Dislocation

Bottom line

Oscar fell 53.4% from $23.27 on 19 September 2024 to $10.85 on 30 March 2026, in three legs on two dated triggers: an industry-wide risk-adjustment shock announced on 1 July 2025, and the expiry of the enhanced ACA premium tax credits. Traded volume peaked at 15.5 times its pre-peak median in July 2025. The stock closed at $28.34 on 27 July 2026 — 161% above the low and 22% above the peak the drawdown started from. The dislocation happened; it is no longer available.

The drawdown, quantified

Source: daily closing prices as reported; drawdown peak, trough, and depth per the run's deterministic capitulation gauge (fit_features.capitulation_gauge.drawdown), which measures the deepest peak-to-trough move in the trailing two years.

The fall took 557 calendar days — 381 trading sessions — which is not the shape of a single repricing. A near-identical prior peak of $23.28 on 21 May 2024 sits just outside the gauge's window and would give a depth of 53.4%, so the measurement does not hang on the choice of September 2024. Against the post-IPO high of $36.77 on 10 March 2021, today's $28.34 is 22.9% lower; against the drawdown's own starting peak, it is 21.8% higher.

Source: month-end closing prices from the run's daily price history, as reported.

Three legs down, two of them on events

The decline is separable into a drift, two event legs, and two rebounds. The distinction matters to the framework, which treats a quiet 10–20% slide as drift rather than a moment.

Source: derived from the run's daily price and volume history; volume multiple is the 20-day average volume at each leg's end date divided by the 2.68 million-share median daily volume in the 180 calendar days before the 19 September 2024 peak, the same denominator the capitulation gauge uses.

The first 21.2% of the fall, from 19 September to 5 November 2024, came on 33 sessions at 1.2 times normal volume with no company disclosure attached — drift. The 5–7 November pair is different: the stock lost 25.6% across the 2024 election result and Oscar's third-quarter release the following morning [1], on roughly four times normal daily turnover.

The trigger

The dated adverse event is 1 July 2025, and it did not originate at Oscar. Centene announced that evening that its first read of 2025 Health Insurance Marketplace data from Wakely — an independent actuarial firm that aggregates market growth and morbidity information for individual-market carriers — covering about 72% of its membership implied market morbidity materially above its risk-adjustment assumptions. Centene later described the announcement as flagging "earnings pressure of $1.8 billion in 2025 as a result of a change in Marketplace risk adjustment transfer assumptions" [2].

The read-across was immediate and graded by exposure to the individual market. On 2 July 2025 Centene closed 40.4% lower, Molina 22.0% lower, Oscar 18.7% lower on 78.0 million shares — 29 times its pre-peak median daily volume — Elevance 11.5% lower, UnitedHealth 5.7% lower and Cigna 4.2% lower. Oscar fell 37.4% in the thirteen sessions from $21.44 on 30 June to $13.42 on 18 July 2025.

Oscar's own confirmation came three weeks after the price move. On 22 July 2025 it filed an 8-K announcing preliminary second-quarter results and "revising its full year 2025 outlook" [3]. Management set out the mechanism on the August call: the Wakely data through 30 April indicated "a meaningful market-wide increase in morbidity in 2025," impacting all carriers and lifting morbidity by mid- to high single digits across Oscar's markets [4]. In accounting terms it landed as an incremental $316 million increase to the 2025 risk-adjustment payable, recognised year-to-date in the second quarter, taking second-quarter MLR to 91.1% [5].

The size of the guidance move is the anchor for the Damage Math tab. Oscar entered 2025 guiding to earnings from operations of $225 million to $275 million on revenue of $11.2–11.3 billion [6]. The 22 July revision replaced that with a loss from operations of $200 million to $300 million on revenue of $12.0–12.2 billion [7] — a $500 million swing at the midpoint on a one-year operating-earnings line, against a $2.1 billion fall in market value over the same thirteen sessions on a constant 262.4 million-share count. The full year came in at a $396.4 million loss from operations and a $443.2 million net loss, $646 million below the original midpoint [8]. The FY2025 10-K attributes the 5.7-point MLR rise to "an increase in average market morbidity that resulted in an increase in the net risk adjustment transfer accrual, as well as higher utilization that was not fully offset by risk adjustment" [9].

