Guide
News that moves stocks
Some kinds of news move a stock the same way time after time; most don't. Be The Puck sorted 24,802 earnings reports, analyst rating changes and company filings on 346 covered companies into 16 kinds, and measured how each stock moved against its market index at the first close after the news. Here is which kinds have a reliable direction, how big the move tends to be, and whether it keeps going.
Earnings reports mostly from 2014 on, rating changes from the last three years, company filings from the last two.
What the record says
Every number below comes from the events and prices as of the Oct 13, 2026 close.
- Four kinds of news cleared the bar for High predictability. After big beats, stocks did better than their index 66% of the time (median +1.9%), on 2,878 events; after upgrades, stocks did better than their index 77% of the time (median +1.4%), on 714 events; after downgrades, stocks did worse than their index 78% of the time (median −1.4%), on 636 events; after new coverage at Buy, stocks did better than their index 66% of the time (median +0.7%), on 628 events.
- With earnings, the size of the surprise matters. Big beats (20% or more above the estimate): 66% did better than their index, median +1.9%; earnings beats (5% to 20% above): 61% did better than their index, median +1.1%; small beats (under 5% above, or matched): 51% did better than their index, median +0.1%. 67% of the 13,802 reports scored beat or matched the estimate, so a small beat is close to what the market expects.
- Misses lean down, but less reliably than big beats lean up. After a miss of any size, stocks did worse than their index 62% of the time (median −1.2%, 4,577 reports). Misses of 10% or more did no better as a signal (63%) than smaller ones (62%).
- Price-target changes are weak signals. A raised target with the rating kept was followed by a gain against the index 55% of the time; a cut target by a drop 53% of the time. In the five sessions before a raise, the stock had already beaten its index by +2.1% on average; before a cut it had lagged by 1.4%. Targets tend to follow the price.
- Company filings have no reliable direction. Leadership-change filings: 56% went the more common way, 6% were big moves; deal filings: 55% went the more common way, 8% were big moves; financing filings: 60% went the more common way, 14% were big moves; other 8-K filings: 52% went the more common way, 10% were big moves. On an ordinary session, 6% of moves are that big.
- The reliable part is mostly in the opening price. Counted from the reaction session's opening price to its close instead of from the close before the news, the share going the expected way falls: downgrades 52% (from 78%), upgrades 51% (from 77%), big beats 55% (from 66%), new coverage at Buy 55% (from 66%). Acting at the open, once the news was out, was close to a coin flip.
- Earnings move stocks the most. 37% to 52% of earnings reactions were big moves (twice a normal day or more), against 6% of ordinary sessions, whichever way the surprise went.
- Does the move keep going? Average moves against the index: big beats +2.6% after one session, +3.0% after five (kept going); earnings beats +1.4% after one session, +1.7% after five (kept going); earnings misses −1.3% after one session, −1.3% after five (held); big misses −1.8% after one session, −2.0% after five (kept going); upgrades +1.8% after one session, +1.6% after five (gave some back); downgrades −1.7% after one session, −1.9% after five (kept going).
Every kind of news, most predictable first
▲ news expected to lift a stock, ▼ news expected to weigh on it, ◆ no expected direction. As expected: the share of events where the stock beat its index on the reaction session (▲) or lagged it (▼), from the last close before the news; for ◆, the share that went the more common way. Median: the middle day-one move against the index. 5-day avg: the average move against the index over five sessions. Big moves: twice a normal day or more (6% of ordinary sessions are that big). From the open: the as-expected share counted from the reaction session's opening price. Click a column to sort.Sort with the menu above the list.
