Medicare Advantage Risk Modeling

Medicare Advantage Actuarial Cost and Revenue Model

Interactive · client-side only Tables only

Scenario

Maintain three presets: Baseline, Conservative, Improved. All inputs below are saved per scenario.

Revenue inputs

Risk adjustment (CMS-HCC v24 / v28 blend)

Year weights default to CMS phase-in pattern: 2024 ≈ 0.67/0.33, 2025 ≈ 0.33/0.67, 2026 = 0/1. Quality bonus and rebate share vary by Stars; enter values used in your bid book.

Clinical RAF and HRA/CR conversion

Literature suggests only ~53–64% of incremental RAF from HRAs and chart reviews maps to spending (Jung et al., 2023; Carlin et al., 2024). This block produces an “effective clinical RAF” that drives utilization.

Medical cost model

Enter events per 1000 member-years and allowed dollars per event. Program effect adjusts utilization by service class.

Service class Events / 1000 / yr Allowed per event ($) Program effect on utilization (%) PMPM ($)
Total medical cost PMPM $0.00

Revenue results (per member per month)

Benchmark adjusted for Stars$0.00
Blended normalized RAF (after 5.9% coding adj)$0.000
Risk-adjusted benchmark$0.00
Risk-adjusted bid$0.00
Rebate PMPM (if bid < benchmark)$0.00
Member premium PMPM (if bid > benchmark)$0.00
Total revenue PMPM (incl. member premium if selected)$0.00

When bid ≤ benchmark, CMS pays risk-adjusted bid and a rebate share of the gap. When bid > benchmark, CMS pays risk-adjusted benchmark and the enrollee pays the difference as premium.

Net results

Medical cost PMPM
$0.00
Non-claims PMPM
$0.00
Revenue PMPM
$0.00
Margin PMPM
$0.00
Admin load (% of revenue)0.0%
Admin dollars PMPM$0.00
Medical loss ratio (claims / revenue)0.0%
Operating margin (% of revenue)0.0%

Compare scenarios

Metric Baseline Conservative Improved

Sources and notes

Show citations

Implementation notes: This tool treats the Stars quality bonus as a multiplicative factor on the county benchmark and applies rebate percentages commonly used in practice. RAF blending follows a simple payment-weighted approach with normalization and a coding pattern offset applied multiplicatively. All are user-editable to match your official bid assumptions.