A second, distinct trigger drove the third leg. The enhanced premium tax credits that underwrite individual-market affordability were scheduled to lapse at the end of 2025; management framed them as the difference between a farmer on $60,000 paying $75 a month and $300 a month for coverage [10]. On 10 November 2025, with the Senate advancing a shutdown-ending deal that did not extend the credits, Oscar fell 17.6% in one session on 49.2 million shares; on 24 November, on a report that a two-year extension would be proposed, it rose 22.3% on 58.2 million shares. Those two sessions are policy repricings, not company events, and no Oscar filing accompanies either — they are sourced here to contemporaneous market reporting rather than to a primary document.

The fear gauge

Source: fit_features.capitulation_gauge.volume_spike — maximum 20-day average volume in the peak-to-trough leg divided by the median daily volume in the 180 calendar days before the peak; derived from the run's daily price history.

Volume spiked 15.47 times. The 20-day average peaked at 41.5 million shares in the window ending 29 July 2025, against a pre-peak median of 2.68 million. That is emotion-priced selling by any reasonable reading — but the date is the part that matters.

Source: derived from the run's daily traded-volume history; pre-peak median daily volume was 2.68 million shares.

The capitulation sits at the July 2025 event, when the price was $13.84 and 40.5% below the peak — not at the $10.85 low eight months later. By 30 March 2026 the 20-day average had fallen to 6.8 million shares, 2.5 times the pre-peak median. The final seven sessions into the low took the price from $13.30 to $10.85, an 18.4% fall, on daily volumes of 8.3, 9.5, 6.8, 6.5, 5.1, 8.2 and 8.1 million shares — every one of them between 1.9 and 3.5 times the pre-peak median. The final leg was an orderly repricing on ordinary volume, not a second capitulation. On 27 July 2026 the 20-day average is 4.3 million shares, 1.6 times the pre-peak median — back to ordinary.

Who was selling

Reported short interest is the clearest positioning evidence, and it tracks the July 2025 event precisely. The run's own short-interest feed returned no rows for OSCR, so the series below comes from FINRA's semi-monthly reported short-interest filings as compiled by public market-data aggregators, cross-checked at two settlement dates against a second compiler; it is not a corpus document and carries no page citation.

Source: FINRA semi-monthly reported short-interest filings, as compiled by public market-data aggregators; percent-of-float figures vary by compiler because float definitions differ.

Short interest went from 12.0 million shares at the 30 September 2024 settlement — around 7% of float — to 62.5 million shares at the 15 July 2025 settlement, a 118% jump in a single fortnight and roughly 35% of float, the highest reading in the series. That is the same fortnight the volume spike peaked. It then fell steadily: 49.6 million by 31 August, 25.1 million by 31 December 2025, and 26.5 million at the 31 March 2026 settlement, about 11% of float. So the price trough was reached with short positioning already down two-thirds from its peak and roughly at its pre-shock level, and with turnover back near normal. Whoever marked the March 2026 low, it was not a crowded short book being pressed or a forced liquidation.

Direct evidence on forced or anchored sellers is thin. No index deletion, fund liquidation, or block-sale disclosure appears in the run's corpus for the drawdown window. The one large sponsor exit on record predates the drawdown entirely: Alphabet sold 6.5 million shares at $8.08 on 17 August 2023, thirteen months before the peak. Thrive Capital, Joshua Kushner's vehicle, amended its Schedule 13D on 13 November 2024, inside the first leg, and T. Rowe Price and Vanguard filed 13G amendments on 14 November 2025 and 27 March 2026 respectively, both inside the third leg; the run's ownership index records these filings but not their contents, so the direction of each change is unverified here.

