2 Beaten-Down Stocks to Avoid Right Now

Teladoc's Q2 revenue fell 4% year over year to $606.9 million, while its per-share loss widened to $0.21 from $0.19; BetterHelp revenue and paying users declined 12% and 11%, respectively. Despite 7% international revenue growth and expanded insurance coverage, competitive and regulatory risks leave the company's turnaround prospects weak after its stock lost 95% over five years. Recursion Pharmaceuticals remains a high-risk AI-drug-discovery play with no approved products or phase 3 assets, and its shares have fallen 21% over the past year as its AI platform has yet to validate superior clinical-development outcomes.

Analysis

TDOC's core issue is not simply weak demand but deteriorating unit economics in a low-switching-cost category. Insurance reimbursement can increase top-of-funnel utilization, yet it also concentrates payer bargaining power and may reduce net revenue per session; absent evidence that covered-member acquisition costs and therapist utilization are improving, broader coverage is more likely to delay than repair margin pressure. Hybrid providers and payer-owned behavioral-health networks have an inherent advantage in referral pathways and care-continuity, leaving TDOC vulnerable to further multiple compression if management cannot stabilize BetterHelp users over the next two quarters.

RXRX remains a venture-style clinical asset priced partly on platform optionality, rather than a proven productivity advantage. The relevant validation event is not additional AI partnerships or preclinical output, but human efficacy data that demonstrates a materially higher probability of technical success or shorter development cycle than conventional discovery. Until then, its cash burn and dilution risk should be evaluated against a catalyst calendar; the stock can rally sharply on AI-sector beta, but that would not constitute fundamental de-risking.

The more investable second-order expression is incumbent pharma: LLY and NVO can deploy AI against large proprietary datasets, established clinical operations, and funded pipelines, so even modest R&D-cycle improvement accrues to already monetized franchises. BMY offers a lower-expectations alternative if pipeline productivity improves, though its patent-expiry exposure means AI narratives alone cannot offset revenue-cliff risk. Consensus bearishness on TDOC is directionally justified, but the short is likely crowded after the drawdown; downside is best pursued only on a failed stabilization signal rather than at indiscriminate lows.

Key Decisions for Investors

• Maintain TDOC as underweight/watch-short, not an immediate momentum short. Add on a post-earnings relief rally only if BetterHelp paying users remain declining and management cannot show sequential improvement in revenue per user or adjusted EBITDA; use a 15-20% stop from entry, as strategic-sale or reimbursement-expansion headlines can create violent squeezes.
• Pair trade over 3-6 months: long LLY or NVO versus short RXRX, sized beta-neutral. The pair captures monetizable AI-enabled R&D productivity at scaled incumbents versus unproven platform valuation; reassess immediately upon RXRX clinical data showing convincing human efficacy or a non-dilutive, economically material partnership.
• For TDOC, monitor the next two quarterly disclosures for covered lives, BetterHelp conversion/retention, therapist utilization, and segment contribution margin. A sequential user stabilization plus improving EBITDA guidance would falsify the bear thesis and remove the short/watch position.
• Avoid treating RXRX as an AI basket proxy. Any long exposure should be catalyst-specific and small until cash runway, expected financing needs, and clinical readout timing are verified; a broad AI rally is a trading catalyst, not evidence of lower drug-development risk.