Bi-Monthly Investment Pipeline Report #

Cycle: 2026-04-03 | Universe: S&P 500

Executive Summary #

Screened 503 S&P 500 stocks through multi-framework pipeline (Graham 7-filter + Greenblatt Magic Formula + Schilit accounting quality + Lynch classification). 9 candidates passed with Graham score >= 5/7. Deep analysis via 9 parallel stock-data-analyst agents applied Dorsey moat, Klarman downside, insider/analyst/earnings assessment.

Bottom line: 1 actionable buy (AIG), 2 watchlist candidates (PHM, HUM), 6 hold/avoid.

The cycle produced a narrow but high-quality output. The US market at current levels offers few deep-value opportunities — most Graham candidates are cyclicals (homebuilders) facing rate headwinds or asset managers facing secular passive-flow pressure.

Composite Ranking #

Weights: Graham 30% | Magic Formula 20% (placeholder) | LLM Deep Analysis 30% | Schilit Clean 10% | Reddit Divergence 10%

LLM composite constructed from: Dorsey moat score, insider signal, earnings momentum, analyst sentiment, balance sheet quality, Klarman downside inversion, shr-034 deploy readiness.

*HUM forward P/E uses management's $9.00 guide, not consensus $15.29. BEN forward P/E assumes earnings normalization from $1.08 to $2.91.

Diversification rules applied:

• Consumer Cyclical: PHM + DHI kept, LEN dropped (worst earnings of the three)
• Financial Services: AIG + TROW kept, BEN dropped (highest Schilit flags, dividend cut risk)

Actionable Recommendations #

1. AIG — American International Group | BUY #

Conviction: HIGH | Only candidate passing shr-034 (RED flags flipping to GREEN)

Why now:

• Forward P/E 8.5x — cheapest in the screen
• CFO bought $232K at $86.54 (underwater, strong conviction signal per shr-002)
• Director Rice bought $770K at $77 (near current price)
• 4-for-4 earnings beats, +43% adj EPS growth in 2025
• $10B buyback authorization = 25% of market cap
• 17 consecutive quarters below 90% combined ratio
• Zero analyst sell ratings (8 Buy, 14 Hold)
• All PTs above current price — even the most bearish analyst ($79) sees upside

Graham IV sensitivity: | Growth | IV (trailing $5.43) | IV (forward $8.85) | MOS | |--------|---------------------|---------------------|-----| | 0% | $46 | $75 | 0% (fwd) | | 3% | $79 | $128 | +41% (fwd) | | 5% | $101 | $164 | +54% (fwd) |

Break-even growth (trailing): 2.7%. Break-even growth (forward): ~0%. The market prices zero growth on forward earnings — too pessimistic for a company buying back 25% of its own shares.

Position sizing suggestion:

• Satellite allocation: 15-20% of available capital
• 2-tranche entry: T1 now ($75 area), T2 on any pullback below $70 or after April 30 Q1 earnings
• If Q1 earnings beat again → add T2 regardless of price (confirms trajectory)

Key risk: Hurricane season (Jun-Nov). AIG's cat exposure means H2 2026 is the risk window. The buyback and earnings momentum provide upside before cat season. Consider trimming 1/3 if stock reaches $87-90 pre-hurricane season.

2. PHM — PulteGroup | WATCHLIST — Pending April 23 Earnings #

Conviction: MEDIUM-HIGH | Best homebuilder candidate, but tariff cost uncertainty

Deploy T1 IF April 23 earnings show:

• Gross margin guidance holds at 24.5-25% (not cut further)
• Tariff cost pass-through guidance is credible
• Order pace stable or improving
• Cancellation rate < 14%

Do NOT deploy if: Gross margin guided below 24%, tariff cost revision upward, or orders decelerating.