The insider record has a hole and a signal. The run's Form 4 extract contains no transactions at all dated in 2024 or 2025 — 365 transactions covering 2021 to 2023 and 2026, none in between — so no claim can be made about insider trading through the fall itself. What the record does show, dated, is the other side: between 14 May and 1 July 2026, with the stock between $21.74 and $31.65, insiders sold 3.82 million shares for $109.4 million under 10b5-1 plans, of which CEO Mark Bertolini sold 2.45 million shares for $70.7 million across 25–30 June 2026 and co-founder Mario Schlosser 1.11 million for $32.4 million. Separately, a company announcement dated 18 November 2024 records a purchase of Oscar shares by Bertolini's foundation, inside the first leg and two sessions before the stock rose 12.7% on 19 November 2024. The buying is at the bottom; the selling, in size, is into the recovery.

Estimates against price

Dated consensus snapshots exist in the run's CapIQ extract only for FY2027 and FY2028, and only back to 28 January 2026. Within that window the sequencing is unambiguous, and it is the framework's signature: consensus was marked up while the price was still falling.

Source: consensus mean estimates from the run's CapIQ estimates extract (data/sp/estimates.json, revision snapshots at 180, 90 and 30 days and current), paired with the closing price on each snapshot date.

Between the 28 January and 28 April 2026 snapshots, consensus FY2027 EPS rose 45% (from $1.00 to $1.45), FY2028 EPS rose 75% ($1.08 to $1.88), and FY2027 revenue rose 39% ($14.1 billion to $19.6 billion). The price on 28 January was $14.87 and on 28 April $18.04 — but on 30 March, between those two marks, it printed $10.85, 27% below the January level. The estimate cycle turned up first; the price made its low afterwards and then ran past the estimates, gaining 65% between 28 April and 26 June against a 16% rise in the FY2028 number.

On the way down, no dated consensus history is available before January 2026, so the July 2025 sequencing has to be read from company guidance rather than from the sell side. There the order is clear enough: the stock fell 37.4% between 30 June and 18 July 2025, and Oscar's own guidance revision was published on 22 July [11]. The price moved on the industry read-across three weeks before the company confirmed the number.

Where the price sits now

The recovery ran ahead of, then with, the fundamentals. Oscar guided 2026 to revenue of $18.7–19.0 billion, MLR of 82.4–83.4% and earnings from operations of $250–450 million on 10 February 2026 [12], and reaffirmed it at the Raymond James conference on 2 March 2026 [13] and again on 8 June 2026 [14]. The stock still fell to $10.85 three weeks after that first reaffirmation. What ended the fall was delivery: on 6 May 2026 Oscar reported first-quarter MLR of 70.5%, 490 basis points better year over year, and $704 million of earnings from operations [15], with net income of $679 million, or $2.07 per diluted share, on 3.2 million members [16]. Consensus for that quarter was $1.10; the stock rose 10.6% on the day and a further 47% over the following seven weeks, to $29.16 on 24 June 2026.

At $28.34 the stock is 161% above its low, 22% above the peak the drawdown started from, 12% below its 2 July 2026 fifty-two-week high of $32.18, and 23% below its March 2021 post-IPO high. Short interest is 8.4% of float, turnover is 1.6 times its pre-shock median, and insiders have sold $109 million into the move. The conditions the framework's entry test looks for — a recent adverse event, a fear-priced volume spike, a price anchored to a cut it has not yet outgrown — were all present between July 2025 and March 2026. None of them is present on 27 July 2026.

One arithmetic note that carries into Fit: at $28.34 on the 262.4 million shares the feature file uses, market capitalisation computes to $7.44 billion. The share count is the FY2025 weighted-average basic figure; the fourth-quarter 2025 weighted-average basic count was 283.2 million [17], which would put the figure near $8.0 billion. Either way it sits below the $10 billion universe line, and it did so at every point in this drawdown.

What would change this read: a fresh dated adverse event — the June 2026 Wakely report landing worse than the pricing assumption, or a 2027 policy change to the individual market — that takes the stock down on a volume spike of the July 2025 order. The drawdown described here is history; the temporary-versus-permanent question about the damage it priced belongs to Damage Math.