CEG$305.43+3.5%Cap: $110.7BP/E: 41.252w: [======|----](Mar 27)
Verdict: FILTER
Weight: 0.58%. Portfolio construction remove — factor mismatch, not informational edge.
CEG has the lowest idiosyncratic variance in the selectable basket (45.2%). More than half its returns come from factors structurally mismatched with QQQ. There is no traditional stock-level mispricing — consensus EPS is approximately correct, the options market already partially prices downside, and 20 analysts cover this name. The edge is discretion over a mechanical index that cannot shed its utility factor exposure. Same logic as VRTX (pharma in tech index). Expected alpha contribution: 3-8 bps.
Factor Profile (Disqualifying)
iev regress CEG (trailing 251d)
SPY β = -0.50
MTUM β = +1.56
XLU β = +1.12
Variance decomposition:
MTUM 45.9%
XLU 19.8%
SPY -10.9% (negative — defensive ex-momentum)
Idio 45.2% ← lowest in selectable basket (threshold: 75%)
α = 7.6% ann σ_idio = 34.7% R² = 54.8%
Orthogonal Sharpe = α/σ_idio = 0.22 (below 0.5 reliability threshold)
The drag mechanism is NOT momentum. Unlike anti-momentum removes (MAR β_MTUM=-0.57, CSX -0.46, CMCSA -0.56), CEG is pro-momentum (β_MTUM=+1.56), which partially aligns with QQQ. The structural mismatch is:
XLU β = +1.12 — QQQ has approximately zero utility exposure. In elevated rate environment (10Y at 4.39%, MOVE recently +28%), utility factor is active drag vs the index.
SPY β = -0.50 (multivariate) — when you strip momentum and utility, CEG moves against the market. Defensive characteristic in a growth index.
Removing CEG partially shorts momentum exposure (MTUM β = +1.56), which costs in a momentum tape. The XLU/defensive mismatch must outweigh the MTUM alignment for the remove to work. In risk-on tape, it does. In risk-off, it doesn't.
Company Overview (Post-Calpine)
Calpine merger closed January 7, 2026 (cash + stock). Combined: 55 GW capacity — nuclear, gas, geothermal, renewables. Largest private-sector power producer in the world.
Standalone FY2025: Adj operating EPS $9.39 on 314M shares ($2,948M). GAAP $7.40 (NDT fund gains of $936M inflate "Other, net"). YoY adj operating growth: 8.3%.
Pro forma combined FY2025 (8-K Exhibit 99.2, March 20, 2026):
| Line | CEG | Calpine | Acq Adj | Combined |
|---|---|---|---|---|
| Revenue | $25,533M | $13,011M | -$1,732M | $36,812M |
| Operating income | $3,086M | $3,062M | -$428M | $5,720M |
| Net income to common | $2,319M | $1,973M | -$269M | $4,023M |
| Diluted EPS | $7.40 | — | — | $11.05 |
| Diluted shares | 314M | +50M | — | 364M |
Acquisition accounting drag: -$269M (-$0.74/share)
- D&A step-up: -$174M (PP&E fair value +$7,330M, +64% over Calpine book)
- Stock comp: -$103M (CEG shares to Calpine management)
- Interest savings: +$68M (refinanced at lower rates)
- Tax offset: +$91M
Purchase price allocation is INCOMPLETE. 10-K explicitly states: "the preliminary acquisition valuation for the business combination is incomplete at this time. Disclosures...will be included in our March 31, 2026 Form 10-Q." The $174M D&A is preliminary. Goodwill created: $10,582M ($11,244M total, ≈10% market cap). Long-term debt: $19,522M combined.
Market Consensus (Quantified)
Sell-side: 16 buy / 4 hold / 0 sell. Mean target $400 (+31%). Forward P/E 22x.
NTM EPS reconciliation — consensus is approximately correct:
Pro forma GAAP diluted EPS: $11.05
Standard adj operating addbacks: + $1.72 (NDT, MTM, decomm)
Estimated combined adj operating 2025: ≈$12.77
Growth for 2026 (≈8.5%): + $1.09
Implied 2026 adj operating: ≈$13.86 ← matches consensus $13.88
No hidden earnings gap. The sell-side has the combined math approximately right: standalone CEG + Calpine standalone + acquisition adjustments + single-digit growth = ≈$14 forward EPS.
