Insider Trading and Retail Attention: Selling Into the Hype

In June and July of 2025, Amazon founder Jeff Bezos sold 25 million shares of Amazon stock for roughly $5.7 billion. Oracle’s Safra Catz cashed out $2.5 billion. Dell Technologies’ Michael Dell sold $2.2 billion. Across the year, tech executives and founders sold a combined roughly $16 billion in company stock during an AI-fueled rally that pulled retail investors in at record pace. Most of the largest sales moved through prearranged Rule 10b5-1 plans, filed in advance and fully compliant. A growing body of research on insider trading and retail attention asks a narrower question about activity like this: not whether the selling was legal, but whether its timing—relative to how much the public was paying attention—carries information the rest of us can actually observe.

A 2025 paper in the Journal of Financial and Quantitative Analysis set out to measure exactly that, using 128,211 firm-months of Google search data, Form 4 filings, and Robinhood account activity. This post walks through what the study found, why the pattern mostly escapes SEC insider-trading enforcement, and how to check it yourself against any Form 4 filing using free tools.

When a Stock Goes Viral, Insiders Often Head for the Exit

The 2025 insider-selling wave didn’t happen in a vacuum. It rode alongside one of the most retail-attention-heavy stretches markets have seen in years—AI stocks trending on social platforms, Robinhood account growth surging in popular names, and Google search interest in mega-cap tech spiking with every earnings beat. Bezos’s, Catz’s, and Dell’s sales landed through pre-arranged trading plans, which is meaningfully different from discretionary, market-timed selling. But the broader pattern—insider selling scaling up alongside retail enthusiasm—is exactly what the research below tests directly, across thousands of companies rather than a handful of famous names.

Fourteen months later, the backdrop looks different. In the week ending September 19, 2026, AAII’s weekly sentiment survey found bullish sentiment at just 28.8% and bearish sentiment at 53.3%—a bull-bear spread of −24.5 points, with bearishness above its historical average for a 32nd straight week. By September 23, the reading eased only slightly, to 32.7% bullish and 48.1% bearish, still well below the 37.5% long-run bullish average. If insider trading and retail attention really move together the way the research below describes, a stretch this depressed is exactly where the pattern should show up in reverse—quieter search volume and more insider buying relative to selling.

The Study: Insiders Sell Into Hype and Buy Into Silence

The paper behind this post, “Investor Attention and Insider Trading”, was written by Sattar Mansi (Virginia Tech), Lin Peng (Baruch College/CUNY), Jianping Qi (University of South Florida), and Han Shi (LSU Shreveport), and published online in the Journal of Financial and Quantitative Analysis in January 2025. Virginia Tech’s Pamplin College of Business summarized the central finding for a general audience: insiders systematically sell shares during periods of heightened retail interest and repurchase shares once that attention fades.

To measure “retail interest,” the authors use abnormal Google search volume (ABSVI)—a stock’s search volume relative to its own recent baseline. That methodology traces back to “In Search of Attention,” by Zhi Da, Joseph Engelberg, and Pengjie Gao, published in The Journal of Financein 2011, which first proposed Google Search Volume Index (SVI) as a direct, real-time proxy for retail investor attention—distinct from trading volume or news coverage, and specifically tilted toward the behavior of individual investors rather than institutions.

The Mansi et al. sample is large: 128,211 firm-month observations running from July 2004 through December 2021, built by combining Google search data with insider trading records, CRSP stock returns, and 13F institutional holdings data. That scope is what separates this from an anecdote about three famous 2025 stock sales—it is a systematic, decades-long test of whether insider trading timing correlates with retail attention.

The Numbers: How Much Attention Moves Insider Trades

The headline finding holds up under a battery of magnitude checks. Here’s what a one-standard-deviation increase in abnormal search volume does to insider trading behavior, according to the working paper draft:

The gap shows up at the firm-month level too: average abnormal search volume is 1.05 in firm-months with net insider selling, versus 0.98 in firm-months with net insider buying—a small-looking difference that is highly statistically significant (t-statistic of 22.41 across a sample this size).

The return evidence points the same direction. Monthly abnormal returns following insider-sale months are −66.8 basis points in the high-attention subsample versus −45.6 basis pointsin the low-attention subsample—insiders selling into hype precede worse subsequent returns than insiders selling quietly. Insider purchases show the mirror image: monthly abnormal returns following purchase months are 91.1 basis points in the high-attention subsample but 111.6 basis points in the low-attention subsample—insider buying is more profitable, on average, when nobody is watching.

Why This Isn’t the Insider Trading That Gets Prosecuted

Insider trading law generally targets trades based on material, nonpublic information. The pattern this study describes is built almost entirely from public inputs: Google search volume anyone can look up, and Form 4 filings the SEC already requires officers, directors, and 10%-plus owners to file within two business days of any transaction. An insider who notices retail attention spiking in their own stock, and sells into it, isn’t obviously trading on anything the public doesn’t also have access to—they’re just reacting to it faster, or more systematically, than most retail investors do.

