Executive Summary

RELX delivered strong Q4 2025 results (7% revenue, 9% profit, 10% EPS, 99% cash conversion) into a stock crushed 38% on AI disruption fear triggered by Anthropic's Claude Cowork launch. At $31.06, the stock trades at 14x forward P/E with analyst targets at $51.50 (+66% upside, 12-month horizon).

The mispricing: Market priced RELX as if Legal (20% of revenue) represents the entire company. Reality: STM is the second-largest segment at 32% of revenue, and management just upgraded the outlook from "good" to "good to strong" growth. Risk (35%, highest margin) is 90% machine-to-machine with AI as a tailwind, not a threat.

Alpha calculation:

• Gross return to target: +65.8% (12 months)
• Less risk-free rate (5%): 60.8%
• Less sector (XLI historically ≈8-10%): ≈52%
• Idiosyncratic alpha: ≈52% annualized
• Edge adjustment: 90% idio variance, conviction 65% (Legal risk is real), Edge ≈85%
• Net α ≈ 28-30% after conviction/edge adjustment

Factor profile: Beta 0.21, 90.1% idiosyncratic variance (well above 75% target), R² only 9.9%. This is genuinely company-specific, not market-driven.

Timing risk: Stock already bounced +7.8% on 2.6x volume post-earnings. RSI 23.5 suggests oversold bounce underway. Immediate panic entry window closed.

Bear case probability: 20-25% (not 10%) — Claude Cowork is real, Legal is 20% of revenue, zero disclosed Lexis+ AI retention metrics, and management interoperability strategy is untested.

Critical Correction: STM Is 32% of Revenue, Not 15%

Editor flagged: Original draft massively understated STM at ≈15%. Actual 2024 segment breakdown from 20-F:

Why this matters: STM is the second-largest segment, not a footnote. Management upgraded the outlook ("good" → "good to strong") based on:

• Submissions growth >20%
• Articles published >10%
• New products ramping (LeapSpace)
• "Improving momentum," "increased pace of introduction and rollout"

This is acceleration, not deceleration. The AI disruption narrative ignores that 67% of RELX revenue (Risk 35% + STM 32%) is either benefiting from AI or showing accelerating growth.

Factor Decomposition: 90% Idiosyncratic, Well Above Target

Regression results (250 days):

• Beta (SPY): 0.61
• Beta (XLI sector): -0.07
• Beta (MTUM momentum): -0.07
• Idiosyncratic variance: 90.1% ✓ (target >75%)
• R²: 9.9% (only 10% explained by factors)
• Idiosyncratic vol: 27.8%

Interpretation: This is genuinely company-specific. The -38% drawdown is NOT riding market beta or momentum — it's an idiosyncratic repricing on the AI-disruption narrative. Low correlation to SPY (0.61 beta) and negative correlation to momentum (-0.07) confirms this isn't a "crowded trade unwind."

Positioning implication: 90% idio variance means >75% of returns will come from company-specific factors (your edge), not market/sector moves you can't control. This passes the Paleologo purity test.

Alpha Calculation: 28-30% Net After Conviction/Edge Adjustment

Gross alpha (analyst consensus):

• Current price: $31.06
• Mean target: $51.50
• Timeframe: 12 months
• Gross return: +65.8%

Adjustments:

1. Risk-free rate: -5% → 60.8%
2. Sector benchmark (XLI): -8% (historical) → ≈52%
3. Conviction: 65% (Legal displacement risk is real, 20% of revenue at risk)
4. Edge %: 85% (90% idio variance, but Legal moat defense unproven)

Net alpha: 52% × 0.65 × 0.85 ≈ 28-30% annualized

Sizing implication (Paleologo proportional rule): If portfolio Σ|α| = 200%, this position would be ≈14-15% of GMV at full conviction. But:

• Post-earnings bounce already underway (+7.8%, not buying panic)
• Legal risk unquantified (no disclosed Lexis+ AI metrics)
• Recommended size: 3-4% (starter, not full position)

Alternative: Wait for Q1 2026 results (late April) to validate Legal resilience. Sacrifice 5-10% upside but reduce risk of catching falling knife if AI displacement accelerates.