Why PHM over DHI/LEN:

• Best balance sheet: 0.10x net debt/EBITDA (vs DHI 0.66x, LEN 0.71x)
• Del Webb 55+ brand provides demographic tailwind and switching costs
• 4 consecutive adjusted earnings beats (vs DHI 2/4, LEN 0/4)
• Analyst support: 56% buy (vs DHI 29%, LEN 14%)

3. HUM — Humana | WATCHLIST — Binary Catalyst April 29 #

Conviction: HIGH RISK / HIGH REWARD | Deepest contrarian opportunity in the screen

Deploy T1 ONLY IF April 29 Q1 earnings show:

• Q1 MLR recovering from Q4's 88.3% toward 83-84%
• Management maintains or raises $9.00 EPS guide
• Any positive Stars trajectory signal

Do NOT deploy if: MLR stays above 87%, guidance cut, or CMS final rate rule materially worse than proposal.

Insider signal is the strongest bullish datapoint: Director Hilzinger bought $2.78M at $170.66 (near 52w low), CenterWell President Shetty bought $2.16M at $185. These are independent officers spending real money per shr-002 — the strongest insider signal in the entire screen.

4. TROW — T. Rowe Price | WATCHLIST — Contrarian Income #

Conviction: MEDIUM | Ultra-contrarian — 0 analyst buys, 8 sells, 5.8% yield

Why not buy now: Zero analyst buys, 13% short interest with 12.4 DTC, secular passive flow headwinds. Per shr-034, no RED flags have flipped. The dividend is safe (55% payout) but the business is in structural decline. Better entry in a market correction when beta 1.5 amplifies the drawdown.

When to deploy: If price drops to $78-82 (near 52w low) on market weakness, the 6.3%+ dividend yield provides a floor and the break-even growth rate drops to <1%.

5. DD — DuPont | WATCH ONLY — Not a Graham Play #

Conviction: MEDIUM | Best moat (FilmTec) but not cheap at 17.9x forward P/E

Included for moat quality (narrow-to-wide) but the screen data error inflated its apparent margin of safety. At $45.48 with forward EPS $2.54, break-even growth is 5% — fair value territory, not deep value. Monitor for a pullback below $40 where the risk/reward improves.

Candidates Excluded #

Sector Distribution (Final Buy List) #

Geographic balance: All US (S&P 500 universe only). EU candidates (STOXX 600) screened separately — existing positions AGN.AS and RI.PA cover EU exposure. No action needed on geographic diversification.

Position Sizing Framework #

Assuming EUR 500 available satellite capital (next deployment tranche from monthly contributions):

Rationale: AIG is the only shr-034-passing candidate. Concentrate capital on highest conviction. PHM is contingent on April 23 data. HUM and TROW are next-cycle candidates.

Upcoming Catalysts Calendar #

Prediction Tracking #

7 predictions recorded in output/pipeline/predictions/predictions.json for Phase 5 feedback loop. Will be resolved after 90 days (target: 2026-07-03).

Pipeline Metadata #

Self-Critique #

What worked well:

• Deep analysis correctly identified DD screen data error (corrupt GAAP EPS from E&I spin-off). Without Phase 2, DD would have appeared as the #1 candidate with "70% margin of safety" — a completely wrong signal.
• shr-034 filtering correctly narrowed 9 candidates to 1 actionable buy. This prevents premature deployment into declining earnings situations (DHI, LEN) or unresolved crises (HUM).
• Insider signal analysis added genuine alpha: AIG's CFO buying at $86 (above current) and HUM's director buying $2.78M near the low are high-information signals invisible to the quantitative screen.

What needs improvement:

• Magic Formula not implemented — all candidates scored 50% (placeholder). This removes 20% of composite weight from differentiation. Priority for next cycle.
• Reddit overlay provided zero signal — value stocks simply aren't discussed on Reddit. The divergence framework is designed for high-sentiment environments; at this market level with these candidates, it's dead weight. Consider adding StockTwits or Seeking Alpha as supplementary sentiment sources.
• EU universe (STOXX 600) not screened this cycle — the full multi-framework screen only ran on S&P 500. Need to add STOXX 600 to the composite screen for geographic balance.
• LLM composite scoring is ad-hoc — the deep analysis produces rich qualitative data but the scoring is manually constructed. Should formalize the scoring rubric for consistency across cycles.

Generated by the bi-monthly investment pipeline. Next phase: Phase 5 (Feedback) available after 30+ days of prediction data.