What Tracking 165,864 Insider Trades Taught Me
What this covers
This 11-minute analysis examines what patterns emerge from tracking over 165,000 insider trades in S&P 500 stocks between 2015 and late 2025. The speaker builds the case by comparing the returns of stocks during periods of heavy insider buying, heavy selling, and neutral activity, asking whether insiders' own trading decisions can reliably forecast stock price movements. The core finding is stark: 95.7% of insider trades are sales, which means nearly constant selling carries no meaningful signal—treating every sale as a red flag would paint the entire market as troubled. By contrast, the rare instances of concentrated insider buying do weakly correlate with higher subsequent returns, though the effect is modest and cannot stand alone as an investment strategy.
The argument hinges on a legal reality that shapes insider behavior: insiders are barred from buying when they hold material non-public information about their companies. This creates a trap in interpretation—the absence of insider buying during a period of optimistic prospects (like Google's during AI uncertainty) does not signal weakness but rather reflects legal caution. The speaker walks through the performance bands at multiple time horizons, showing that heavy sell periods cluster near neutral returns with rarely more than a 2% gap, while heavy buy periods outpace neutral in nearly every case. The Google and Warner Brothers examples illustrate how raw trades can mislead without context. The analysis filtered to spontaneous open-market purchases, excluding options exercises and pre-planned trades, to isolate genuine individual decisions. The takeaway is pragmatic: occasional insider buying in your own holdings can be reassuring, but chasing insider activity as a standalone strategy misses the primary work of finding fundamentally sound businesses trading at fair valuations.
Analyzing 165,864 insider trades shows that insider selling is near-constant and largely meaningless as a signal, while rare bursts of insider buying weakly predict higher future returns — but legal restrictions and base rates mean insider activity should never be a standalone investing strategy.
- 95.7% of qualifying insider trades are sells, so sells carry almost no predictive signal
- Buy bands beat the neutral band in nearly every horizon, supporting 'insiders buy only because they expect gains'
- Insiders are legally barred from buying when holding material non-public information (e.g. Google/AI), so absence of buying is not a bearish signal
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Insiders are legally allowed to buy their own stock only when not in possession of material non-public information; breakthroughs like AI progress affecting revenue would qualify as material, so insiders are often barred from buying even when they know prospects are good — meaning lack of insider buying is not evidence of weak prospects.
“Insiders at a public company are allowed to buy their own stock, but only when they are not in possession of material non-public information.”
Across 165,864 qualifying open-market insider trades in S&P 500 stocks from 2015 to Q3 2025, 95.7% were sells and only 4.3% were buys, meaning insiders are almost constantly selling — so treating every sell as a red flag would make every stock look like it is failing.
“95.7% of those trades were sells. Only 4.3% were buys. Insiders are almost constantly selling stock all year, every year.”
During the period when Google/Alphabet was widely seen as cheaply valued due to AI threats to search advertising, insiders sold roughly $189 million and bought less than $5 million (the lone buy being a fund purchase), yet the stock subsequently rose 80% — showing that absence of insider buying did not signal weak prospects.
“In total, $189 million worth of sales and just less than $5 million worth of buys.”
Insider sell bands (light, moderate, heavy) all return roughly equal gains close to the neutral band, with the heavy sell band consistently lowest but never more than 2% below neutral — supporting Peter Lynch's view that insiders sell for many personal reasons, so even prolonged or intense selling does not signal a bad investment.
“the sell bands, light, moderate, and heavy all return pretty equal gains, and all of them are quite closely in line with the neutral band”
Unlike sell bands, buy bands beat the neutral band in nearly every case (except light buy after 5 years), with moderate and heavy buying periods outperforming consistently — supporting the view that insiders buy for only one reason: they think the price will rise.
“the buy bands beat neutral in every case except for light buy after 5 years, but moderate and heavy buying periods outperform neutral consistently.”
Heavy insider buy periods were followed by an average share price gain of 3.5% over the next 3 months, versus only 1.8% for heavy sell periods, indicating buying intensity is associated with stronger near-term returns.
“the heavy buy periods were followed by an average share price gain of 3.5% over the following 3 months, while heavy sell periods only achieved a gain of 1.8% over the next 3 months.”
Buy orders are rare and high-intensity periods of buy orders are super rare, so investors do not need to see lots of insider buying to feel comfortable investing, and insider sell orders do not indicate one should sell.
“buy orders are rare and a high intensity period of buy orders is super rare. So, there are lots of reasons why I don't think you need to see lots of buy orders in a stock in order for you to feel comfortable investing in it.”
Insider trades can give occasional interesting insights — a buy in a stock you own is reassuring — but they are not a strategy in themselves; it is better to start by finding fundamentally good businesses at fair to low valuations and flatly ignoring overvalued stocks.
“If an insider buys a stock that you already own, it's reassuring, but it's not a strategy in of itself. I much prefer to start my search by finding great businesses trading at fair to low valuations”
Because raw buy/sell counts are dominated by sells, each stock's 30-day periods were ranked into percentiles separately for buys and sells, with top-15% intensity periods labeled 'heavy', so each stock is compared against its own historical behavior to remove the sell-skew.
“For every stock, I ranked every 30-day period into percentile rankings. One set for buy events and one set for sell events.”
Warner Brothers' last insider buy about four months before a takeover bid (it rose ~140% after Netflix and Paramount bids) looks suspicious but is likely coincidence, because Warner Brothers has bought consistently since 2021 and the Netflix offer reportedly came in October, months after the buy.
“Warner Brothers has a consistent history of insider buying since 2021. Buy orders aren't that rare for Warner Brothers like they are for different stocks. And the Netflix offer came months after this buy order.”
The analysis filtered to only open-market buys for shares directly owned by insiders, removing pre-planned trades, options exercises, and fund purchases, to isolate spontaneous individual insider decisions.
“I only looked at open market buys for shares directly owned by the insiders themselves that removes pre-planned trades, options exercises, purchases by funds, a whole bunch of things.”
Repeating the analysis with slightly different thresholds would give slightly different results, but the overall story would likely be the same.
“If you repeated this analysis with slightly different thresholds, you'd get slightly different results, but I still think the overall story would be the same.”