| Kind of news | Events / stocks | As expected | Median | 5-day avg | Big moves | From the open | Predictability |
|---|---|---|---|---|---|---|---|
| ▼DowngradeAnalyst rating. A firm lowers its rating (say, from Buy to Hold). | 636 / 155 | 78% | −1.4% | −1.9% | 13% | 52% | High |
| ▲UpgradeAnalyst rating. A firm raises its rating (say, from Hold to Buy). | 714 / 163 | 77% | +1.4% | +1.6% | 14% | 51% | High |
| ▲Big beatEarnings report. Earnings per share 20% or more above analysts' average estimate going in. | 2,878 / 308 | 66% | +1.9% | +3.0% | 47% | 55% | High |
| ▲Initiated at BuyAnalyst rating. A firm starts covering the stock at Buy, Outperform or Overweight. | 628 / 158 | 66% | +0.7% | +1.1% | 6% | 55% | High |
| ▼Big missEarnings report. Earnings per share 10% or more below the estimate. | 2,705 / 278 | 63% | −1.3% | −2.0% | 37% | 56% | Moderate |
| ▼Earnings missEarnings report. Earnings per share below the estimate, by less than 10%. | 1,872 / 305 | 62% | −1.1% | −1.3% | 48% | 55% | Moderate |
| ▲Earnings beatEarnings report. Earnings per share 5% to 20% above analysts' average estimate going in. | 3,587 / 322 | 61% | +1.1% | +1.7% | 52% | 52% | Moderate |
| ◆Restructuring or write-downSEC filing (8-K). 8-K items 2.05 and 2.06: restructuring costs or a material impairment. | 11 / 8 | 64% up | +0.7% | +1.1% | 18% | 36% | Lowtoo few events |
| ◆FinancingSEC filing (8-K). 8-K items 2.03 and 3.02: new debt, or shares sold privately. | 43 / 20 | 60% up | +0.6% | +1.9% | 14% | 47% | Lowclose to a coin flip |
| ◆Leadership changeSEC filing (8-K). 8-K item 5.02: an executive or director arrives or leaves. | 504 / 116 | 56% down | −0.2% | +1.7% | 6% | 54% | Lowclose to a coin flip |
| ▲Price target raisedAnalyst rating. A firm keeps its rating and raises its 12-month price target. | 4,768 / 215 | 55% | +0.2% | +0.7% | 6% | 52% | Lowclose to a coin flip |
| ◆Deal or material agreementSEC filing (8-K). 8-K items 1.01, 1.02 and 2.01: a material agreement signed or ended, or an acquisition or sale completed. | 304 / 92 | 55% down | −0.3% | −0.9% | 8% | 53% | Lowclose to a coin flip |
| ▼Initiated at Hold or SellAnalyst rating. A firm starts covering the stock at Hold, Neutral, Sell or the like. | 343 / 142 | 53% | −0.2% | −0.2% | 4% | 50% | Lowclose to a coin flip |
| ▼Price target cutAnalyst rating. A firm keeps its rating and cuts its 12-month price target. | 2,458 / 197 | 53% | −0.1% | +0.1% | 6% | 50% | Lowclose to a coin flip |
| ◆Other material eventSEC filing (8-K). 8-K items 7.01, 8.01, 3.01 and 4.02: other disclosures the company deems material. | 591 / 99 | 52% down | −0.2% | 0.0% | 10% | 54% | Lowclose to a coin flip |
| ▲Small beatEarnings report. Earnings per share matched the estimate or beat it by less than 5%. | 2,760 / 305 | 51% | +0.1% | 0.0% | 51% | 50% | Lowclose to a coin flip |
Headlines: an early read
The same measure for 206 news headlines sorted by their wording. The headline archive only started recently, so most rows have too few events to grade; they will firm up as it grows.
| Kind of news | Events / stocks | As expected | Median | 5-day avg | Big moves | From the open | Predictability |
|---|---|---|---|---|---|---|---|
| ◆Layoffs Early readNews headline. A headline about job cuts or layoffs. | 3 / 3 | 100% down | −2.4% | −1.7% | 0% | 100% | Lowtoo few events |
| ▲Dividend raised Early readNews headline. A headline says the dividend was raised, or a special dividend declared. | 4 / 4 | 75% | +0.7% | – | 25% | 50% | Lowtoo few events |
| ▼Lawsuit or investigation Early readNews headline. A headline about a lawsuit, class action, investigation or probe. | 24 / 17 | 58% | −0.7% | −1.7% | 0% | 62% | Lowtoo few events |
| ◆Acquisition news Early readNews headline. A headline about an acquisition, merger or takeover involving the company. | 81 / 55 | 58% up | +0.3% | +0.1% | 7% | 54% | Lowclose to a coin flip |
| ▲Contract or partnership Early readNews headline. A headline about a contract or order won, or a partnership. | 20 / 17 | 55% | +0.2% | −2.8% | 5% | 65% | Lowtoo few events |
| ▲Beat, in the headlines Early readNews headline. A headline says results beat or topped estimates. | 6 / 6 | 50% | +1.7% | −1.6% | 17% | 50% | Lowtoo few events |
| ▼Dividend cut Early readNews headline. A headline says the dividend was cut or suspended. | 2 / 1 | 50% | −0.2% | −0.4% | 0% | 50% | Lowtoo few events |
| ▲Buyback Early readNews headline. A headline about a share buyback or repurchase plan. | 26 / 19 | 46% | −0.1% | +1.7% | 0% | 54% | Lowtoo few events |
| ▼Miss, in the headlines Early readNews headline. A headline says results missed or fell short of estimates. | 11 / 11 | 45% | +0.1% | −2.2% | 0% | 64% | Lowtoo few events |
| ▲Guidance raised Early readNews headline. A headline says the company raised its outlook or forecast. | 20 / 13 | 45% | −0.2% | +0.7% | 15% | 55% | Lowtoo few events |
| ▼Share offering Early readNews headline. A headline about selling new shares or convertible notes. | 3 / 2 | 33% | +0.4% | +2.2% | 0% | 33% | Lowtoo few events |
| ▼Guidance cut Early readNews headline. A headline says the company cut or withdrew its outlook, or warned. | 5 / 5 | 20% | +1.4% | −15.9% | 20% | 40% | Lowtoo few events |
| ▲Regulatory approval Early readNews headline. A headline about an FDA or other regulator's approval or clearance. | 1 / 1 | 0% | −0.5% | – | 0% | 0% | Lowtoo few events |
How events are sorted
Earnings reports come from Yahoo Finance's earnings calendar: the earnings per share each company reported against analysts' average estimate going in. The surprise sorts the report.