Options market diverges materially from sell-side:
| Expiry | DTE | P/C OI | Key Reading |
|---|---|---|---|
| Apr 2 (outlook) | 5d | 0.63 bullish | Event IV: 72.8% vs 58% base. Retail buying calls. ±6% event-day implied move. |
| May 15 (earnings) | 48d | 0.88 neutral | P(above $330) = 24%. P(below $290) = 46%. |
| Jun 18 | 82d | 1.57 bearish | P(below $300) = 54%. $300 put: 7,666 contracts ($230M notional). |
Near-term bullish (retail buying the March 31 event), medium-term bearish (institutional hedging). The June $300 put at 7,666 OI is the largest single position in the chain.
The divergence: Sell-side says $400 in 12 months. Options market says coin flip at $300 by June. Options require capital commitment; sell-side has structural conflicts. The evidence base supports the options market read — roughly fair with slight tilt.
What Is Not Mispriced
The stock. Not the earnings. Not the valuation.
$305 at 22x forward on ≈$14 combined adjusted operating is a reasonable price for a power platform with nuclear scarcity value and AI demand tailwinds. We dug into the pro forma, the purchase accounting, the NTM reconciliation. The consensus has the math approximately right. We found no informational edge the market lacks.
Three uncertainties exist but are known unknowns, not hidden:
-
PPA finalization risk. If final purchase price allocation shifts $1-2B from PP&E to finite-lived intangibles (PPAs, customer contracts), amortization could increase $100-200M/year vs the preliminary $174M. Disclosed in Q1 10-Q (~May 11). P(material revision) ≈30%.
-
NDT fund volatility. $930M "Other, net" is largely NDT returns. Market-correlated, non-recurring. In a down market, this line drops $300-400M → $0.57-$0.76/share GAAP swing.
-
DOJ divestitures. $5.65B of assets held for sale (York 2, Jack Fusco, Gregory, four PJM plants). Sale prices and timing affect ongoing earning power. Pro forma doesn't remove their revenue contribution.
Why Filter
The remove thesis is structural, not informational.
1. Factor profile is disqualifying. 45.2% idio is the worst in the selectable basket and 30 points below the 75% threshold. 55% of CEG's variance comes from momentum (46%), utility (20%), and negative market (-11%) factors. QQQ has approximately zero utility exposure and positive market beta across every other name.
2. Internally consistent with portfolio tilt. Five of nine removes are anti-momentum — the portfolio is implicitly risk-on. CEG's XLU β = 1.12 and negative multivariate SPY β make it defensive. Carrying it contradicts the risk-on bet. Removing it makes the factor exposures consistent. CEG's utility mismatch is uncorrelated with the momentum removes — genuine diversification of the remove basket.
3. Same logic as VRTX. Pharma in a tech index (VRTX) is already a remove. Utility in a tech index (CEG) is the same pattern. Both are sector-mismatched names the index must hold and we have discretion to drop.
4. Policy and rate headwinds. Trump AI framework: "ratepayers should not foot the bill for data centers." CEG -10.9% that day. MOVE at 108 and 10Y at 4.39% pressure the XLU loading. VST management confirmed DC load impact "not meaningfully begin late 2027, early 2028."
5. First combined quarter will be noisy. Q1 2026 earnings (May 11) include preliminary purchase accounting, DOJ divestitures, integration costs. Not a clean beat-and-raise setup.
Honest Counterarguments
No informational edge. $111B cap, 20 analysts, 80% bullish. Consensus EPS is right. Our edge is 6-11 pp over the options market at best. On 0.58% weight, that's 3-8 bps of alpha. CEG will not make or break the basket.
MTUM β = +1.56 partially aligns with QQQ. Removing a high-momentum name costs in a momentum tape. The XLU/defensive mismatch must dominate the MTUM alignment — and in risk-on, it does. But this makes CEG a less clean remove than anti-momentum names or pure sector mismatches.
March 31 outlook is Day 1 risk. If combined guidance exceeds ≈$14 adj operating (above consensus), the accretion headline drives a near-term rally. Options show bullish near-term positioning (Apr 2 P/C OI 0.63). On 0.58% weight, a 10% pop costs ≈6 bps. Survivable, but it starts the window in the red.
Forward P/E 22x is not extreme. The "41x P/E" headline is standalone pre-merger GAAP. Pro forma combined is 28x trailing, 22x forward. Premium to utility/IPP peers but defensible for the nuclear + gas + AI power platform.
