LIVE- METALS GOLD$- - SILVER$- - FOREX EUR/USD- - USD/JPY- - GBP/CAD- - CRYPTO BTC$- - ETH$- - XRP$- - SOL$- - MARKETS S&P 500- - DOW- - RUSSELL- - VIX- - SPXU$- - INDICATORS SPAXX 3.29% 7-day yield · Fidelity MMF 30YR MTG 6.65% Freddie Mac · weekly LIVE- METALS GOLD$- - SILVER$- - FOREX EUR/USD- - USD/JPY- - GBP/CAD- - CRYPTO BTC$- - ETH$- - XRP$- - SOL$- - MARKETS S&P 500- - DOW- - RUSSELL- - VIX- - SPXU$- - INDICATORS SPAXX3.29%7-day yield · Fidelity MMF 30YR MTG6.65%Freddie Mac · weekly
Insider Buying Strategy Backtest: 5,605 Cluster Buys Tested
INVESTING STRATEGY November 9, 2026

Insider Buying Strategy Backtest: 5,605 Cluster Buys Tested

Insider buying strategy, tested: 5,605 SEC cluster buys since 2016, bought the next day. They did not beat the S&P 500.

An insider buying strategy sounds like the cleanest edge a regular investor can get. When a CEO and two directors all buy their own company’s stock in the same month, they’re betting their money on something they understand better than anyone. And it’s all public: the SEC posts every purchase within two business days.

So we copied all of them. 5,605 cluster buys from 2016 to 2026.

They didn’t beat the S&P 500.

We ran this because insider-tracking apps and newsletters are everywhere right now, and the academic papers behind them are mostly old. We wanted the answer for today’s market, with real costs, compared against simply holding the index over the same dates. It’s the same test we ran on copying Congress’s stock trades, which also failed.

How we built the test

We pulled every Form 4 filing from the SEC’s public data sets, 2016 through September 2026. That’s 254,820 open-market purchases by officers and directors.

A cluster buy counted when 3 or more different insiders each bought at least $10,000 of their company’s stock, at $2 a share or more, within 30 days. That left 5,605 events across 2,807 companies.

For each one, we:

  1. Bought at the open the day after the filing, the earliest a follower could act.
  2. Held for 60 trading days, about three months.
  3. Compared it with the S&P 500 (SPY) over exactly the same days.
  4. Subtracted 0.5% for trading costs, both ways.

We locked all of these rules in writing before calculating a single return, so we couldn’t tweak them after seeing results.

Does insider buying beat the market?

No. Here’s what the cluster buys did against the S&P 500 over the same days, after costs.

Holding periodAverage vs S&P 500, after costs
20 trading days-0.09%
60 trading days+0.37%
120 trading days-1.19%

At 60 days, the median trade trailed the index by 1.96%, and only 45.5% of trades beat it.

The 60-day average is positive, barely, but it’s well within luck. The range of results we could reasonably expect runs from about -1.4% to +2.1%. And the typical trade lost to the index, by almost 2%.

We also checked whether pre-scheduled buying muddied the signal. Since 2023, insiders must flag purchases made under a 10b5-1 trading plan, which are set up months ahead and say little about today’s price. Only 16 of our events included one, so routine plan buying isn’t what dragged the results down.

Stronger clusters didn’t help either. Events with 5 or more insiders buying averaged -0.60%, though there were only 26 of them.

Where the small positive average came from

A few big winners. Cheap stocks priced $2 to $10 averaged +2.02%, but the median trade among them was -3.68%. That’s a lottery ticket pattern: most lose, a few win big.

The years tell the same story:

YearAverage vs S&P 500Beat the index
2016+5.21%58%
2019-3.94%38%
2020+6.91%50%
2021-4.24%41%
2023 to 2026-0.90%about 42%

Two years did most of the lifting: 2016 and 2020, when insiders bought heavily into the COVID crash and stocks snapped back. Since 2023, it’s been negative.

The catch that makes it look better than it was

Only 65% of the events had usable prices. Free price data tends to drop companies that went bankrupt or were delisted, so our test mostly saw the survivors. In 2016, fewer than half of the events could be priced. By 2026, 95% could.

That bias only pushes one way: up. The companies missing from our data are disproportionately the ones that failed. So the true result of following insider cluster buys is probably worse than +0.37%, not better. Our breakdown of backtest mistakes that fake an edge covers why this matters so much.

Why doesn’t it work anymore?

Our honest guess: everyone’s watching. The filings are free, fast and easy to scrape, and dozens of apps flag cluster buys within minutes. By the time a regular investor buys the next morning, the price has usually moved.

The academic studies that found an edge mostly used older data, when collecting Form 4s took real effort. The edge seems to have been paid out to whoever got there first.

What we’d do instead

  1. Treat insider buying as a research lead, not a trade signal. It can tell you where to look. It doesn’t tell you to buy.
  2. Ignore the average, check the median. A strategy where most trades lose and a few win big is a lottery ticket.
  3. Ask any signal seller one question: compared with what, over which dates, and after which costs?

If you still want to experiment, keep it small and run it next to a plain index fund for a year. Our portfolio stress test shows what a few bad picks do to the whole account.

This is education, not financial advice. Past results, including ours, don’t predict future returns.

Frequently asked questions

Does following insider buying beat the market?

Not in our test. We copied 5,605 cluster buys, where 3 or more officers or directors bought their own company's stock within 30 days, from 2016 to 2026. Bought the day after the filing and held 60 trading days, they beat the S&P 500 by 0.37% on average after costs. That's statistically indistinguishable from zero, and the typical trade lagged by 1.96%.

What is a cluster insider buy?

It's when several insiders at one company buy shares on the open market around the same time. In our test, that meant at least 3 different officers or directors, each buying $10,000 or more, within 30 days. The idea is that one insider can be wrong or routine, but several buying together suggests they know the stock is cheap.

How fast do insiders have to report stock purchases?

Within two business days. Company officers, directors and big holders file a Form 4 with the SEC, and anyone can read it on the SEC's website for free. Congress members get up to 45 days, which is one reason copying insiders was supposed to work better than copying politicians.

Why don't insider buying signals work for regular investors?

Our best guess is that the signal is public within two days and widely tracked, so prices adjust quickly. In our data the average was pulled up by a few huge winners, mostly in cheap stocks, while the typical trade lost to the index. The years with the most complete data, 2023 to 2026, came out negative.

What is survivorship bias in a stock backtest?

It's when your data only includes companies that survived. Free price data often drops stocks that were delisted or went bankrupt, so a backtest sees the winners and quietly skips the failures. In our test only 65% of events had usable prices. That bias pushes the result up, so the real result is likely worse than +0.37%.

RELATED READING

→ TradingAgents Review: The AI Bot's 26% Claim, Checked → Roth Conversion Ladder Example: $56,000 a Year at 4% → Tax-Loss Harvesting Example: $6,000 Loss, $1,728 Saved → How Much Interest Does $10,000 Earn in a Money Market?
#insider buying strategy#insider buying backtest#form 4 insider purchases#cluster insider buying#do insider purchases beat the market#copy insider trades#insider trading signals#survivorship bias backtest#SEC form 4#copy trading stocks
← Back to all posts