Verdict: KEEP | Weight: 0.42% (rank ≈42/50) | LR: 1.0

Factor Profile

Trailing 250-day regression against SPY, MTUM, XBI:

α = -3.4%, but t = -0.10. Statistically meaningless — SE is 34.6% over 250 days. The betas are the signal. The intercept is noise.

Anti-momentum drag estimate (15 weeks): If momentum continues at historical rates (SR ≈0.5), expected drag is ≈2.3%. Point estimate with huge confidence interval. Momentum reversal flips this to +2.3% tailwind.

What Consensus Thinks (And Why They're Right)

28 analysts. $80B cap. Every argument on both sides is fully priced.

The bull (75% of Street, mean target $873): Dupixent growth ($18.4B → $22B+ by 2030) offsets EYLEA erosion. Sanofi reimbursement step-up mid-2026 drives ≈19% EPS growth ($41.48 → ≈$52.81). Pipeline is deep (fianlimab, linvoseltamab, cemdisiran, obesity GLP-1+PCSK9). Fortress balance sheet ($16.2B cash). Buybacks shrink float 5-6%/yr. At 14x forward, you're paying value-stock multiples for double-digit growth.

The bear (nobody, really — the 25% are Holds): EYLEA franchise -27% YoY despite HD conversion. Biosimilar wave arriving H2 2026 (Amgen already at $700M/yr, Mylan/Biocon, Sandoz, Formycon, Celltrion all entering). Anti-momentum loading creates structural QQQ drag. Tax rate stepped up permanently (7.7% → 13.9%). Revenue essentially flat (+1% YoY) because EYLEA erosion eats Dupixent growth at the consolidated level.

Both sides are right. There is no disagreement on the trajectory. The only disagreement is whether 14x forward is cheap enough given EYLEA headwinds.

The Consensus I Stress-Tested

Went looking for the edge nobody found. Here's what the cross-reference showed:

Sanofi's CFO (Jan 29 call): Reimbursement to REGN running at €400M in 2026, up from the €300M originally modeled. His words: "Dupixent faster faster than, accelerate determination reimbursement earlier." If payoff pulls into early Q2 instead of mid-2026, there's a minor positive EPS timing surprise. Worth maybe 50bps. At 0.42% basket weight, that's 0.2 basis points of portfolio impact.

Amgen's EYLEA biosimilar (PAVBLU): $700M in FY2025. "Continues gain momentum... retina specialists responded positively to prefilled syringe format." Real share capture, plus antitrust counterclaims alleging REGN obtained EYLEA patents "fraudulently." Litigation timeline is years — not in our 15-week window.

EyePoint's DuraVu: Phase 3 data mid-2026 for six-month dosing anti-VEGF. Used EYLEA as control and DuraVu was numerically better (+2.1 letters vs EYLEA). Long-term threat to EYLEA HD's durability positioning. But filing → approval → launch = 2028 at earliest.

Sanofi's amlitelimab (OX40L): Phase 3 in atopic dermatitis with quarterly dosing. Sanofi's CEO called it "long-term immune normalization." REGN's founder Schleifer responded with an unusually extensive defense of Dupixent's safety profile on the Q4 call, then directly attacked OX40L: "somewhat disappointing, not only efficacy side, perhaps most importantly safety side." You don't pre-emptively defend your franchise unless you see the threat coming. But amlitelimab readouts are H2 2026, commercialization 2028+.

Dupixent competitive moat: Cross-corpus search found ≈10 companies pursuing IL-4/IL-13/OX40L pathways (KYMR, ZLAB, ABCL, CRVS, NKTR, LBTSF). AbCellera's assessment of their own OX40L asset: "probably important drug but not Dupixent-like." Bio penetration only 18% after 8 years — competition expands the market more than it steals share.

Options market: The near-term P/C ratio of 1.84 is noise (471 contracts, 6-day expiry). The post-earnings chain (May 15, 3,036 OI) shows P/C OI of 0.90 (neutral) with call volume running 3x put volume (bullish recent flow). ATM IV 48.6% vs 30-day HV 27.9% — pricing ±7-8% earnings move, consistent with history. Max pain at $760, market centered on current price. No unusual positioning. No directional bet.

Edge assessment: Zero. Everything I found, the market already knows. The Sanofi acceleration, the Amgen litigation, the amlitelimab threat, the biosimilar timeline — all mapped by 28 analysts covering an $80B name.

The Only Question: Momentum Regime

For 15-week basket filtration, REGN reduces to one variable:

This is a factor bet, not a stock decision. There is no idiosyncratic mispricing to exploit. The hold/remove decision depends entirely on your momentum regime view — and that view affects every anti-momentum name in the basket simultaneously, not just REGN.

Expected value of the decision:

At 0.42% weight: removing REGN captures ≈1 basis point if momentum continues, costs ≈1 basis point if it reverses. The decision is below the noise floor of the portfolio.

6 In-Window Catalysts (All Consensus)

1. EYLEA HD prefilled syringe FDA — Late April. Convenience upgrade, not clinical.
2. Q1 2026 earnings — Apr 29. Consensus $9.10. Management guided Q1 low (reauthorizations + $30M inventory absorption). Beat history: 3 of 4 quarters.
3. Fianlimab Phase 3 melanoma — H1 2026. Powered for non-inferiority to current standard of care. Yancopoulos: "hope achieve better." Carefully worded — not guiding superiority.
4. DB-OTO gene therapy FDA — H1 2026. 11/12 children with hearing gain. Tiny market (profound genetic deafness).
5. Sanofi reimbursement completion — Mid-2026 (possibly earlier per Sanofi CFO). ≈$7.50/share annualized EPS uplift.
6. Cemdisiran gMG submission — Q1 2026 (imminent). "Best result C5 inhibitors date" per management.

All known. All modeled. All priced at 28-analyst coverage depth.

Verdict: KEEP

Not because REGN is a great QQQ name. It isn't — it's a pharma stock in a tech index with anti-momentum loading that creates structural drag in trending markets.

KEEP because:

1. No impairment signal. α is statistically zero, not negative. The business isn't broken — Dupixent is one of the best drug franchises in pharma. EYLEA decline is priced and managed.
2. Weight is immaterial. At 0.42%, removing REGN frees ≈$4,200 of capital to redistribute. The expected value difference is ±1 basis point.
3. Natural hedge optionality. β_MTUM = -0.50 means REGN gains when momentum reverses. In a basket that's long QQQ survivors (inherently momentum-tilted), a small anti-momentum allocation provides portfolio-level insurance for free.
4. No edge means default to hold. Removing introduces active risk for no expected reward when edge is zero. At zero edge, the null hypothesis is: don't touch it.

If the basket needs 15+ removals from 50 (deeper cut), REGN is on the chopping block — but only because of factor profile, not because the business is impaired. Prioritize removing names with actual deterioration signals, not pharma misfits with quality franchises.

Process Note

Single-stock regression α is noise at standard sample sizes. 250 daily observations with 34.5% idio vol gives SE(α) ≈ 34.6%. You'd need |α| > 69% for significance at 95% confidence. Never lean on single-stock α from a 1-year daily regression. The betas are the signal; the intercept is not.