Keros Therapeutics (KROS) is a clinical-stage biotech with one internal program (rinvatercept, chimeric ActRIIA/IIB ligand trap, DMD/ALS) and one Takeda-licensed Phase 3 asset (elritercept, MDS). Stock $10.50 vs cash/share $14.27 — net cash $254.8M, EV -$48M. We had this DONE in March; three filings in the last 40 days changed the structural setup.

What the filings say

Q1 2026 10-Q (filed May 14): Cash $281.5M (-$5.9M Q/Q vs -$23.7M net loss — gap is a one-time $14.5M CRO refund from terminated cibotercept/elritercept contracts). True pre-Phase-2 quarterly burn ≈$17-18M. Runway "into first half of 2028" reconfirmed. ATM unavailable (not WKSI). Phase 2 DMD trial committed to Q3 2026 initiation with H1 2027 initial data — first time the data date is in a filed document. ALS regulatory engagement scheduled H2 2026. Poison pill expired April 9, 2026 — board did not renew (Note 6, no announcement).

Form 4 filed April 17: Jean-Jacques Bienaimé (director, former BioMarin CEO) bought 1,000 shares at $11.70 on April 15, code P. First open-market insider buy since the February 2026 C-suite selldown. Six days after the pill expired.

Schedule 13G filed May 15 (one day after 10-Q): Nantahala Capital Management at 6.90% (1,360,419 shares as of March 31, 2026). Filed passive, but Nantahala is a Connecticut small-cap special-situations fund, not pure index. Schedule 13G filed April 7: BlackRock at 10.1% — mechanical passive accumulation post the Pontifax/Adar1 capital-return exit.

What the market thinks

Stock at $10.50 = 26% discount to cash/share. Market is pricing the rinvatercept + ALS + Takeda + M&A bundle at -$3.77/share. Working backwards through the catalyst scenarios, market-implied weights look like ≈50-55% DMD clear-negative, ≈12-15% mixed, ≈5-7% positive, ≈3-5% M&A approach in 12 months.

Our weights, building cash + Takeda + ALS at $11.60/share base and probability-weighting the DMD outcome at 35% clear negative / 25% mixed / 10% positive / 30% delay plus 20% M&A optionality independently, yield forward EV ≈$14.10 by H1 2027. Gap to entry: 25-40% over ≈13 months. Range, not point — depends on how the M&A leg weights.

The cleanest disagreement is on M&A optionality: we're at ≈20%, market is implying ≈4%.

Why the gap exists

Three specific reasons, none of which require the market to be "wrong":

1. Time. Pill expired five weeks ago. Nantahala 13G was filed yesterday. Bienaimé buy was four weeks ago. Sell-side coverage updates lag; the consensus "no insider buying / pill in place" framing from the March DONE verdict is now empirically obsolete but hasn't propagated.
2. Anchoring. The bear narrative is anchored to myostatin pathway failures — BHVN taldefgrobep Phase 3 RESILIENT in SMA, plus 20 years of single-target anti-myostatin failures (stamulumab, trevogrumab, bimagrumab). KROS rinvatercept is a chimeric ActRII ligand trap — same modality class as Acceleron's approved Reblozyl and Winrevair, and as SRRK apitegromab (Phase 3 SAPPHIRE hit in SMA). The pathway is not foreclosed when paired with correct molecular design and indication. Less bearish than market, not bullish.
3. Cohort divergence un-resolved. XBI +69% 1Y, KROS -25% 1Y. Most of that is idio destruction priced (cibotercept failure Jan 2025, restructuring May 2025, TSA wind-down). With idio overhangs cleared and ALS optionality newly disclosed, the divergence may pause even without thesis confirmation.

Risks (ranked)

1. DMD trial outcome. Largest variance contributor. We have no edge on the biology — bear case from cross-ticker failures is real. Probability-weighted contribution to EV is small (+$1.58/share); the cash floor + Takeda residual cap downside at roughly $9-11/share.
2. Cash burn beyond guidance. Phase 2 enrollment + ALS prep could push quarterly burn from ≈$18M to $25-30M faster than guided. ATM unavailable — any raise needs full S-1/S-3 registration.
3. Hostile lowball. Pill is gone. Stock at deep cash discount. A below-fair-value bid could succeed if board pressure isn't there. Floor caps at roughly current price.
4. Sector beta drag. 44.3% idio variance — more than half of KROS variance is XBI/SPY. Long-only exposure is a partial biotech sector bet.

Catalysts

What would change our mind

• CEO/CFO open-market buy (code P) — bumps insider factor LR materially; the M&A optionality leg becomes a live thesis rather than positioning bet
• Any 13D filed — M&A factor flips from optionality (we weight 20%) to live (re-weight 40%+)
• Spike in Q2 10-Q "transaction and other costs" SG&A footnote vs Q1's $1.0M baseline — corroborates hidden strategic process per the "Hidden M&A via SG&A" pattern
• Phase 2 trial NOT initiated by Sep 30 — catalyst horizon slips into 2027; reduce
• Cash discount widens past 35% without new bear evidence — market sees something we don't; reduce
• XBI breaks down >15% — mechanical KROS drag; entry timing matters more

Sizing implication

44.3% idio variance is below the 75% target. If carried, this should be paired with an XBI short at ≈0.6× NMV to isolate the idio thesis — otherwise it's a closet biotech sector long with idio noise. Gross expected return ≈31% annualized, idio Sharpe ≈0.45-0.50 with hedge. Right-skewed distribution with cash floor at ≈$9-12/share by Q4 2026.

Not a Sharpe trade. An asymmetric tail-option trade with floor protection. Scaling entry across $10.50 / $9.50 / confirmation-breakout makes more sense than a full position today given negative momentum (-25% 1Y, -10% 1M) and thin insider corroboration (one director, $11.7K).

Evidence