Step D — Five-model valuation ?
Five conservative, fully-inspectable models per name, screened across the three most-beaten sectors (Energy, Real Estate, Technology; US c-corps and equity REITs, yield 4-12%, cap over $1B; mortgage REITs, MLPs, royalty trusts and preferred series excluded by design). Provider black-box DCFs are reference-only and never enter the average - this run they claimed VICI was worth $134.02, GLPI $136.14, LPG $184.84, OKE $237.33 and UPBD $139.53, and produced a negative levered DCF for SAFE. Outliers (over 2x or under 0.5x the valid median, or non-positive) are excluded. The decisive result this week is VICI: the price fell to $25.42, the analyst consensus eased to $29.80, and the discount is 25.9% on the conservative five-year growth basis - 26.7% on the seven-year basis and 27.8% on the engine's capped reading. All three clear 25%, and the published figure is the lowest of them. Two new names that looked cheaper than VICI on the screen failed inspection: FIS (P/E 6.4) earned 70% of its trailing twelve months in a single quarter, and SAFE (P/E 9.3) has a ten-year series that belongs to two different companies. SPG's P/E of 14.3 is a one-off $3.0bn quarter; on clean earnings it is 38% above fair value.
| Name | EPS-DCF | Dividend | Graham № | Hist. P/E | Analyst tgt | Average | Discount |
|---|
| VICI | $36.08 | $31.20 | $39.37 | $35.03 | $29.80 | $34.30 | 25.9% |
| GLPI | $45.21 | $38.62 | $38.73 | $66.04 | $49.60 | $47.64 | 12.0% |
| UPBD | $13.36 | $15.60 | $21.52 | $20.56 | $28.00 | $19.81 | 4.2% |
| LPG | $63.50 | $39.50 | $70.33 | $53.41 | $55.00 | $56.35 | 2.0% |
| OKE | $92.85 | $53.73 | $69.08 | $111.96 | $97.78 | $85.08 | -12.2% |
| SPG | $95.12 | $152.53 | $47.95 | $134.54 | $223.60 | $151.45 | -38.3% |
| SAFE | n/a | n/a | n/a | n/a | $16.00 | — | — |
| FIS | n/a | n/a | n/a | n/a | $49.00 | — | — |
VICI: IN THE ZONE - FIRST PICK OF THE PROCESS, AND RE-CHECKED FROM SCRATCH TODAY. All five models valid and clustered between $29.80 and $39.37, none flagged. The engine's 27.8% was again rejected because its 89.2% ten-year growth rate comes from a 2016 pre-IPO base of $0.0084; scored at the real five-year growth of 8.2% the discount is 25.9%, and at the seven-year 8.95% it is 26.7% - every defensible basis clears 25%, and the lowest is the one published. Nothing in the fair value rose: the only input that moved was the analyst consensus, down from $30.22 to $29.80, and the price fell 1.7% to $25.42 against a 52-week low of $25.34. Trailing EPS of $2.537 is below the four-quarter sum of $2.58 (0.48, 0.82, 0.57, 0.71 - no outlier). Free cash flow positive and rising in all five years from $894m to $2,509m, covering the dividend every year; interest coverage 4.33x; long-term debt 0.60x equity and falling; net margin 69%. Ten filings in the quarter with no equity shelf or takedown - the $1.75bn of new 5.40% and 5.75% notes refinance 2026 maturities. The dividend was raised for a ninth consecutive year this week, to $0.46 quarterly. The margin is thin: 0.9 points over the threshold and $0.30 under the rung.
GLPI: THE QUALITY ANCHOR, DISCOUNT 12.0%. EPS up every single year from 2.27 to 2.95, net margin flat at 51-52%, ROE the highest of the group at 17.8%, free cash flow positive in all five years, leverage down from 2.06x to 1.61x equity, no dilution and the quietest filing record of the group. The dividend model is again scored at the real post-recovery growth of 1.68% rather than the engine's 15.5% measured off the 2020 cut, which moves fair value from $50.87 to $47.64. The May 2026 raise from $0.78 to $0.82 quarterly is not yet credited. Price fell 1.5% on the week; the buy rung is $35.73.
UPBD: THE ONLY TECHNOLOGY SURVIVOR, AND A THIN ONE. Quarterly EPS is clean (0.38, 0.63, 0.35, 0.23) and the 8.2% yield is real, but the payout is 98% of trailing GAAP earnings, the quarterly dividend has been frozen at $0.39 for seven payments since December 2024, and free cash flow in FY2024 ($48m) did not cover the $82m paid out. The engine's 1.2% five-year dividend growth is a calendar artifact - five payments landed in 2020 - and on the quarterly rate the real figure is 6.1%, which under the rule would give a $27.04 dividend model and a 14.1% discount. It is not credited: a stalled dividend gets a zero-growth model ($15.60), fair value $19.81, discount 4.2%. Fiscal-year EPS runs 2.37, 0.23, -0.09, 2.26, 1.30 - noise, not a trend - and long-term debt is 2.26x equity with interest covered 1.98x. Carried because the sector assignment is the provider's and nothing better survived in it.
