Do insider buys or analyst revisions predict stock returns?
By Mehmet Saruhan, Founder·Opulence Alpha Research · Published · Data through
Mostly no. Across 22 measures of Form 4 insider buying and selling on US stocks since 2003, none showed a reliable edge within three months; analyst upgrades and price targets, at most a faint tilt. The one exception is analyst coverage, measured since 2012: over three months, the least-covered fifth of stocks beat the median stock 52 times in 100, the most-covered fifth 49. A small tilt, not a forecast.
Keywords: insider buying, does insider buying predict stock returns, Form 4 insider purchases, CEO insider buying, insider cluster buying, analyst upgrades·JEL classification: G11, G12, G14, C12, C58·Concepts: Multiple testing, Point-in-time data, Backtest overfitting
- Universe
- The same for every study: 1,767 US common stocks in the 11 GICS sectors, 422 of them since delisted; S&P 500 members since 1996 plus large and mid-sized companies outside the index
- Period
- Every Wednesday from 4 Jan 1995 to 19 Aug 2026: 1,634 Wednesdays across 7,959 trading sessions
- Sample
- n = 1,763 stocks of the 1,767-stock universe; 1,889,758 stock-weeks, a median of 1,148 stocks per Wednesday
- Outcome
- Return over the next 1, 5, 21 and 63 trading sessions against the same day's median stock
- Inference
- t on non-overlapping dates; proven only when |t| ≥ 3 and the same sign in at least 60% of years
1Introduction
Insider purchases and analyst actions are among the most watched events in the market: an officer putting their own money into the company, or an analyst raising a rating, is read as information the price has yet to reflect. The literature is more measured. Seyhun, and Lakonishok and Lee, found that insider purchases were followed by better returns, mostly in smaller companies and over horizons longer than a few months; Hong, Lim and Stein found that news travels more slowly in stocks few analysts follow. This study measures both families on one universe, Form 4 insider trades since 2003 and analyst actions since 2012, against the same day's median stock from the next session to three months.
2Data and method
2.1Sample design
Population and frame. The population is US common stocks listed on the NYSE and Nasdaq; funds, ETFs, trusts, preferred shares, warrants and units are excluded. The sampling frame is a fixed universe of 1,767 companies, drawn once when the platform was built and not re-sampled since, in two strata. Stratum 1 is a census of the S&P 500: every company in the index at any time since 1996 whose price history could be recovered, 1,027 companies of which 413 have since delisted; it holds 76% of the index's members in 1996 and at least 96% in every year from 2010. Stratum 2 is 740 large and mid-sized companies outside the index, selected in proportion to the market's sector weights from the stocks that passed a minimum share price of $15 and a minimum average daily trading value of $25 million; 9 of them have since delisted.
Sample. The unit of observation is a stock-week: one stock on one Wednesday. A stock enters a Wednesday's cross-section when it has a valid close that day, a value of the signal and a measured outcome; bars flagged as bad data and returns that cross a change of issuer are left out. This study's sample is n = 1,763 stocks of the 1,767: 1,889,758 stock-weeks on 1,634 Wednesdays from 4 Jan 1995 to 19 Aug 2026, a median of 1,148 stocks per Wednesday (range 820 to 1,371). 331,954 stock-weeks (17.6%) come from the 422 companies that have since delisted. Every stock and every Wednesday carries equal weight.
Representativeness. Table 1 gives the sample by GICS sector beside the S&P Composite 1500: 7.9% of companies would have to change sector for the two to match exactly. By latest market value, 55% of the listed companies with a recorded market value are large (at least $10bn), 38% mid ($2–10bn) and 6% small. Because Stratum 2 was chosen from companies listed at construction, its history carries survivorship bias. Section 4 repeats the lead result among the stocks that were S&P 500 members on each date, the part of the sample largely free of that bias.
Companies in this study's sample, of them those since delisted, the size of the listed companies with a recorded market value, the sample's share of stock-weeks and the sector's share of the S&P Composite 1500.
