Direct Answer
Estimate revision breadth is (up revisions − down revisions) / (up revisions + down revisions) × 100, a number from -100 to +100 that summarizes whether more covering analysts raised or cut their estimates over a defined window. It describes the net direction of analyst sentiment, not a prediction of future stock performance.
Key Takeaways
Estimate revision breadth answers a narrower question than it sounds like: not "are estimates going up," but "of the revisions that happened in a given window, what share leaned up versus down." It's a ratio built from counts of revisions, not from the size of those revisions, so a single small nudge and a single dramatic estimate cut count identically in the formula. That makes it a clean sentiment gauge and a poor magnitude gauge at the same time, useful for both, misleading if you expect it to do the other job.
- The formula only counts revisions that moved up or down; unchanged estimates are excluded from both the numerator and the denominator in this convention.
- +100 means every revision in the window was an increase; -100 means every revision was a cut; 0 means the count was even.
- Revision breadth aggregates the individual revisions covered in Swoopr's Earnings Estimate Revisions guide into one summary figure for a stock, sector, or index.
- The window length (30 days, 90 days, or another period) is a real methodology choice, a stock can show strongly positive breadth over 30 days and roughly flat breadth over 90 days for the same underlying estimate history.
- An alternative convention divides by all observations including unchanged estimates, producing a different number for the same underlying data, always confirm which convention a data source uses before comparing figures across providers.
- Positive or negative breadth is a description of sentiment direction, never a guarantee of how the stock will perform.
What Is Estimate Revision Breadth?
Estimate revision breadth is computed as:
breadth = (up revisions − down revisions) / (up revisions + down revisions) × 100
Every term in the formula is a count of revisions, not a dollar figure. "Up revisions" is the number of individual analyst estimate changes in the window that raised a forecast; "down revisions" is the number that lowered one. The result is bounded between -100 and +100 by construction: if every revision in the window pointed the same direction, the numerator and denominator become equal in magnitude and the ratio hits an extreme. A breadth of 0 means the up and down counts were exactly equal, regardless of how many revisions happened in total.
Why unchanged estimates are excluded
This formula, as implemented, deliberately excludes analysts who left an estimate unchanged from both the numerator and the denominator. That is one documented convention, not the only valid way to measure revision breadth. A second, equally legitimate convention computes "up share minus down share of all observations," dividing by the total number of analysts covering the stock, including the ones who didn't move their number at all. Because the denominators differ, the two conventions produce different numbers from identical underlying revision data, and neither one is more "correct" in the abstract. Whenever you see a revision breadth figure from a data provider, check which convention it uses before comparing it to a number computed a different way.
What breadth aggregates
A single earnings estimate revision, one analyst moving one EPS or revenue number, is the atomic unit this metric is built from. Revision breadth takes every one of those atomic events for a stock (or, applied across a group of stocks, for a sector or index) that fell inside a defined window and reduces them to one net number. The window is not a fixed, universal setting; a 30-day window and a 90-day window measure genuinely different things and can disagree with each other for the same stock at the same moment, because a burst of cuts early in a 90-day window can be offset by a run of increases in the most recent 30 days. State the window whenever you cite a breadth figure, "30-day revision breadth" is a complete statement; "revision breadth" alone is not.
Worked Examples
Illustrative figures, not live data.
The two examples below use the exact formula above with hypothetical revision counts, to show how the number reads at each end of the range.
Example 1: Strongly positive breadth
Suppose a stock had 15 up revisions and 3 down revisions over a 90-day window.
breadth = (15 − 3) / (15 + 3) × 100 = 12 / 18 × 100 = 66.67
A breadth of +66.67 indicates that, of the 18 revisions that occurred, a large majority, 15 of 18, were increases. This reads as strongly positive: most of the covering analysts who moved their estimate at all moved it up over the period measured.
Example 2: Strongly negative breadth
Suppose the same stock, in a different quarter, had 4 up revisions and 14 down revisions.
breadth = (4 − 14) / (4 + 14) × 100 = -10 / 18 × 100 = -55.56
A breadth of -55.56 indicates the opposite pattern: of 18 total revisions, 14 were cuts. This reads as strongly negative, most analysts who revised their number over the window lowered it.
Common mistake
The common mistake is reading these two examples as symmetric mirror images of the same underlying conviction. They aren't necessarily, 18 total revisions happened in each case, but the split (15 vs. 3 in the first, 4 vs. 14 in the second) tells you nothing about how large any individual revision was. A breadth of +66.67 driven by fifteen one-cent nudges is a very different situation from the same breadth driven by fifteen large upward revisions, and the formula can't distinguish between them. Breadth measures direction and consensus among revisions, not the magnitude of the underlying change.
How Should Revision Breadth Be Used?
