Will Oscar Health's FY2026 risk adjustment as % of direct premiums remain at or below 21%?

Management guides 20% of direct premiums for FY2026, and the threshold is 21% (100 bps buffer). In FY2025, risk adjustment was 18.5% after a 390 bps YoY increase. The question is whether a further 100+ bps increase beyond the already-elevated FY2025 level occurs. Oscar's continued growth into younger/healthier populations mechanically increases risk adjustment payable. However, the company is investing in Wakeley reporting and third-party data to improve estimation. The 21% threshold provides some buffer above the 20% guide. Near coin-flip because the structural forces pushing risk adjustment higher (market share growth, younger demographics) compete with improved estimation tools.

The 2025 experience shows 390 bps of YoY increase, from ~14.6% to 18.5%. Management now guides 20% for 2026, another 150 bps increase. But the 2025 miss was driven by market-wide morbidity shock from Medicaid redetermination. If that shock is now in the rearview mirror, the 2026 market may be more stable. However, the subsidy cliff is introducing a different kind of market disruption — healthier members leaving, potentially worsening the remaining pool's morbidity. This could push risk adjustment above 21%. The uncertainty is genuinely high. Slightly below 50%.

Management's 20% guide already builds in a significant increase from FY2025's 18.5%. The 21% threshold adds another 100 bps buffer. For risk adjustment to breach 21%, it would need to increase 250+ bps from FY2025's level — approaching the magnitude of the FY2025 surprise itself. While possible, a second consecutive year of that magnitude of surprise is less likely if Oscar has learned from the experience and improved its estimation process. Slightly above coin-flip for YES (staying below 21%).

This is the most genuinely uncertain market in the set. Risk adjustment is inherently unknowable until CMS publishes final data. The CFO himself calls it 'the most difficult thing we have to do.' The 21% threshold provides some buffer above the 20% guide, but the 2025 experience showed that 390 bps surprises are possible. If the market morbidity worsens from the subsidy cliff (healthier members leaving), risk adjustment could breach 21%. True coin-flip.

Management has proactively increased the risk adjustment assumption from 18.5% to 20%, showing they've incorporated the 2025 lesson. The 21% threshold requires the estimate to miss by 100 bps. While this happened in 2025 (actual exceeded prior year's level by 390 bps), the company is investing in third-party data (Wakeley) and has a better understanding of the market dynamics. The subsidy cliff may actually help if it removes the sicker Medicaid crossover population — reducing market morbidity. Slightly lean YES.

The structural dynamic is concerning: Oscar's market share nearly doubled (17% to 30%) with younger/healthier members, which mechanically increases risk adjustment payable. The larger the market share, the larger the payable. In 2026 with even higher membership (3.0M vs 2.0M), the absolute dollar risk adjustment is massive. If the remaining ACA pool (after subsidy cliff) is sicker on average, Oscar's healthier book creates an even larger transfer obligation. Slightly lean NO — the structural forces favor risk adjustment exceeding 21%.

Management guides 20%, threshold is 21% — 100 bps buffer. Management has learned from 2025 and invested in better estimation. But risk adjustment is structurally uncertain. Slight lean YES given the buffer.

Risk adjustment is the single hardest variable to predict. The 21% threshold provides some buffer but the 2025 experience showed estimates can miss by hundreds of bps. True coin-flip with low confidence.

Management has already built in a 150 bps increase from FY2025 to FY2026 guidance. For the market to resolve NO, risk adjustment would need to exceed 21% — meaning 250+ bps above FY2025's actual. While possible, this would be a truly exceptional year, exceeding the already exceptional 2025. The most likely scenario is that risk adjustment lands near 20% with normal variance. Lean YES.