That distinction shows up directly in the enforcement data: attention-based insider sales are “considerably less sensitive” to SEC insider-trading enforcement actions than unconditional insider sales. The result holds even after excluding every trade made under a prescheduled Rule 10b5-1 plan, and it persists across both the more permissive Bush-era SEC and the more aggressive Obama-era SEC—suggesting the pattern isn’t an artifact of one regulatory regime. The same mispricing-exploitation logic extends beyond personal trades, too: firms are more likely to conduct seasoned equity offerings following periods of high retail attention.

The Research Lineage Behind the Finding

This isn’t a finding that appeared out of nowhere. It sits directly on top of two older, foundational lines of research MarketPeel has covered before in the context of filtering opportunistic insider trades from routine ones.

The first is Cohen, Malloy, and Pomorski’s “Decoding Inside Information”, which established that more than half of all insider trades are “routine” and carry essentially no predictive power, while the “opportunistic” subset generates 82 basis points per month in value-weighted abnormal returns. NBER’s digest summary of that paper notes opportunistic trades predict future firm-specific news, while routine, calendar-driven trades don’t. Mansi et al. essentially ask: what drives a meaningful chunk of that opportunistic subset? Their answer is retail attention timing.

The second is Barber and Odean’s “All That Glitters”, published in The Review of Financial Studies, which found that individual investors are net buyers of attention-grabbing stocks—names in the news, with abnormally high trading volume, or extreme one-day returns—while institutional investors show far less sensitivity to the same cues. That paper explains the demand side of the mispricing: retail buying pressure chases attention. Mansi et al. supply the supply side—insiders appear to recognize that retail-driven demand and time their own sales into it.

Robinhood Herding: Insiders Trading Against the Crowd

The study’s most concrete evidence comes from Robinhood account data. Using herding measures built by Barber, Huang, and Odean (2022), the researchers classify months as “Buy Herding” (a sharp increase in the number of Robinhood accounts holding a given stock) or “Sell Herding” (a sharp decrease). The relationship is close to a mirror image of what retail traders are doing:

Insider sales and abnormal search volume are both highest during months when retail Robinhood accounts pile into a name, and lowest when accounts leave it. Insider purchases run the opposite way, spiking during Sell Herding months. Put plainly: insiders are measurably more likely to sell exactly when a crowd of retail accounts is piling in, and more likely to buy when that crowd heads for the exits.

How to Check This Pattern Yourself, for Free

None of the inputs behind this research require a paid data terminal. Here’s a repeatable, three-step workflow for cross-checking any Form 4 filing against a retail-attention proxy:

• 1
  Pull the Form 4 filing. Search SEC EDGAR’s full-text search for the company or insider name, note the transaction date, code, and size. (See MarketPeel’s guide to reading a Form 4 if the filing format is unfamiliar.)
• 2
  Check Google Trends for the same window. Pull up Google Trends for the company’s ticker or name and look at the weeks immediately surrounding the filing date. A sharp, unusual spike relative to the stock’s own baseline is the retail-attention signal the study measures with ABSVI.
• 3
  Add broader market context from AAII. The weekly AAII Investor Sentiment Survey won’t tell you anything about a single stock, but it tells you whether retail attention broadly is elevated or depressed right now, which is useful context for how much weight to put on any one company-level search spike.

A Form 4 sale that lands during a visible search-volume spike, from an insider whose company also shows up in a Robinhood Buy Herding wave, is the closest a retail investor can get to replicating the exact conditions this research flags as most informative—without needing access to the underlying Google or Robinhood datasets themselves.

What This Changes About Reading an Insider Sale

This research doesn’t replace the opportunistic-versus-routine framework for reading Form 4 filings—it adds a layer on top of it. Once you’ve established that a trade is opportunistic (a departure from the insider’s normal pattern, not tied to a 10b5-1 plan), the next question this study suggests asking is: was retail attention in this stock unusually high or unusually low right before the trade? A sale during a hype spike and a sale during a quiet stretch are not the same signal, even if the transaction code and dollar size look identical on the filing itself.

None of this is a buy-or-sell recommendation, and the researchers frame it the same way: this is a description of a measurable pattern in how insider trading timing correlates with retail attention and subsequent returns, not a trading strategy to replicate. Transaction costs and the possibility that some attention-driven trades rest on genuine nonpublic information both complicate turning this into an actionable edge. What it does change is how much scrutiny a Form 4 sale deserves when it lands inside a visible retail-attention spike—and how much more interesting an insider purchase looks in the kind of quiet, below-average-sentiment stretch the market is in right now.