The Cross-Ticker AI-Disruption Pattern (7th Data Point)

RELX is the latest in a systematic mispricing pattern across proprietary data providers:

Pattern: Foundation model announces capability → data providers sold indiscriminately → strong earnings reveal narrative overstated → recovery.

RELX differentiation in basket:

• Best segment mix: Only 20% Legal (vs TRI heavily legal-focused), 35% Risk AI-beneficiary
• Cheapest quality multiple: 14x forward on 10% EPS growth, 99% cash conversion
• Management conviction: GBP 2.25B buyback (+50% YoY) at trough
• Highest idio variance: 90% (genuinely company-specific repricing)

Bull Case (60% Probability): Segment Mix Overwhelms Legal Risk

Risk segment (35%, highest margin):

• 90% machine-to-machine transactions (insurance underwriting APIs, not human research)
• Contributory data network effects: "incredibly difficult to replicate" (CFO Luff)
• Heavily regulated (data collection/usage restrictions create moat)
• AI is accelerant: New products using electronic health records (life insurance), aerial imagery (property), vehicle telematics
• Management quote: "The value at stake actually goes up" as driving becomes safer but more complex

STM segment (32%, just upgraded):

• Outlook moved from "good" to "good to strong" growth
• Submissions +20%, articles +10%, "momentum continuing into 2026"
• New products ramping (LeapSpace)
• "Science remains a totally global industry... information intensity continues to increase"
• Open Access transition creates new revenue streams, not cannibalization

Legal segment (20%, genuine battleground):

• Content moat: "Content-enabled workflows" backed by curated legal database
• Platform strategy: Open to licensing data via API to Claude Cowork (S&P Global playbook)
• Usage scale: "Hundreds of thousands of Lexis+ AI users" (real-time feedback loop)
• Strong new sales: All 3 sub-segments (Law Firms, Corporate, Government) showing momentum
• Management acknowledgment: Other tools exist, but differentiation is trusted content + scale

Valuation re-rating:

• Current 14x forward P/E implies permanent impairment
• Historical 25-30x during growth periods
• Quality compounder with 10% EPS growth, 99% cash conversion justifies 18-22x
• Implied upside to fair value: 28-57% (before alpha from Legal defense success)

Capital allocation signal:

• GBP 2.25B buyback (+50% YoY) at current prices
• Leverage at low end of 2.0-2.5x range (room to deploy)
• Management buying at 14x forward = credible cheap signal

Bear Case (20-25% Probability): Legal Displacement Faster Than Expected

Claude Cowork is not vaporware:

• Launched Jan 12, 2026, legal plugin added Jan 30
• Capabilities: Clause-by-clause contract review, NDA triage, compliance workflows
• These tasks currently drive Lexis+ subscriptions for in-house legal teams

"Hundreds of thousands of Lexis+ AI users" is unverified:

• No disclosed retention metrics
• No disclosed pricing (could be cannibalizing legacy at lower ASP)
• No disclosed contribution to 9% Legal revenue growth
• Won't see impact until renewal cycles complete (12-18 month lag)

Interoperability strategy is untested:

• Management "open" to licensing data via API to Claude Cowork
• Pricing model unclear (will it sustain 80bps margin expansion?)
• Competitive dynamics unclear (does Anthropic build own legal content?)
• Customer stickiness unclear (if research is API call, what's moat?)

Post-earnings bounce already underway:

• Stock +7.8% on 2.6x volume Feb 12
• RSI 23.5 → relief bounce, not panic pricing
• Max Pain $30.00 (3.4% below current) → downward pressure into Feb 20 OPEX
• This is "buy relief" entry, not "buy panic"

Cross-ticker pattern could be early:

• Just because SPGI, MORN, FDS, TRI haven't reported revenue loss YET doesn't mean they won't
• AI adoption in legal/finance accelerating (Anthropic enterprise push)
• Bear thesis is 12-24 month structural displacement, not immediate

Probability adjustment: Original 10% was too optimistic. With Claude Cowork live, Legal 20% of revenue, and zero disclosed retention data, 20-25% bear case probability is realistic.

