own.
Building that into a repeatable process means treating the filing as the beginning of a query rather than the end of one. FMP's Insider Trade Statistics API establishes whether a given fortnight sits inside or outside a ticker's normal run of insider activity, which is what turned Chime from a co-ordinated-looking cluster into four separate stories. The Search Insider Trades API then allows the same data to be sliced by filer and transaction type, so that a sponsor distributing after a listing is never averaged together with an executive making a discretionary call.
The market's own expectations complete the picture. Where the Financial Estimates API shows forward revenue and earnings paths still rising while insiders sell on schedule, the two facts are compatible and the filing is largely noise. Where estimates are drifting and a board-adjacent holder sells off-plan in size, as at Septerna, the filing earns closer reading. The Key Metrics TTM API allows businesses as different as a sequencing company, a consumer fintech, an infrastructure provider and a clinical-stage biotech to be compared on returns, leverage and cash generation rather than on the size of the block that traded, and the depth of coverage available through FMP is what makes that comparison practical across a screen rather than one name at a time. Historical price and volume data from the Historical Price EOD API supplies the last piece, showing how much of each disposal the market actually absorbed.
The general point is that insider selling is close to useless as a sentiment measure and quite useful as a classification problem. Nobody in this month's five told you what they think about their company. Between them, though, they told you exactly who chose to act, who was following a calendar, and where in each stock's own range that happened, which is a narrower claim and a far more durable one.
Building a Repeatable Insider Research Workflow
Monitoring insider filings manually breaks down quickly once coverage expands, so the first step is to systematize the intake. The workflow starts by pulling recent transactions through the Latest Insider Trading API, which returns a standardized feed of filings — including insider role, transaction type, share count, and execution price — in a format that can be filtered and stored without additional normalization.
If you don't already have one, you'll need to generate your API key before making your first request.
Endpoint:
|
https://financialmodelingprep.com/stable/insider-trading/latest?page=0&limit=100 |
Example Response:
|
[ { "symbol": "APA", "filingDate": "2025-02-04", "transactionDate": "2025-02-01", "reportingName": "Hoyt Rebecca A", "typeOfOwner": "officer: Sr. VP, Chief Acct Officer", "transactionType": "M-Exempt", "securitiesTransacted": 3450, "price": 0, "securityName": "Common Stock" } ] |
Once those individual transactions are captured, the analysis shifts from event-level review to pattern recognition. This is where the Insider Trade Statistics endpoint becomes useful. By aggregating activity at the ticker level, it surfaces whether insiders are, on balance, increasing or reducing exposure over time rather than reacting to a single filing.
Endpoint:
|
https://financialmodelingprep.com/stable/insider-trading/statistics?symbol=AAPL |
In practice, this two-step process — first capturing transaction-level data, then summarizing it into net activity — converts a stream of filings into something analytically usable. It allows you to separate one-off trades from sustained accumulation patterns, which is where insider data starts to carry interpretive weight.
From Individual Signals to Institutional Research
What begins as an individual analyst's workflow often reveals its real value once it's shared. Insider data, when consistently structured and interpreted, tends to surface the same questions across teams: Is this activity isolated or persistent? Is it showing up across related names? How does it align with broader positioning and capital flows? At that point, the challenge is no longer access — it's standardization.
When insider analysis moves beyond a single desktop and into a shared framework, it becomes easier to align interpretation across research, portfolio management, and risk. Centralized datasets allow teams to work from the same signal definitions, apply consistent filters, and review changes over time without rebuilding the logic each cycle. What was once an analyst's custom screen becomes a repeatable input that supports investment discussions across desks.
This is where institutional infrastructure matters. Formalizing insider monitoring through shared pipelines — supported by auditable data sources and common taxonomies — reduces duplication and improves accountability. It also enables downstream integration with portfolio tools, compliance workflows, and internal dashboards, so insights persist beyond the individual who surfaced them. For firms looking to move from ad hoc analysis to a durable research layer, frameworks like the FMP's enterprise plan provide the scaffolding to operationalize that transition without disrupting existing workflows.
The real shift isn't about adding more data; it's about creating continuity. When insider activity is captured, contextualized, and distributed through a common system, it becomes part of the firm's collective intelligence rather than a one-off insight.
Keeping the Plan Status in View
The difference between a decision and a calendar entry is one footnote deep, and it changes the meaning of every figure above it. Reading FMP's Latest Insider Trading feed with that field in hand is what stops a month of routine scheduled selling from being mistaken for a verdict, and what makes the handful of genuinely discretionary filings visible when they arrive.
For additional trading ideas backed by data, explore: Weekly Signals Desk | Price-Target Gaps Identified via the FMP API (Aug 17-21)
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.