LPG: CYCLICAL AT PEAK EARNINGS, NOW AT ITS 52-WEEK HIGH - BUY RUNG UNRELIABLE. TTM EPS of $7.56 against a five-year fiscal-year average of $4.08 (1.79, 4.31, 7.63, 2.14, 4.55), with the June 2026 quarter alone earning $3.25. The stock rose 10.9% on the week to $55.20, its 52-week high and 2.3x its low. Scored on normalised earnings the average is $41.84 and the discount is minus 31.9%, so the honest reading is a range from plus 2.0% to minus 31.9%. The balance sheet remains the cleanest here - 0.40x long-term debt to equity, coverage 7.09x, positive free cash flow in all five years - but the dividend is cycle-dependent and takes 97% of normalised earnings. Only the STRONG and EXTREME rungs carry real margin of safety.
OKE: THE BEST BUSINESS IN THE SWEEP, 12% ABOVE FAIR VALUE. Quarterly EPS reconciles exactly (1.53, 1.23, 1.55, 1.50 against a reported TTM of 5.80), fiscal-year EPS has been 5.19-5.49 for three years, free cash flow has been $1.7-2.9bn positive every year, and the dividend was raised to $1.07 quarterly in February. It trades 12.2% above a conservative $85.08. Two cautions: the share count has grown from 446m to 630m on acquisitions and the company filed two automatic shelves and a takedown this quarter; and 2025's $2.45bn of free cash flow no longer covered $2.58bn of dividends after capex rose to $3.15bn. On the ladder to fill the Utilities vacancy - the rungs are far below the price.
SPG: FLAGGED - THE P/E OF 14.3 IS ONE QUARTER. Q4-2025 EPS of $9.35 ($3,049m of net income) against $415-668m in every other quarter; trailing EPS of $14.65 is double any prior fiscal year. Rescored on the H1-2026 run rate of $5.94 the fair value is $151.45 and the stock trades 38% above it. Engine headline of 11.1% discarded. Nothing wrong with the company's filings (no shelf, no dilution) - it is simply not cheap.
SAFE: DATA-DISTORTED - NO AVERAGE, NOT ON THE WATCHLIST. The most interesting new name on the screen: a ground-lease REIT at a real trailing P/E of 9.3 with six consecutive clean quarters of $0.39-0.42. But the ten-year EPS series reads 0.36, 9.75, 0.64, 28.17, 1.17, 1.32, 2.17, -0.82, 1.48, 1.60 - it belongs to a predecessor entity and a 2023 reverse merger, so the engine's 18% ten-year growth rate, its 10.5x own median P/E and its minus 18.9% dividend growth all describe a company that no longer exists. Three of five models corrupted; the engine's 23.2% discount is meaningless and the provider's own levered DCF comes out negative. Re-examine when the post-merger entity has five clean fiscal years.
FIS: DATA-DISTORTED - NO AVERAGE, NOT ON THE WATCHLIST. The P/E of 6.4 and the engine's 26.7% discount rest on a single quarter: Q1-2026 EPS of $4.59 ($2,366m) against $231-511m in the surrounding quarters and a loss in Q2-2025. The ten-year series contains impairment years of -27.68 and -11.26, so there is no usable growth rate and no own median P/E. Rescored on FY2024's clean $1.42 the fair value is $34.73 and the stock trades 21% above it. The same failure mode as EIX two weeks ago, in a different sector.
Model assumptions (fully inspectable): EPS-DCF: 10y at historical EPS growth (capped 0-10%) + 10y terminal at 4%, discounted at 11%. Dividend model: DPS x (1+g)/(10% - g), g = min(5y dividend CAGR, 4%), measured on the quarterly rate rather than calendar-year sums, and set to zero where the rate has been frozen for more than a year. Graham Number: sqrt(22.5 x EPS x BVPS). Historical-PE: EPS x own 10-year median P/E. Analyst: consensus price target. Cyclical rule: where TTM EPS sits at a cycle peak, the discount is shown as a range (peak and 5-year normalised EPS) and the BUY rung is declared unreliable. One-off rule: where a single quarter's below-the-line item dominates the TTM, the name is rescored on the clean run rate and the headline discount discarded. Merger rule: a definitive cash merger agreement caps the market price and voids the model. Listing-history rule: before using the engine's ten-year EPS CAGR, confirm the oldest years in the series are real operating years of the same entity - an IPO stub year or a reverse merger produces a growth rate the DCF silently caps instead of rejecting. REIT caveat: EPS-based models understate REIT cash earnings; treated as extra margin of safety, not corrected upward.