| Sector | Sample | Delisted | Large | Mid | Small | Stock-weeks | S&P 1500 |
|---|---|---|---|---|---|---|---|
| Information Technology | 282 | 67 | 122 | 76 | 18 | 13.9% | 12.7% |
| Financials | 256 | 60 | 114 | 74 | 8 | 14.6% | 17.2% |
| Industrials | 255 | 53 | 126 | 67 | 11 | 15.9% | 17.5% |
| Health Care | 237 | 47 | 92 | 81 | 16 | 12.4% | 10.9% |
| Consumer Discretionary | 209 | 39 | 69 | 87 | 12 | 12.3% | 12.9% |
| Energy | 105 | 30 | 39 | 31 | 5 | 5.9% | 4.7% |
| Consumer Staples | 101 | 38 | 40 | 17 | 5 | 5.9% | 4.9% |
| Materials | 94 | 34 | 35 | 24 | 2 | 5.6% | 5.1% |
| Real Estate | 84 | 10 | 40 | 32 | 2 | 5.7% | 6.9% |
| Communication Services | 80 | 31 | 28 | 13 | 5 | 3.9% | 3.3% |
| Utilities | 60 | 13 | 35 | 12 | 0 | 4% | 4% |
| All sectors | 1,763 | 422 | 740 | 514 | 84 | 100% | 100% |
Sample: n = 1,763 of the 1,767 companies. Size by latest market value for the 1,338 listed companies with one recorded: large ≥ $10bn, mid $2–10bn, small < $2bn. S&P 1500 shares count the constituents of the S&P 500, MidCap 400 and SmallCap 600 (1,506 companies, lists read 25 Sep 2026). Sectors are each company's current GICS sector.
2.2Signals
| Signal | Definition | Since | Wednesdays |
|---|---|---|---|
| Purchases as a share of insider trades, 90 days | Open-market insider purchases as a share of all open-market insider trades, from Form 4 filings once public | 30 Jul 2003 | 1,193 |
| A CEO or CFO purchase, 30 days | The CEO or CFO made an open-market purchase in the last 30 days | 30 Jul 2003 | 1,193 |
| Cluster purchases, 30 days | Three or more insiders made open-market purchases within 30 days | 30 Jul 2003 | 1,191 |
| Insider purchases, 90 days | The number of open-market insider purchases in the last 90 days | 30 Jul 2003 | 1,193 |
| Analyst coverage | Research firms that published any rating or price-target action on the stock in the last 180 days | 15 Feb 2012 | 748 |
| Net upgrades, 63 days | Upgrades minus downgrades, as a share of all rating actions | 15 Feb 2012 | 748 |
| Gap to the median price target | How far the median analyst price target sits from the price | 15 Feb 2012 | 748 |
2.3Measurement
Outcome. The return from the close on the Wednesday to the close 1, 5, 21 and 63 trading sessions later, on closes adjusted for splits and dividends, compared with the same day's median stock; half of all stocks beat the median by construction, so chance is 50 in 100 on every date. No delisting return is added.
Each Wednesday the stocks are ranked on the signal. The study reports how often each fifth of that ranking beat the median stock, and the rank correlation (IC) between the signal and the return that followed.
The t-statistic uses non-overlapping dates only. A result is called proven when |t| is at least 3, it held in at least 60% of years and it is large enough to matter; with 1,470 tests across the studies, a looser bar would pass dozens by luck.
Full data and methods3Results
Insider buying and selling, measure by measure
From Form 4 filings we built 22 measures, counting only open-market purchases and sales (grants, option exercises and gifts are left out), each from the session after the filing was public: how many insiders bought or sold and for how much, whether a CEO or CFO bought, whether several insiders bought at once, and how long since the last trade. Each was tested over the next session, week, month and three months: 88 readings in all, and 0 of them passed our bar for a measured tendency.
| Measure | Day | Week | Month | 3 months |
|---|---|---|---|---|
| Buying | ||||
| Purchases as a share of insider trades, 90 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| A CEO or CFO bought, 30 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Cluster buying, three or more insiders, 30 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Number of insider purchases, 30 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Value of insider purchases, 30 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Number of insider purchases, 90 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Value of insider purchases, 90 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Distinct insiders buying, 30 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Value of CEO or CFO purchases, 30 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Purchases weighted by role, 30 days | Faint | No measurable edge | No measurable edge | No measurable edge |
| Selling | ||||
| Number of insider sales, 30 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Value of insider sales, 30 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Number of insider sales, 90 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Sales weighted by role, 30 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Cluster selling, three or more insiders, 30 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Net buying and holdings | ||||
| Purchases minus sales, by number, 30 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Purchases minus sales, by value, 30 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Net buying weighted by role, 30 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Change in insider holdings, 90 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| How recent | ||||
| Days since the last insider purchase | Faint | No measurable edge | Faint | Faint |
| Days since the last insider sale | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Days since any insider trade | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
Insider measures by their strongest reading at any horizon
The measures people ask about
Three measures come up most: buying as a share of all insider trades over 90 days, a CEO or CFO buying in the last 30 days, and cluster buying, when three or more insiders buy within 30 days. Over the next week, month and three months, each showed no measurable edge. The only tilt any of them showed was over the next session: faint, not reliable on its own.
