Is insider buying a good sign? We followed 3,638 purchases in the S&P 500
Is insider buying a good sign for a stock? We took every purchase of $100,000 or more by the officers and directors of S&P 500 companies from 2010 to March 2026, 3,638 in all, and followed the shares for 60 trading days. Bought at once, they beat the index by 0.28% a trade. Bought once the share had started to rise, they fell 0.32% behind it.
The belief
“When a company's insiders buy its shares, buy them too, once the share starts to rise and shows they were right.”
Waiting until the share stood 5% above the insiders' price and was still rising, then holding for 60 trading days, earned 1.30% a trade after all costs, but 0.32% less than the S&P 500 over the same days, in both halves of the test. Shares no insider had bought, traded the same way, did about as badly. Buying at once did only a little better than the index: 0.28% a trade.
What insider buying is
Insiders are a company's officers and directors: the chief executive, the finance chief, the board. They may buy their own company's shares, but in the United States they must report each trade to the SEC on a Form 4, and the reports are public. A purchase with their own money is read as a vote of confidence from the people who know the company best. Is insider buying a good sign for the share, then? Many trading sites list the latest insider purchases for that reason.
A popular refinement says: do not copy at once. Wait until the share starts to rise, so the market confirms the insiders were right, and give up the bottom to avoid their failed bets. We tested that rule, and buying at once beside it.
How we tested it
We took the companies in the S&P 500 as the index stood on each day, and every open-market purchase their officers and directors reported on a Form 4, leaving out large holders who own 10% or more. One company's purchases on one filing day made one event. We kept the events worth $100,000 or more at the market price, filed from January 2010 to March 2026: 3,638 events at 488 companies.
The waiting rule watched each share for 60 trading days after the filing. It bought at the first close at least 5% above the close on the insiders' buying day, if the share was also above its average of the last 20 days. It then held for 60 trading days, with a stop-loss below the entry, and one trade at a time in each company. The plain rule bought at the open on the first day after the filing and held the same way.
Every trade paid 6 US cents a share for dealing, and the result in money also paid a financing charge each night, as on a share CFD. We compared each trade with SPY, the S&P 500 index fund, over the same days, dividends included. And we ran the same waiting rule 200 times on shares of the same index that no insider had bought in the 90 days before. The rules for a pass were written down before any result was read.
What we found
The waiting rule made 2,042 trades and earned 1.30% a trade after all costs. That looks like a profit, but the index rose too: over the same days, the trades did 0.32% worse than SPY. The gap was the same in both halves, −0.32% from 2010 to 2015 and −0.32% from 2016 to 2026. Shares no insider had bought, traded by the same rule, trailed the index by 0.43% in the middle of the 200 draws, so the insiders' shares did not stand out from them.
Buying at once, at the first open after the filing, made 2,742 trades and earned 2.19% a trade after all costs, 0.28% more than the index. That is the better of the two ways, but the margin is thin: +0.05% a trade from 2010 to 2015 and +0.37% from 2016 to 2026, before financing.
An average trade against the S&P 500 over the same 60 trading days, 2010–2026, %
| 2010–2015 | 2016–2026 | All | |
|---|---|---|---|
| Waiting rule: trades | 552 | 1,490 | 2,042 |
| Waiting rule: after all costs | +0.49% | +1.60% | +1.30% |
| Waiting rule: against the index | −0.32% | −0.32% | −0.32% |
| No insider bought: against the index | −0.07% | −0.56% | −0.43% |
| Bought at once: trades | 765 | 1,977 | 2,742 |
| Bought at once: after all costs | +1.19% | +2.58% | +2.19% |
| Bought at once: against the index | +0.05% | +0.37% | +0.28% |
Why waiting did not help
Waiting does what it promises: it avoids most of the purchases that go wrong. Counting every event on its own, 899 of 3,637 never rose 5% with the share still climbing within 60 trading days. Bought at once, those lost 6.68% against the index. The 2,738 that did confirm, bought at once, beat the index by 2.21%. But the waiting rule bought them only after the rise, and from there they trailed the index by 0.59%. Most of the gain the wait was looking for came before it bought.
What the wait gives up: each event on its own, against the index over 60 trading days, %
Bigger purchases, several insiders, other settings
None of them changed the answer. With the waiting rule, purchases of $500,000 or more trailed the index by 0.25% a trade (910 trades), and events where two or more insiders bought within 30 days trailed it by 0.59% (607 trades). Taking every purchase of $25,000 or more, the gap was −0.41%. We also set nine combinations of margin (2%, 5% or 10%) and holding time (20, 60 or 120 trading days) before the run; none beat the index in both halves.
The trades themselves were lopsided. With the waiting rule, 45% of trades won, by 12.94% on average, and the losers lost 8.10%; 48% ended at the stop-loss.
Year by year
Bought at once, the insiders' shares beat the index in 9 of the 17 years; with the waiting rule, in 6. The waiting rule's worst stretch came at the end: it trailed the index in every year from 2023 to 2026. The year 2026 counts only purchases filed to March.
| Year | Waiting: trades | Waiting | At once |
|---|---|---|---|
| 2010 | 78 | +0.95% | +0.64% |
| 2011 | 97 | −1.19% | −0.45% |
| 2012 | 82 | −0.03% | +0.47% |
| 2013 | 114 | −0.14% | +1.97% |
| 2014 | 82 | −0.43% | −0.83% |
| 2015 | 99 | −0.84% | −1.01% |
| 2016 | 131 | +1.11% | +2.91% |
| 2017 | 120 | −0.35% | −0.48% |
| 2018 | 128 | +0.14% | +0.28% |
| 2019 | 180 | +0.06% | −0.68% |
| 2020 | 185 | −0.44% | +1.76% |
| 2021 | 105 | +0.31% | +0.85% |
| 2022 | 148 | +0.82% | +0.72% |
| 2023 | 156 | −1.13% | −1.41% |
| 2024 | 105 | −1.65% | −1.08% |
| 2025 | 172 | −0.86% | +1.06% |
| 2026 | 60 | −3.00% | −2.30% |
Is insider buying a good sign, then?
