What Star Ratings cut points are

Cut points are the numeric thresholds that translate a measure's score into a 1 to 5 star rating. CMS sets them each year from the distribution of all contracts' scores for that measure, so a plan does not know the exact cut point in advance. A contract earns a given star on a measure only once its score crosses the threshold CMS draws for that star level.

Because the thresholds are derived from the whole field rather than fixed in regulation, the same raw performance can earn a different star from one year to the next. If competitors improve, the bar to hold a star rises. If the field slips, the bar can fall. That moving target is the core operational fact of Stars work.

How CMS sets cut points

For most measures, CMS uses a clustering algorithm. It takes every contract's score on a measure, plots the distribution, and groups the scores into five clusters so that scores within a cluster are close together and the gaps between clusters are wide. The boundaries between those clusters become the cut points for 1, 2, 3, 4, and 5 stars.

A simplified version makes the idea concrete. Suppose contracts on a measure score anywhere from 60 to 95. The algorithm finds the natural breaks in that spread and might draw the 4 star boundary at 82 and the 5 star boundary at 90. A contract at 81 sits one point under the line and earns 3 stars on the measure; a contract at 83 earns 4. The exact break depends entirely on where this year's field clusters, which is why the same 81 can be a 4 star one year and a 3 star the next.

The mechanics matter for planning. Cut points are set retrospectively, after the measurement year closes and every contract's score is in. The 2026 cut points reflect how the field performed on 2024 service, not a number any plan could have looked up while the work was still open. CMS publishes the final values in the Star Ratings Technical Notes alongside the release. CAHPS patient-experience measures follow a different statistical method, so the clustering, Tukey, and guardrail rules described here apply to the non-CAHPS measures that make up most of a contract's Stars work.

1 to 5
Star levels a cut point assigns from a measure's numeric score
3.0x IQR
Tukey outer-fence distance CMS uses to delete extreme outliers before clustering
5 points
Maximum a cut point can move year over year under the guardrail for most non-CAHPS measures

Tukey outlier deletion

Before clustering most non-CAHPS measures, CMS removes statistical outliers using the Tukey outer-fence method. It deletes contract scores that fall more than 3.0 times the measure-specific interquartile range below the first quartile or above the third quartile, then runs the clustering algorithm on the scores that remain. See the Tukey outlier deletion glossary entry for the full calculation.

The effect is to keep a handful of extreme scores from dragging the cluster boundaries up or down. A few very low outliers used to pull the lowest cut points down and make a 4 star easier to reach on some measures. Removing them produces cut points that sit closer to where the bulk of the field actually performs, and that move less from one year to the next.

The 5 percentage point guardrail

CMS also caps how far a cut point can move from one year to the next. Under 42 CFR 422.166(a)(2)(i) the guardrail limits the year-over-year shift on a non-CAHPS measure to 5 percentage points in either direction. For measures not scored on a 0 to 100 scale the cap is the restricted range version: 5 percent of the prior year's spread between the highest and lowest score, with outer-fence outliers excluded. The cap works together with Tukey deletion to make cut points more predictable.

The order of operations in the 2026 Technical Notes is Tukey outer-fence deletion first, then mean resampling with hierarchical clustering, then the guardrail applied to the resulting cut points.

Two exceptions matter more than the rule. CMS does not apply the guardrail to the Part C and Part D improvement measures, which carry a weight of 5. It also does not apply it to any measure that has been in the program 3 years or less, and a measure returning after a substantive specification change counts as new for this purpose. On those measures the cut point can move as far as the field moves.

For an operator, the guardrail is still the planning lever everywhere it applies. The worst the bar can get next year is the prior year's cut point plus the guardrail. A team that builds to that ceiling, rather than to last year's published number, protects the star even when the field tightens. On an unguardrailed measure there is no ceiling to build to, so the only safe posture is to sit well clear of the boundary.