- ▲Big beat Earnings per share 20% or more above analysts' average estimate going in.
- ▲Earnings beat Earnings per share 5% to 20% above analysts' average estimate going in.
- ▲Small beat Earnings per share matched the estimate or beat it by less than 5%.
- ▼Earnings miss Earnings per share below the estimate, by less than 10%.
- ▼Big miss Earnings per share 10% or more below the estimate.
- ▲Upgrade A firm raises its rating (say, from Hold to Buy).
- ▼Downgrade A firm lowers its rating (say, from Buy to Hold).
- ▲Initiated at Buy A firm starts covering the stock at Buy, Outperform or Overweight.
- ▼Initiated at Hold or Sell A firm starts covering the stock at Hold, Neutral, Sell or the like.
- ▲Price target raised A firm keeps its rating and raises its 12-month price target.
- ▼Price target cut A firm keeps its rating and cuts its 12-month price target.
- ◆Leadership change 8-K item 5.02: an executive or director arrives or leaves.
- ◆Deal or material agreement 8-K items 1.01, 1.02 and 2.01: a material agreement signed or ended, or an acquisition or sale completed.
- ◆Financing 8-K items 2.03 and 3.02: new debt, or shares sold privately.
- ◆Restructuring or write-down 8-K items 2.05 and 2.06: restructuring costs or a material impairment.
- ◆Other material event 8-K items 7.01, 8.01, 3.01 and 4.02: other disclosures the company deems material.
Rating changes (Yahoo Finance, last three years) are kept to one event per kind per stock per day, however many firms acted. A rating change within 2 sessions of an earnings report is left out of the scores, since the report, not the analyst, usually moved the stock; the same goes for filings and headlines. That set aside 3,644 events, 3,120 of them price-target changes.
Company filings are 8-Ks from SEC EDGAR, sorted by their lead item. Toronto listings aren't in the filing archive.
Headlines from Google News are sorted by their wording. Questions, previews, lists and fund-holding filler are left unsorted.
- ▲Guidance raised A headline says the company raised its outlook or forecast.
- ▼Guidance cut A headline says the company cut or withdrew its outlook, or warned.
- ▲Beat, in the headlines A headline says results beat or topped estimates.
- ▼Miss, in the headlines A headline says results missed or fell short of estimates.
- ▲Buyback A headline about a share buyback or repurchase plan.
- ▲Dividend raised A headline says the dividend was raised, or a special dividend declared.
- ▼Dividend cut A headline says the dividend was cut or suspended.
- ▼Share offering A headline about selling new shares or convertible notes.
- ▼Lawsuit or investigation A headline about a lawsuit, class action, investigation or probe.
- ▲Regulatory approval A headline about an FDA or other regulator's approval or clearance.
- ▲Contract or partnership A headline about a contract or order won, or a partnership.
- ◆Layoffs A headline about job cuts or layoffs.
- ◆Acquisition news A headline about an acquisition, merger or takeover involving the company.
How a reaction is measured
The reaction session is the first session whose 4 pm New York close comes after the event. A report after the close, or on a weekend, is read on the next session.
The move is the stock's close-to-close change that session minus its index's (the S&P 500, or the S&P/TSX for Toronto listings). That strips out the day the whole market had.
Five days runs from the last close before the event to four sessions after the reaction, again against the index: did the move keep going or fade?
A big move is at least twice a normal day for that stock, where a normal day is the typical size of its moves against the index over the 60 sessions before.
The odds are a binomial test against a coin flip: how likely a record this lopsided would be if the news told you nothing about direction.
What High predictability means, and what it doesn't
High: moved the expected way at least 65% of the time, on at least 50 events, with less than a 1% chance of a record that lopsided from coin flips. Moderate: at least 57%, 30 events and under a 5% chance. Low: anything else, either because the split is close to 50/50 or because there are too few events to tell.
It describes past reactions, averaged across many stocks. A kind graded High tended to move stocks the expected way; any one stock, on any one day, can go the other way, and the median move is the middle of a wide spread.
It is not a forecast. Whether earnings beat or miss is only known once the report is out. The predictable part is the reaction once the surprise is known, and it is measured from the last close before the news, a price no one could trade at once the news was out. Counted from the opening price instead, the kinds graded High went the expected way 51% to 55% of the time: most of the reaction was already in the opening price.
Other limits. Events on the same day across many stocks are not independent, which makes the odds look better than they are. Analysts sometimes change a rating on the same day as other company news, which this can't separate. Yahoo's earnings time put the report on the same session as the company's 8-K filing for 870 of 875 reports where both exist; where they differ, the earlier time is used. Companies that were delisted or acquired are not in the sample. The 20% line for a big beat was drawn after seeing that larger beats ran more one-sided, so that row may flatter a little. Headlines are sorted by wording, which misreads some.
Each stock page shows its own record under What moved it lately: its biggest recent moves, the news on each, and how that kind of news has moved it and every other stock.
This analysis does not constitute trading advice. Please meet with an advisor or independently review sources before making any decision.