What the faint readings say
Of the 13 faint readings, 11 were over the next session, and they ran slightly the other way: stocks with more insider buying did a touch worse the next day. The one longer hint: stocks whose insiders had bought more recently did slightly better over one and three months, the same way in 17 of 24 years over three months. Each is a faint tilt, not reliable on its own.
What this does and does not say
A Form 4 purchase is a real, dated event: an officer or director bought shares with their own money, and the filing says so. Our question is narrower. Once the filing was public, did stocks with more insider buying beat the median stock over the next day to three months? Across the stocks we cover, we could not measure that they did. This study does not test longer horizons, a single filing read in context, or companies outside our universe. Research, not advice.
More signals that describe a stock without predicting itAnalyst coverage: the one measured tendency
Coverage here is the number of research firms that published any rating or price-target action on a stock in the last 180 days, reiterations included. Every Wednesday since 2012 we split the stocks into fifths by it. Over the next three months, the least-covered fifth beat the median stock 52 times in 100 and the most-covered fifth 49 times in 100. On average the least-covered fifth earned +1.56 percentage points against the day's average stock, the most-covered fifth -0.72. Measured across every stock, not only the two end fifths, the relation kept its direction in 14 of 15 years; the gap between the two end fifths alone pointed the same way in 10 of 15. Over a day and a week it showed no measurable edge; over a month, a faint tilt. It is a tilt across hundreds of stocks and a nudge for any one of them. Coverage tends to rise with company size, and this study did not separate the two.
Stocks in fifths by the number of research firms covering them; lowest = fewest. Out of 100; the dashed line is chance.
Percentage points against the same day's average stock, per fifth; lowest = fewest research firms.
For the curious: the rank correlation (IC) between coverage and the next three months' excess return is -0.029, with a t-statistic of -3.4 on non-overlapping dates.
Upgrades, price targets and estimate revisions
Net upgrades over the last 63 days (upgrades minus downgrades, as a share of all rating actions) showed a faint tilt over the next week, in the direction the upgrades pointed, and no measurable edge over the next day, month or three months; over the week it kept its direction in 9 of 15 years. Net price-target raises over the same 63 days and the gap between the median price target and the price showed no measurable edge at any horizon. How long since any analyst last acted showed a faint tilt over three months, the same way as coverage: names analysts had left alone longer did slightly better. Earnings-estimate revisions could not be measured yet: we have collected them for too short a time.
| Measure | Day | Week | Month | 3 months |
|---|---|---|---|---|
| Research firms covering the stock, 180 days | No measurable edge | No measurable edge | Faint | Proven |
| Net upgrades, 63 days | No measurable edge | Faint | No measurable edge | No measurable edge |
| Net price-target raises, 63 days | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Gap to the median price target | No measurable edge | No measurable edge | No measurable edge | No measurable edge |
| Days since the last analyst action | No measurable edge | No measurable edge | No measurable edge | Faint |
| Earnings-estimate revisions, 30 days | Not enough history | Not enough history | Not enough history | Not enough history |
4Robustness
Table 3 re-measures the two lead relations. Analyst coverage over three months kept its negative sign in 10 of 11 sectors and in 2 of the decades its history covers (-0.029, t -3.4), but among S&P 500 members on the date it is not measurable (0.001, t -0.3): the tilt rests on the smaller, less followed companies outside the index, consistent with coverage standing in partly for company size. The number of insider purchases over 90 days showed no measurable edge over the next month, overall (0.003, t 0.7) or among S&P 500 members (0.001, t 0.2).
Rank IC and its t-statistic on non-overlapping dates.
| Sample | IC | t | Wednesdays |
|---|---|---|---|
| Analyst coverage, next three months | |||
| All stocks, whole period | -0.0291 | -3.4 | 735 |
| By decade: 1995–2004 | |||
| By decade: 2005–2014 | -0.0314 | -1.2 | 148 |
| By decade: 2015– | -0.0286 | -2.1 | 587 |
| S&P 500 members on the date | 0.0011 | -0.3 | 735 |
| Sectors with the overall sign | 10 / 11 | ||
| Insider purchases, 90 days, next month | |||
| All stocks, whole period | 0.0027 | 0.7 | 1,189 |
| By decade: 1995–2004 | 0.0199 | 1.4 | 75 |
| By decade: 2005–2014 | 0.0007 | 0.3 | 518 |
| By decade: 2015– | 0.0023 | 0.8 | 596 |
| S&P 500 members on the date | 0.0011 | 0.2 | 1,189 |
| Sectors with the overall sign | 10 / 11 |
5Limitations
- The universe and period are the same as in every study, but the data starts later: insider trades with Form 4 filings from 2003, analyst actions in 2012. Both families therefore cover fewer years, and each reading starts the session after the filing or action was public.