The Reward Factor in 2026 and 2027

For the 2026 Star Year, CMS applies the Reward Factor, an integrated bonus of up to 0.4 added to a contract's overall and summary ratings for consistently high performance with low variance across measures. It rewards plans that are both high-scoring and steady. CMS describes the calculation in the 2026 Technical Notes.

The Reward Factor continues into 2027. The 2024 final rule had finalized an Excellent Health Outcomes for All reward, previously named the Health Equity Index, to start with the 2027 Star Ratings and to replace the Reward Factor. On April 2, 2026, in the Contract Year 2027 final rule, CMS finalized its decision not to implement that reward and to continue the historical reward factor instead. The current regulation reflects it: 42 CFR 422.166 carries the reward factor at (f)(1) with values of 0.4, 0.3, 0.2, 0.1 and 0.0, and there is no (f)(3). We cover the rest of the 2027 changes in 2027 Medicare Star Ratings changes.

Why you cannot wait for final cut points

Because cut points are set retrospectively from the field and can move up to the guardrail each year, you cannot wait for the final numbers to act. By the time CMS publishes the 2026 cut points, the 2024 measurement year is closed and the gaps are locked. The only workable approach is to manage to a margin: the prior year's cut point plus the guardrail, with a buffer on top.

The margin math is straightforward once you accept the uncertainty. If last year's 4 star cut point on a measure was 80, the bar next year could move up to 85 under the guardrail. A team that targets 80 risks losing the star on any upward move in the field. A team that targets 85 plus a small buffer holds it in almost every scenario. The cost of the extra gaps closed is small next to the revenue tied to a half-star swing on a weighted measure.

That means forecasting the glide path during the year, not auditing it after. A plan needs a running estimate of where each measure will land, how far it sits from the likely 4 star and 5 star thresholds, and how many gaps it has to close in the time left to clear the margin. The teams that hold their stars are the ones working that forecast every week while the year is open, not the ones reading the published cut points after the year has closed.

What to expect for the 2027 cut points

There are no 2027 cut points yet. CMS sets them from the measurement year 2025 field and publishes them in the 2027 Technical Notes at release. What is already settled is the methodology they will be produced with, and that is enough to forecast against.

  • Tukey outlier deletion is on. It has applied since the 2024 Star Ratings and nothing in the Contract Year 2027 rulemaking touched it.
  • The 5 percentage point guardrail is on. It stayed in 42 CFR 422.166(a)(2)(i) through the April 2026 amendments. CMS did not propose or finalize its removal.
  • Four measures have no guardrail. The respecified Colorectal Cancer Screening measure, Care for Older Adults Functional Status Assessment, Concurrent Use of Opioids and Benzodiazepines, and Polypharmacy in Older Adults all enter the 2027 ratings as new, so their cut points can move without a cap. The two weight-5 improvement measures never get one.
  • The reward factor is unchanged. CMS is not implementing the Excellent Health Outcomes for All reward and is keeping the historical reward factor, worth up to 0.4.
  • You get to check CMS's math. Beginning with the 2027 Star Ratings, CMS has to release de-identified contract-level sample data for one measure of each type during the second plan preview, so a plan can replicate the cut point calculation before the ratings publish.

On timing, CMS has not announced a date. The 2026 ratings went up on Medicare Plan Finder on October 9, 2025, and the 2025 ratings in October 2024, so expect the 2027 ratings in October 2026, ahead of Open Enrollment. Those ratings drive 2028 quality bonus payments. We will update this page with the published 2027 cut points on release day.

How Pelica helps

Pelica is the AI-native execution layer for value-based care: one live member record and a copilot next to every team. Most platforms show you what needs to happen. Pelica actually does it. For Stars, that gap between knowing the cut point math and clearing it is the whole problem.

The Quality & Stars copilot forecasts the glide path to each measure's likely cut points and works the open gaps while time remains, closing roughly 90% of BCS and KED gaps in-year. It does not stop at a gap list. It prioritizes the members who move a measure across its threshold, makes the outreach, and follows up until the gap is resolved or escalated. At HealthCare Partners, the largest IPA in the country, teams run this on 175,000+ patients live without adding headcount.

Sources