Revision breadth is most useful as a descriptive summary of analyst sentiment momentum, layered alongside, not instead of, the individual revisions and the consensus estimate itself. A stock moving from flat breadth to strongly positive breadth over consecutive windows suggests the covering group's overall view is shifting upward, which is a different and often earlier signal than waiting for the consensus EPS number itself to visibly move. The same logic applies in reverse: deteriorating breadth ahead of a scheduled report has historically been treated by some market participants as one input into forming expectations around that report, though it says nothing on its own about whether the market has already priced the shift in.
Practical checklist
- Always state the window (e.g., "30-day breadth" or "90-day breadth") when citing or comparing a figure.
- Confirm which denominator convention a data source uses, including or excluding unchanged estimates, before comparing across providers.
- Read breadth alongside the size of the underlying revisions where available; two stocks with identical breadth can have very different magnitudes behind it.
- Treat breadth as a description of analyst sentiment direction, not a trading signal or a forecast of stock performance on its own.
Common mistake
The common mistake is treating any nonzero breadth reading as automatically meaningful. A breadth of +66.67 from 18 revisions and a breadth of +66.67 from 3 revisions (2 up, 1 down) are the same number but represent very different amounts of underlying analyst activity, a small sample size makes the ratio swing dramatically on just one or two revisions. Check the total revision count behind a breadth figure, not just the percentage itself.
Misconceptions Versus Reality
| Misconception | Reality |
|---|---|
| Revision breadth is one universal, standardized number | It depends on a documented convention (whether unchanged estimates are included in the denominator) and on the window length; different sources computing "revision breadth" can disagree for the same stock |
| Positive breadth means the stock will beat estimates or outperform | Breadth describes the direction analysts have been revising estimates, not what will actually happen at the next report or how the stock will trade around it |
| A bigger magnitude breadth number always reflects stronger conviction | Breadth only counts the number of up versus down revisions, not their size, a handful of small revisions can produce the same reading as a handful of large ones |
| Revision breadth and the consensus estimate measure the same thing | Consensus is the aggregated level of the estimate itself; breadth measures the recent direction of change in that estimate, which is a distinct signal that can move independently of the consensus level |
Risks, Limitations, and Exceptions
- Breadth is a count-based ratio; it does not capture the dollar or percentage size of any individual revision.
- With a small number of covering analysts, one or two revisions can swing the reading dramatically, making the figure noisy for thinly covered stocks.
- The unchanged-estimates-excluded convention used here is one documented choice; a different denominator convention produces a different number from the same underlying data, so figures are not directly comparable across data providers without checking methodology.
- Window length is a real methodology choice; a 30-day and a 90-day breadth figure for the same stock can disagree and both be "correct" under their own definitions.
- Revision breadth describes past and recent analyst behavior; it is not a forecast of future estimate direction or stock performance and should never be used as a standalone trading signal.
- The worked examples on this page use illustrative, hypothetical revision counts, not live data for any specific stock.
Aggregating Breadth Across an Industry or Sector
The same formula scales from a single stock to a whole industry or GICS sector: instead of counting up and down revisions for one company's covering analysts, count every up and down revision across every constituent company in the group over the same window, then apply the identical calculation, (up − down) / (up + down) × 100. A GICS sector such as Industrials with 70 constituent companies might generate several hundred individual estimate revisions in a 90-day window across all of those companies' covering analysts; rolling all of them into one net figure produces an industry-level or sector-level revision breadth reading.
This is a distinct signal from sector leadership breadth, which measures how many constituent stocks are participating in a sector's price move (commonly, the percentage trading above their own 50-day moving average). Sector-level estimate revision breadth instead measures how many fundamental estimate changes across the sector's analyst coverage were positive versus negative. The two can diverge: a sector's price breadth can be narrow (a handful of mega-cap names driving the index) while its estimate revision breadth is broadly positive across many smaller constituents, or the reverse. Reading both together, one price-based and one fundamentals-based, gives a more complete picture of whether a sector move is supported by improving fundamentals, not just improving prices.
Worked example: comparing two industries
Suppose a hypothetical Semiconductors industry group recorded 120 up revisions and 40 down revisions across its constituents over a 90-day window: (120 − 40) / (120 + 40) × 100 = 50.0. Over the same window, a hypothetical Regional Banks industry group recorded 30 up revisions and 90 down revisions: (30 − 90) / (30 + 90) × 100 = -50.0. The same formula, the same window length, and directly comparable results: analyst sentiment on balance improved across semiconductor coverage and deteriorated across regional bank coverage over that period. As with the single-stock case, the count-based ratio says nothing about the size of any individual revision, and a small constituent count in a narrowly defined industry group makes the reading more sensitive to any single company's cluster of revisions (for example, one company issuing guidance that triggers revisions from a dozen analysts at once).
Because industry and sector definitions vary by classification standard, always specify which taxonomy was used, see Swoopr's GICS Sector Taxonomy guide, since a broader sector-level grouping and a narrower industry-level grouping under the same standard can produce different breadth readings for what looks like "the same" area of the market.
Counting Analysts Without Weighing Them
Breadth counts analysts and does not weigh them. A large cut from a well-informed contributor and a token increase from a stale one each register as one vote, which is the measure's simplicity and its principal limitation. Reading it alongside the magnitude of the changes recovers most of what the counting throws away.
