Five Executives, One Day

Five Sunrun executives sold stock on January 6, 2026. All at $17.80 per share. The CFO, the COO, an SVP, a VP, and the General Counsel. A director sold additional shares two weeks later. The stock was down significantly within months as tariff exposure became undeniable.

Every one of those transactions was public. Filed as Form 4 with the SEC within two business days, as required by law. Available on EDGAR. Free to read.

One insider selling means nothing. C-suite officers sell stock for dozens of reasons: tax planning, diversification, real estate purchases, divorce settlements, estate planning. Any single sale is noise.

But sometimes it is coordinated timing. Vesting schedules, tax-year-end planning, and lockup expirations can produce clusters that look meaningful but aren’t. The question is whether you can rule those explanations out. When five executives sell on the same day at the same price and the stock isn’t at all-time highs, the benign explanations thin out quickly.

What follows is how to find these patterns, how to separate them from routine noise, and how to use them.

Form 4 Basics

Federal securities law requires insiders to report their stock transactions. “Insiders” means three categories: officers (CEO, CFO, COO, etc.), directors (board members), and beneficial owners of more than 10% of a company’s stock.

When any of these people buy or sell shares, they must file a Form 4 with the SEC within two business days. The form shows:

  • Who: name and title of the insider
  • What: common stock, options, restricted stock units, or other equity
  • How many shares: exact quantity
  • At what price: per share price for the transaction
  • On what date: the transaction date (not the filing date)
  • Transaction type: open market purchase, open market sale, option exercise, gift, or other. The transaction codes on the form: S = open-market sale, P = open-market purchase, M = option exercise, F = tax withholding on vest, A = grant/award. S and P are the ones I care about most.

The SEC makes this data available through free APIs. One query returns all Form 4 activity for a given ticker over a specified time window. Names, titles, transaction types, share counts, prices, dates. What used to require a paid subscription service is now available to anyone who knows where to look.

Signal vs Noise: Filtering Out the Mechanical Sales

Most insider selling is noise. Understanding why is the prerequisite to finding signal.

10b5-1 plans: the scheduled sales

SEC Rule 10b5-1 allows executives to set up automatic trading plans at a time when they don’t possess material nonpublic information. Once the plan is established, trades execute on a schedule regardless of what the executive learns afterward.

10b5-1 sales dominate the Form 4 data. An executive selling $500,000 of stock every quarter under a 10b5-1 plan tells you nothing about their view of the company. The plan was set up months or years ago.

The footnotes on Form 4 filings distinguish these. Look for: “Pursuant to a Rule 10b5-1 trading plan adopted on [date].” If you see that, the sale was pre-scheduled. Less informative.

If there’s no 10b5-1 mention in the footnotes, I treat it as more likely discretionary, though disclosure practices vary. The executive probably chose to sell on that specific day. Much more informative.

Vesting and option exercise: the mechanical sales

This is where most insider cluster screens fall apart. When restricted stock vests or options reach their exercise date, executives often sell immediately to cover the tax bill. These “sell-to-cover” transactions look like insider selling in the data, and when multiple executives have the same vesting schedule (common at companies that grant equity on the same annual cycle), you get a cluster that looks meaningful but is entirely mechanical.

How to distinguish:

  • Option exercise + same-day sale: usually tax-driven. The executive exercised options and sold enough shares to cover the tax liability. The Form 4 will show both the option exercise and the sale.
  • RSU vest + same-day sale: same logic. RSUs vest, executive sells some to cover taxes.
  • Open-market discretionary sale: no option exercise, no vest. The executive decided to sell shares they already owned. This is the real signal.

Most “insider cluster” screens you’ll find online are noisy precisely because they don’t filter out vesting-related sales. If you’re not normalizing for vesting schedules, you’re mostly detecting compensation events, not informed selling.

The 2023 rule changes

The SEC tightened 10b5-1 rules in 2023. Key changes: mandatory 90-day cooling-off period between plan adoption and first trade, no overlapping plans, one single-trade plan per year, mandatory disclosure of plan adoption and termination. These changes make the data cleaner but don’t eliminate the noise entirely.

Scoring a Cluster

Rather than treating insider selling as a binary signal, I use a rough scoring rubric. This isn’t precise, but it forces structured thinking about whether a cluster is meaningful or mechanical.

Positive signals (add points):

  • 3+ unique officers or directors selling within 10 trading days: +2
  • Discretionary sales (no 10b5-1 footnote, no option exercise): +1 per seller
  • Percentage of holdings sold exceeds 20%: +1
  • First sale in 12+ months (seller hasn’t sold recently): +1
  • Stock price below 52-week high by 20%+ (not an ATH cash-out): +1

Noise indicators (subtract points):

  • Sale immediately follows option exercise or RSU vest: -2
  • All sellers have same vesting date (annual grant cycle): -2
  • Stock is at or near all-time highs (tax-rational selling): -1
  • 10b5-1 plan footnote on all transactions: -1

A cluster that scores +4 or higher after subtracting noise indicators is worth investigating further. Below that, it’s probably mechanical.

The Sunrun example scored

The January 2026 Sunrun cluster: 5 sellers on the same day (+2), discretionary sales (+1 each for at least 3 sellers = +3), stock not at all-time highs (+1), several were first-time sellers (+1). No option exercise or RSU vest on the same day. Total: roughly +7. That’s a strong signal before you even look at what was happening with tariffs.