- Horizons stop at three months. The insider literature finds its effects mostly over longer horizons, which this study does not test, and it does not separate analyst coverage from company size.
- Before costs. Averages exclude trading costs, taxes and market impact.
- Same-close timing. Returns start at the close the signal is computed from; a real trade would start later.
- Survivors among smaller companies. The non-index names were chosen from companies listed when the universe was built; no delisting returns are added.
6Conclusion
Within three months, neither insider purchases nor analyst upgrades pointed to stocks that beat the median stock by a measurable margin: 0 of 88 insider readings passed the bar. The one measured tendency is analyst coverage, and it favours the least-covered stocks, not the most; it is small, not measurable among S&P 500 members and may partly stand in for company size. A Form 4 purchase remains a real event; over these horizons it did not add up to a measurable edge.
References
- Lakonishok, J. and Lee, I. (2001). Are Insider Trades Informative? Review of Financial Studies 14(1).
- Seyhun, H. N. (1986). Insiders' Profits, Costs of Trading, and Market Efficiency. Journal of Financial Economics 16(2).
- Hong, H., Lim, T. and Stein, J. C. (2000). Bad News Travels Slowly: Size, Analyst Coverage, and the Profitability of Momentum Strategies. Journal of Finance 55(1).
- Benjamini, Y. and Hochberg, Y. (1995). Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society, Series B 57(1).
- Grinold, R. C. and Kahn, R. N. (2000). Active Portfolio Management, 2nd ed. McGraw-Hill.
- Harvey, C. R., Liu, Y. and Zhu, H. (2016). … and the Cross-Section of Expected Returns. Review of Financial Studies 29(1).
- Shumway, T. (1997). The Delisting Bias in CRSP Data. Journal of Finance 52(1).
Appendix A. Questions readers ask
- Does insider buying predict stock returns?
- Not in a way we could measure. Across 22 measures built from Form 4 open-market purchases and sales on US stocks since 2003, none showed a reliable edge over the next day, week, month or three months: 0 of 88 readings passed our bar. 11 measures showed a faint tilt somewhere, not reliable on its own.
- Does CEO insider buying predict stock returns?
- In this data, no. When a CEO or CFO had bought shares in the open market in the last 30 days, the stock showed no measurable edge over the next week, month or three months, measured on US stocks since 2003. The only reading was a faint one-session tilt, not reliable on its own.
- Is cluster buying by insiders a reliable signal?
- Not one we could measure. When three or more insiders bought within 30 days, the stock showed no measurable edge over the next week, month or three months on US stocks since 2003; over the next session, a faint tilt that was not reliable on its own.
- Do analyst upgrades predict stock returns?
- Barely. Net upgrades over the last 63 days, upgrades minus downgrades as a share of all rating actions, showed a faint tilt over the next week and no measurable edge over a month or three months, on US stocks since 2012. Net price-target raises and the gap to the median price target showed no measurable edge at any horizon.
- Do stocks with fewer analysts do better?
- Slightly, as a tilt across many stocks. Since 2012, the fifth of US stocks covered by the fewest research firms beat the median stock over the next three months 52 times in 100; the most-covered fifth, 49 times. Measured across every stock rather than only the two end fifths, the relation kept its direction in 14 of 15 years. Coverage tends to rise with company size, and this study did not separate the two.
- Does this mean insiders have no information?
- No. A Form 4 purchase is a real event, and this study asks a narrower question: whether, once the filing was public, stocks with more insider buying beat the median stock within three months. On US stocks since 2003, 0 of 88 readings passed our bar. Longer horizons and single filings read in context were not tested.
Data availability and citation
Every figure in this paper is quoted from one frozen snapshot, published as JSON with the sample description, the robustness results and the test counts. The same snapshot feeds the research console, so the two cannot disagree. The paper and the snapshot are licensed under CC BY 4.0: share and adapt them with credit to Opulence Alpha and a link to this page.
Mehmet Saruhan (2026). Do insider buys or analyst revisions predict stock returns? Opulence Alpha Research, Opulence Alpha Studies, 25 Sep 2026. https://opulencealpha.ai/studies/insiders-and-analysts
Research, not advice. A measured tendency across hundreds of stocks is a nudge for any one of them, never a forecast of its price.
More evidence
All studiesWhere this is read on the S&P 500 company pages
The company pages show each S&P 500 company's readings from the latest close; a ranking lists every company by one measure.