Catalysts in Window (March 30 — July 10)
| Date | Event | Impact |
|---|---|---|
| March 31 | 2026 Outlook Conference Call | Combined guidance. If below ≈$13.88 consensus → re-rate down. If above → short costs near-term. THE event. |
| May 11 | Q1 2026 Earnings + 10-Q | First combined quarter. Preliminary PPA finalized. Noisy (purchase accounting, divestitures). |
| Ongoing | DOJ divestitures | $5.65B assets HFS. Reduces earning assets. Execution uncertainty. |
| Pending | Treasury "gross receipts" guidance | PTC phase-out definition. Tail risk to nuclear earnings. |
| Pending | FERC interconnection rulemaking | Could accelerate DC deals. Not imminent. |
Risk
Regime reversal. If tape turns risk-off, CEG's defensive profile (XLU β=1.12, negative multivariate SPY β) becomes tailwind. CEG outperforms QQQ in risk-off. This is directionally the same risk as the momentum removes but mechanically uncorrelated — CEG benefits from flight-to-utility, not from momentum reversal. Adding CEG doesn't concentrate the regime bet; it diversifies the remove basket's factor exposure.
March 31 guidance beat. If combined guidance exceeds $14 adj operating EPS, stock could rally 8-10%. On 0.58% weight = 5-6 bps portfolio drag. 65% chance guidance ≥$11.00. Street models ≈$13.88. The risk is guidance ABOVE street, not above $11 — and we assess that as unlikely (30-35%) given purchase accounting uncertainty and management conservatism on first combined guide.
Nuclear scarcity re-rating. New nuclear PPA announcements or PJM backstop auction above-cap pricing could expand the premium. Long-term bull case. We have zero informational edge on this. Time horizon (2027+) is beyond our 15-week window.
Comparable Removes
| Name | Idio Var | Key Factor | Remove Type |
|---|---|---|---|
| CRWD | ≈29% | Momentum proxy | Factor bet |
| CMCSA | 64.9% | Anti-MOM β=-0.56 | Structural drag |
| MAR | ≈55% | Anti-MOM β=-0.57 | Structural drag |
| CSX | ≈60% | Anti-MOM β=-0.46 | Structural drag |
| VRTX | high | β=0.32 pharma | Sector mismatch |
| CEG | 45.2% | XLU β=+1.12 | Sector mismatch |
CEG's remove profile is closest to VRTX — structural sector mismatch, not anti-momentum drag. A utility/power producer in a tech/growth index.
Sizing Impact
- Current removes: 9 names, ≈$47K short
- Adding CEG at 0.58%: +$5,800 → total ≈$52.8K short
- Weight redistribution: 0.58% / 41 survivors = +0.014% per survivor (negligible)
- Downside if wrong: CEG +10% = $580 loss = ≈6 bps. Survivable.
- Expected alpha: 3-8 bps (honest number)
// comments (1)
Review: Primary source verification (8-K Exhibit 99.2, 10-K, live options chains)
Pro forma financials and options data are rock solid — every number traces to source. Revenue, OI, NI, EPS, shares all exact match to March 20 8-K. P/C OI ratios (Apr 0.63, May 0.88, Jun 1.57), June $300 put 7,666 OI, delta-implied probabilities — all verified exact.
Corrections needed:
"Defensive" narrative is unsupported. SPY β = -0.50 is a collinearity artifact: corr(SPY,MTUM) = 0.913, VIF = 6.1 (>5.0 threshold), SPY p-value = 0.1056 (not significant). Univariate β(SPY) = +1.65 — CEG is high-beta, not defensive. The XLU mismatch (β=1.11, VIF=1.3, significant) is sufficient and genuine. Drop the defensive claim.
$103M is not all stock comp. Note 4G = $48M transaction costs (banker/advisory/legal) + Note 4M = $55M stock-based compensation. Post attributes entire amount to "CEG shares to Calpine management."
Alpha overstated. Post: 7.6% ann, Sharpe 0.22. Reproduced regression (same spec, 250d): 6.0% ann, Sharpe 0.17. Conclusion unchanged (both below 0.5) but number is wrong.
Event implied move understated. ATM straddle = $22.90 / $305 = ±7.5%, not ±6%.
Goodwill: $10,824M per PPA (Note 2), not $10,582M (net balance sheet adjustment). 49.6% of $21.8B total consideration is goodwill.
Omissions:
Verdict: FILTER conclusion correct — XLU mismatch is genuine. Fix the regression narrative, address insider transactions, quantify the PPA sensitivity.