Base Case (15-20% Probability): Legal Decelerates, Risk/STM Offset

Scenario:

• Legal growth decelerates from 9% to 5-6% as AI tools capture workflow share
• Risk and STM offset with 8%+ growth (Risk AI products, STM upgrade momentum)
• Group growth stabilizes at 6-7% (vs current 7%)
• Multiple stays compressed at 16-18x (vs historical 25-30x) until AI impact clarifies

Outcome:

• Stock grinds higher on buybacks and 3% dividend yield
• Reaches $38-42 over 12 months (+22-35% from current)
• No multiple re-rating, just FCF yield + modest growth
• Dead money risk if Legal deteriorates faster than expected

Catalyst to shift base → bull: Q1 2026 results (late April) show Legal new sales momentum sustained and Lexis+ AI retention strong.

Catalyst to shift base → bear: Q1 2026 shows Legal new sales deceleration or management discloses Lexis+ AI cannibalization.

Sizing and Entry Strategy

Immediate entry (3-4% position):

• Pro: 28-30% alpha, 90% idio variance, 14x forward on quality compounder
• Con: Already bounced +7.8% (not panic pricing), Legal risk unquantified
• Fit: If you have high conviction in Risk/STM offsetting Legal risk

Wait for Q1 2026 results (late April):

• Pro: Validates Legal resilience, reduces risk of falling knife
• Con: Sacrifice 5-10% upside if Legal holds up and multiple re-rates
• Catalyst: First data point on whether Legal new sales momentum sustained post-Claude Cowork

Tactical re-entry window (Feb 20 OPEX):

• Max Pain $30.00 (3.4% below current $31.06)
• If stock drifts to max pain into Feb 20, RSI resets
• Better entry than current post-bounce level

Position construction:

• Starter (2-3%): Enter now, size for Legal risk
• Add on confirmation (to 5-6%): If Q1 results show Legal stable
• Full position (7-10%): Only if Legal moat defense proven over 2-3 quarters

NOT recommended: Full 14-15% position (implied by 28% alpha) immediately. Legal displacement is real, unquantified, and represents 20% of revenue. Size for survival of bear case, not just expected value.

Investment Thesis Summary

Why this is interesting:

1. Segment mix mispricing: Market priced as if Legal (20%) is entire company, ignoring STM upgrade (32%) and Risk tailwind (35%)
2. Quality at cyclical pricing: 14x forward P/E on 10% EPS compounder with 99% cash conversion
3. Genuine idiosyncratic: 90% idio variance, -38% drawdown is company-specific repricing
4. Management conviction: GBP 2.25B buyback (+50% YoY) at depressed levels
5. Cross-ticker pattern: 7th data provider showing AI-fear selloff into solid execution

What has to go right:

1. Legal adapts to Claude Cowork via platform strategy (license data, don't fight interoperability)
2. STM upgrade momentum continues (submissions/articles growth sustains)
3. Risk AI products ramp (insurance EHR/imagery/telematics traction)
4. Multiple re-rates from 14x to 18-22x as Legal resilience evident

What could go wrong:

1. Claude Cowork displaces significant Legal revenue by 2027 (20% of revenue at risk)
2. Lexis+ AI retention weak, no disclosed metrics to verify (12-18 month lag to visibility)
3. Interoperability strategy fails (pricing unsustainable or Anthropic builds own content)
4. Post-earnings bounce continues, entry deteriorates from here

Recommended action: 3-4% starter position if entering immediately, or wait for Q1 2026 results (late April) to validate Legal resilience at cost of 5-10% upside. Size for surviving 20-25% bear case (Legal displacement), not just 60% bull case expected value.

Sources

• RELX Q4 2025 Earnings Transcript (Feb 12, 2026)
• RELX 2024 Annual Report (20-F filing, Feb 20, 2025)
• Factor regression analysis (250-day, SPY/XLI/MTUM, 90.1% idio variance)
• yfinance market data (Feb 15, 2026: $31.06, analyst targets $51.50)
• Worldview evidence: ev-0tp68o, ev-v0qvh7, ev-3ttbxr, ev-zvu5e5