Not one a trader can earn much from in large US companies. Buying S&P 500 shares right after their insiders bought beat the index by 0.28% a trade over 60 trading days, and by only 0.05% from 2010 to 2015, before financing. Waiting for the share to prove the insiders right turned that small lead into a shortfall of 0.32% a trade, in both halves, and the insiders' shares did not stand out from shares nobody had bought: those random shares did worse from 2016 to 2026, −0.56%, but better from 2010 to 2015, −0.07%.
The test covers only the largest US companies. The research that made insider buying famous found the effect mostly in small companies, and this test cannot speak for them. One more caution for anyone reading the forms: on 646 of 31,107 purchase rows with a price, the typed price lay a fifth or more outside that day's trading range, so we valued every purchase at the market's close.
Questions people ask
Is insider buying a good sign?
Only a weak one in large US companies. From 2010 to March 2026, shares of S&P 500 companies bought at the first open after an insider purchase of $100,000 or more beat the index by 0.28% over the next 60 trading days, on 2,742 trades, before financing.
Does insider buying mean anything?
It shows what the insiders did, not what the share will do next. Of 3,637 purchases, 2,738 (75%) saw the share rise 5% and keep climbing within 60 trading days, and 899 never did. Bought at once, the ones that never confirmed lost 6.68% against the index.
Does following insider trading work?
Not with the popular rule of waiting until the share rises. On 2,042 trades it trailed the S&P 500 by 0.32% a trade in both halves of 2010–2026, and none of nine combinations of margin and holding time beat the index in both halves.
Do bigger insider purchases work better?
Not on this evidence. With the waiting rule, purchases of $500,000 or more trailed the index by 0.25% a trade, and events where two or more insiders bought within 30 days trailed it by 0.59%.
Is insider buying illegal?
No. Officers and directors may buy their own company's shares on the open market, as long as they do not trade on important news the public does not have yet, and in the United States they must report the purchase on a Form 4, which the SEC publishes. Those public reports are the data of this test.
The bottom line
Copying company insiders once the share has started to rise did worse than simply owning the S&P 500: 0.32% a trade over 60 trading days, in both halves of 2010–2026. Buying at once did a little better than the index, by 0.28% a trade. In large US companies, insider buying is a weak sign. MYTH.
How we tested
- Belief
- Following company insiders' purchases, once the share starts to rise, beats the market.
- Data
- Form 4 filings to the SEC: open-market purchases by officers and directors, without holders of 10% or more, filed 1 January 2010 to 31 March 2026, at companies in the S&P 500 as the index stood on the filing day. One company on one filing day is one event: 3,638 events of $100,000 or more at 488 companies, valued at the market's close. Daily prices with dividends to 25 September 2026.
- Rule
- From the day after the filing, for 60 trading days: buy at the first close at least 5% above the close of the insiders' last buying day, with the share above its average close of the last 20 days. Stop-loss 3 times the average true range of the last 20 days below the entry; exit at the close 60 trading days after it. One trade a company at a time.
- Periods
- By the entry day: 2010–2015 (552 trades) and 2016–2026 (1,490 trades).
- Costs
- 6 US cents a share for the round trip; for the result in money also a financing charge each night of the 3-month Treasury bill rate plus 2.5% a year. The comparison with the index counts the dealing cost and leaves financing out on both sides.
- Compared with
- SPY with dividends over the same days; the same events bought at the first open after the filing; the same rule on a share of the index with no insider purchase filed in the 90 days before, drawn at random, 200 times.
- Pass rules
- Six conditions written before any result was read, among them: above zero after costs and above the index in both halves, above 95% of the random draws, better than buying at once, at least six of the nine margin and holding combinations above the index in both halves, and enough trades. Two were met: above zero after costs, and enough trades.
- Left out
- Form rows with the trade day after the filing day or more than 90 days before it, or with no share count (1,290); purchases with no close within five days of the trade day (948).
Insider purchases: SEC EDGAR Form 4 filings (public domain). S&P 500 membership as it stood: rebuilt from Wikipedia's list of S&P 500 component changes (CC BY-SA 4.0). Share and SPY figures are our own statistics from daily closing prices; no prices are shown and no insider is named.
This is research on past prices. It is not advice to buy or sell anything, and past results do not promise future ones.
Keep reading
More tests
- Test 006MythCopy the leaders with the best past year and drop the ones who fade. On 1,310 real-money records from 2023 to 2026, the best fifth lost less than the rest, but still lost 7.7% in the average quarter.
- Test 005MythDoubling the trade after every loss is meant to win everything back. On EUR/USD since 1999, about half of the martingale accounts were wiped out within a year.
- Test 004MythA famous 2003 study found stock markets did better on sunny mornings. In ten markets from 2004 to 2025 the sunny days did not win, and trading the weather lost money even before costs.