Stoke Therapeutics (STOK) posted a third-quarter loss of $1.08 per share, missing the consensus estimate of a $0.88 loss by 22.7%. Earnings fell 66.2% year over year as the net loss nearly doubled to $73.6 million from $38.3 million. Revenue of $11.26 million beat the $8.28 million consensus by 35.9% and grew 5.9%. For a company whose value rests almost entirely on one pivotal readout, the miss matters less than what sits beneath it. Stoke is spending heavily ahead of a Phase 3 readout in the third quarter of 2027, and it secured an FDA agreement that makes that readout more likely to produce an approvable package. The central tension is that the regulatory story improved while the financial cushion got thinner.

The revenue beat deserves little weight. All of it comes from the Acadia and Biogen collaborations, so it reflects how contractual obligations are recognized rather than any commercial demand. Year-to-date revenue of $26.8 million compares with $183.0 million a year ago, when Stoke booked $150.8 million tied to the Biogen IP license. That comparison underscores how lumpy partner income can be. The spending side drove the miss:

• Research and development rose to $59.8 million from $37.7 million as EMPEROR advanced.
• Sales, general and administrative expenses jumped to $28.7 million from $16.0 million on personnel and launch-readiness costs.
• Total operating expenses reached $88.6 million versus $53.7 million a year earlier.

This is deliberate spending rather than an operating stumble. Even so, the CFO guided fourth-quarter spending to step up by a similar amount, so losses are still widening.

The call's most important development was regulatory. The FDA agreed that the key cognition-and-behavior secondary endpoint can be a multicomponent assessment averaging four Vineland-3 subdomains: expressive communication, receptive communication, interpersonal relationships and personal skills. Management said the composite carries no testing hierarchy and no alpha split, replacing the hierarchical subdomain approach discussed last quarter. Assuming 150 evaluable patients, management said the study is powered to detect a 46-53% seizure reduction and a 1.8-2.1 point difference on the composite. Management also does not expect an advisory committee, and it said the FDA treated inclusion of open-label extension data in the label as a review issue.

Trial execution looks clean. About 110 of 162 patients are through the Week 28 primary endpoint, up from roughly 60 last quarter and 18 the quarter before. About 20 have completed Week 52, all are continuing into open-label treatment, and no patient has discontinued.

Bears still have legitimate ammunition. In the 12-month Phase 1/2 analysis, three of the four individual Vineland subdomains missed statistical significance. Management cited a composite p-value below 0.01, but leaning on an averaged endpoint will draw scrutiny. Europe will not use the composite; Biogen will instead pursue a subdomain-based net treatment benefit, and it flagged that ex-U.S. pricing may depend on proving a cognition benefit. A rolling review is not guaranteed. STK-002 efficacy data for autosomal dominant optic atrophy have slipped from "end 2026/early 2027" to the first half of 2027, although the third dose cohort is now underway.

The cash story is the clearest negative. Stoke ended the quarter with $358.8 million, and share count growth this year shows the balance was replenished with equity. Runway language, however, has drifted:

• Mid-2028 a year ago.
• "Through potential U.S. launch in early 2028" last quarter.
• "Into early 2028" now.

With a launch targeted for early 2028 and spending still climbing, investors should expect a financing need near the commercial inflection point. Offsetting that, the commercial build has moved from planning to execution:

• A payer account team is deployed and a head of patient services has been hired.
• Top treatment centers are proactively requesting launch readiness.
• Claims data show 500-600 newly diagnosed Dravet patients per year.

Investor sentiment remains negative but improved meaningfully, rising from -0.39 to -0.22 since the prior report. The stock has not followed. Shares at $26.35 are down 8.4% from the open after the August report and sit about 15% below the 200-day moving average of $30.94. Earnings Whispers price and AVWAP trend readings are both negative, consistent with a market that has been reluctant to pay ahead of binary data despite the steady stream of de-risking news.

The bottom line is that the EPS miss is a byproduct of a company spending toward a pivotal readout and launch, not evidence of a broken plan. The FDA composite endpoint, the absence of an expected advisory committee and zero dropouts strengthen the zorevunersen setup. Shrinking runway language, a cognition case built on an averaged endpoint, and a share price still below its long-term trend mean the stock has yet to be rewarded for that progress.