Context Multipliers

Certain contexts make a scoring cluster even more informative:

  • Active DOJ/SEC investigation + selling: executives selling while the company is under investigation. Legally sensitive. Informationally rich.
  • Pre-earnings window: companies impose blackout periods before earnings. Selling just before the blackout begins suggests urgency to get the trade done before the window closes.
  • Recent 8-K amendments + selling: the company is amending prior disclosures AND insiders are selling. The amendments might be correcting something insiders already knew was wrong.
  • Board member first-ever sale: board members rarely sell. They hold because their positions are relatively small and largely symbolic. A board member selling for the first time in their tenure is unusual enough to investigate.

The cluster that burned me was a technology company where four executives sold within a week. It scored well on my rubric: discretionary, first-time sellers, stock not at highs. I started building a thesis. Then I dug deeper and found that all four had the same equity grant date three years earlier, and the three-year cliff vest had just triggered. The cluster was a compensation event, not informed selling. I would have caught it faster if I’d checked the vesting schedule first.

Now I treat “check the vesting schedule” as step one, before I even score the cluster. The SEC’s 2023 rules now require companies to disclose equity award grant dates in proxy statements, which makes this easier than it used to be.

The Buy Side

Insider BUYING is rarer and almost always informative.

Executives buy stock with their personal money. Not option exercises, not RSU vesting. Actual open-market purchases where the executive writes a check. This is uncommon. When it happens, the executive is putting personal capital at risk because they believe the stock is undervalued.

The asymmetry: selling has a hundred innocent explanations. Buying has almost none. An executive doesn’t spend $200,000 of personal money on their company’s stock for tax planning.

The absence signal

Sometimes the most informative data point is the one that doesn’t exist. A company’s stock drops 50%. No insider buys. Not the CEO, not the CFO, not a single director. They all sat on the sidelines while the stock halved.

If management believed in the recovery story they’re telling analysts on earnings calls, they’d be buying. Absence of insider buying after a significant decline tends to suggest management doesn’t believe their own narrative. It’s not conclusive (there may be blackout periods, or legal counsel may have advised against it), but it’s a data point worth tracking.

Activist convergence

When a 13D filing (activist stake above 5%) arrives simultaneously with insider buying, you have a potential catalyst stack going the other direction. The activist is pushing for changes while insiders are buying because they think the stock is cheap. This is a bullish convergence.

For grey swan research, this matters as a counter-signal. If your bearish thesis on a company is running headfirst into insider buying and activist accumulation, reconsider the timing and conviction of your short thesis.

Building the Cluster Timeline

A practical step-by-step workflow for any ticker:

Step 1: Pull insider data. Query the last 180 to 365 days of Form 4 activity. You want enough history to see patterns, not just snapshots.

Step 2: Filter out mechanical sales. Remove option exercise + same-day sales. Remove RSU vest + same-day sales. What remains is discretionary activity.

Step 3: Plot on a timeline. Sort the remaining transactions by date. Group by 2-week windows. Identify clusters: periods with 3+ sellers.

Step 4: Score the cluster. Apply the rubric above. First-time sellers? Percentage of holdings? Stock price relative to 52-week range?

Step 5: Check for vesting events. Is there a company-wide equity grant anniversary or cliff vest that explains the timing? Check the most recent proxy statement for grant dates.

Step 6: Cross-reference dates. Pull 8-K filings from the same period. Check court docket data for new litigation filed in the same window. Check news for relevant developments. A cluster coinciding with an 8-K amendment chain or a denied motion to dismiss in an active lawsuit compounds the signal.

Step 7: Check the options chain. If conviction is building, look for puts covering the catalyst window. Pricing, liquidity, and expiration alignment (all covered in Guide 1).

The insider selling cluster is one signal. It becomes more useful when stacked with filing anomalies, pending litigation, or supply chain concentration. No single data point builds a thesis. The convergence of multiple independent signals is what separates informed analysis from speculation.

Failure Modes

Insider selling analysis has real limitations:

  • 10b5-1 plans obscure intent. An executive can set up a 10b5-1 plan, then learn bad news, and the pre-scheduled sales continue. The plan provides legal cover for what is effectively informed selling. You’ll see the sales tagged as “10b5-1” and dismiss them, even though the plan was adopted with knowledge that the next quarter would be bad. The 2023 cooling-off period helps but doesn’t fully solve this.
  • Insider buying is too rare to use as a regular screen. It’s informative when it happens, but months can pass without a meaningful open-market purchase. You can’t build a systematic strategy on data that shows up once a quarter.
  • False positives from vesting events. As noted above. If you don’t normalize for this, half your “clusters” are compensation events.
  • The signal is lagging. Form 4 filings are due within two business days, but that means you’re seeing sales from earlier in the week. And the insider’s information advantage may have been priced into the stock by the time you see the filing.
  • Small sample sizes. At many companies, only 5-8 insiders file Form 4s. A “cluster” of 3 sellers out of 5 total might look strong, but with such a small universe, coincidence is harder to rule out statistically.

If You Only Do Three Things

  1. Filter out the mechanical sales first. Remove option exercises, RSU vests, and 10b5-1 plan transactions before looking for patterns. The signal is in what remains.
  2. Score clusters, don’t just count them. Three executives selling means different things depending on whether they’re first-time sellers, how much they sold relative to holdings, and where the stock price is. Use the rubric.
  3. Cross-reference the timing. A cluster is one data point. A cluster that coincides with 8-K amendments, new litigation, or risk factor changes is a thesis. Check the other data sources in this series before acting on insider data alone.