Price Targets Move Twice as Often as Ratings

Upgrades and downgrades get the headlines. Price targets do most of the moving. Across 574 US stocks between 22 April and 22 September 2026, the mean analyst price target changed on 21.6% of stock-days. The consensus rating changed on 9.5%. Targets moved more than twice as often as ratings, and on 15.0% of all stock-days the target moved while the rating consensus did not move at all.

So a consensus rating that has not budged in a month is not evidence that nothing changed. On most stocks, something did.

What we counted

Every trading day we store a snapshot of the published analyst consensus for each covered stock: how many analysts sit in each rating bucket, the consensus rating score, and the mean, high and low price target. For this piece we compared each snapshot with the same stock's previous one. That gives 43,718 stock-day comparisons across 574 stocks and 106 trading days, with funds excluded.

A rating change here means a change in the consensus rating score; the underlying rating counts changed on a near-identical 9.8% of stock-days. A target change means any change in the mean target, however small.

The headline numbers

• Consensus rating score changed: 9.5% of stock-days
• Mean price target changed: 21.6%
• Target moved, rating did not: 15.0%
• Rating moved, target did not: 2.9%
• Per stock over the window: a median of 13 target changes against 7 rating changes
• Stocks with more target changes than rating changes: 89.7%

Most target moves are small

The raw count needs a qualification straight away. The median target change was 0.37%, and 79.0% of target changes were under 1%. That is what one analyst out of twenty nudging a number looks like after averaging.

Moves of 1% or more happened on 4.5% of stock-days. They were not balanced: raises of 1% or more came on 3.03% of stock-days, cuts of 1% or more on 1.51%. Twice as many raises as cuts. Across all target changes, of any size, 61.3% were upward.

Over the full five months the drift adds up. From each stock's first snapshot to its last, the median mean target rose 2.3%, and 62.5% of stocks ended the window with a higher target than they started with.

Why ratings look quieter than they are

Part of the gap is how the source reports ratings. Changes in the rating mix clustered at the start of each month: 17.1% of stock-days in the first eight calendar days showed a rating change, against 6.9% for the rest of the month. Target changes showed no such rhythm, 22.3% early in the month and 21.4% later.

That pattern fits rating tallies refreshed in monthly batches while targets update as they arrive. So the rating consensus is not only slower because analysts change ratings less often, which they do. It is also slower because it is published in steps. Anyone reading a rating consensus in the third week of a month is often reading something assembled at the start of it.

What a quiet consensus hides

Our view is that the target is where the movement shows, and the rating is where it gets confirmed, if it ever does.

An analyst who is warming to a stock can raise the target several times without touching the label, and the label may never change at all. We covered that pattern in how analyst upgrades and downgrades move markets: a raised target on an unchanged "hold" is a real update, even if it produces no upgrade headline. The data is consistent with that. On nine stocks in ten, targets moved more often than ratings, and the gap between the two is exactly the stretch where a rating consensus says "unchanged" while the numbers behind it drift.

The same logic applies to consensus estimates. A consensus figure is an average, and the average can shift a little every day through individual revisions that never show up as a headline. The dispersion and the direction of those small moves often say more than the level. Our earlier piece on what price targets are covers the other half of the caution: a target is a model output with wide error bars, not an appointment the price has agreed to keep. A rising average target describes what analysts are publishing. It does not describe what the stock does next.

Limits

• Snapshots, not notes. We see the published consensus once a day, not each analyst's note. Two revisions on the same day count as one change, and a revision that the source picks up late is dated late.
• Source rhythm. The monthly clustering of rating changes looks like a reporting cycle at the source. It inflates the apparent quietness of ratings mid-month.
• One period. Five months in a mostly rising market, which likely explains some of the tilt toward raises.
• Descriptive. No claim is made about what prices did after a target moved.

EquityBias is a market data research tool. Nothing here is financial advice. These figures describe how our data layers behaved over a fixed past period, not predictions about any stock or any future period.