The window is doing more work than it appears to. A short lookback is dominated by the reporting calendar, and a long one dilutes recent changes with older ones the market has already absorbed. Fixing the window in advance, and holding it constant across comparisons, is what makes the series interpretable at all.
The denominator deserves attention too. Thin coverage produces a figure that swings between extremes on a single revision, and an extreme reading in that situation reflects the small count rather than any strength of conviction.
Aggregating breadth across a group compounds all of this, because the group reading becomes an average of measures that were individually unstable.
Frequently Asked Questions
What is estimate revision breadth?
Estimate revision breadth is a single summary number describing the net direction of analyst estimate changes for a stock over a defined window: (up revisions − down revisions) / (up revisions + down revisions) × 100. It ranges from -100, meaning every revision in the window was a cut, to +100, meaning every revision was an increase, and it describes analyst sentiment direction, not a prediction of stock performance.
How is revision breadth different from a single earnings estimate revision?
A single earnings estimate revision is one analyst changing one number, such as raising a quarterly EPS estimate from $1.10 to $1.15. Revision breadth aggregates every individual revision from every covering analyst over a chosen window into one net figure, so it answers a different question, not "what did one analyst just do" but "what is the covering group doing on balance, and how lopsided is it."
Does positive revision breadth guarantee the stock outperforms?
No. Positive revision breadth describes analyst sentiment direction, more analysts raising estimates than cutting them, not a guaranteed or even reliable predictor of future stock performance. A stock can have strongly positive breadth and still underperform if the improved estimates are already priced in, and it can have negative breadth and still rally if results beat the now-lowered bar. Treat it as one descriptive input, never a standalone signal.
Can revision breadth be measured for a whole industry or sector, not just one stock?
Yes. The same formula applies by counting up and down revisions across every constituent company in an industry or sector over the same window: (up − down) / (up + down) × 100. This industry-level or sector-level reading is a fundamentals-based signal, distinct from sector leadership breadth, which measures price participation across constituents. Because sector and industry definitions vary by classification standard, the taxonomy used (see Swoopr's GICS Sector Taxonomy guide) should always be specified alongside the figure.
What window length is appropriate for measuring revision breadth?
Common windows run from one to three months, and the choice is a trade-off. A short window responds quickly but can be dominated by a handful of revisions clustered after a single event. A long window smooths that noise but can still be reporting a reaction to news that is no longer current. The important discipline is fixing the window before looking at results, then keeping it constant, rather than selecting whichever length produces the cleaner picture.
How does thin analyst coverage distort a breadth reading?
The metric is a ratio, so with three covering analysts a single revision moves it in large increments and two revisions in the same direction produce a maximum reading. Those extreme values look identical to a strongly positive reading built from twenty analysts, but they describe far less agreement. Publishing the contributor count next to the breadth figure prevents a small-sample result from being read as a broad shift in sentiment.
Can revision breadth be positive while the consensus estimate falls?
Yes, because breadth counts analysts and ignores the size of their changes. Six analysts nudging their forecasts up by a fraction of a cent and four cutting theirs sharply produces positive breadth alongside a falling mean. This is the built-in limitation of any count-based measure. Reading breadth next to the actual change in the consensus figure catches the cases where direction and magnitude are telling different stories.
Should one analyst revising twice inside the window count twice?
Counting each revision event lets a single active analyst move the ratio more than a colleague who revised once, which pushes the measure away from describing how many analysts changed direction. Counting each analyst once, using their net change over the window, keeps the denominator equal to the covering universe. Neither convention is wrong, but they produce different numbers from the same underlying data, so the rule needs recording alongside the result.
What does a breadth reading of exactly zero indicate?
It means up revisions and down revisions were equal in number over the window, which describes a split rather than an absence of activity. That is a different situation from no revisions at all, where the formula has no denominator and the metric is undefined. Distinguishing a genuine even split from a period with no revision activity matters, because the two are frequently displayed the same way in a summary table.
References
The formula and worked examples on this page match the estimateRevisionBreadth function implemented and unit-tested in Swoopr's analyst-estimates calculation module. The two-convention distinction (excluding versus including unchanged estimates in the denominator) reflects a genuine methodology split used across sell-side and data-vendor practice; readers comparing figures across sources should confirm which convention that source documents. This guide describes long-standing, widely documented sell-side research conventions around how and why estimates get revised. Key reference sources include:
- U.S. Securities and Exchange Commission, Regulation FD: sec.gov: the fair-disclosure framework governing how companies communicate financial information relevant to estimate revisions to analysts and the market.
- CFA Institute, Equity Research and Valuation: cfainstitute.org: professional standards and methodology references for equity research practice, including revision-tracking conventions referenced in this guide.
This content was reviewed by the Swoopr Editorial Team in August 2026. The worked examples are illustrative and do not represent live or current data for any specific stock.