PRG is the unusual case where we are more bearish than Stephens at spot ($35.88) despite finding the consensus bull driver — Four BNPL inflection — more credible than the consensus does. The reason is a footnote in the same Q1 2026 10-Q that nearly doubled the GAAP amortization drag from the Purchasing Power acquisition: $14M/yr of GAAP earnings drag that sell-side models built off the original 10-K mostly have not rebuilt. Bull tail fatter, base case shifted lower, mean similar; entry at $35.88 doesn't reward us for either insight. Pass at $36. Watchlist trigger at $31-32 (50DMA + max pain confluence) where the asymmetry restores.

Two non-obvious findings

Four BNPL is idio share capture, not sector tailwind. Q1 2026 EBT $11.4M (+478% YoY), revenue +142%, EBT margin 32.6% vs 13.7% prior year. SG&A as % of revenue fell 1,020 bps in a year. The cross-ticker test is unforgiving: UPBD/Acima Q1 cal 2026 — the cleanest comparable LTO/BNPL peer — printed +1.8% revenue, -5.9% GMV, +30bps EBITDA margin. AFRM Dec 2025 quarter +31% GMV (still strong, nowhere near Four). Four is gaining share at the LTO/BNPL boundary from both directions: from Acima on the LTO side AND from AFRM/Klarna on the BNPL side. The dual-vector growth only shows up when peer prints are aligned by calendar quarter rather than fiscal. Most desks don't run that comparison on a $1B-cap.

PPA amortization correction is footnote-buried. The original 10-K disclosed preliminary 20-year client-relationship lives implying ≈$17M/yr amortization. Q1 10-Q refined to 10-year lives — quarterly D&A of $8.085M annualizes to ≈$31.25M/yr. That's $14M/yr of additional GAAP drag (≈$0.25/share after-tax on 40M shares) not in models built off the prior filing. Q2 will force the rebuild because $8M/quarter cannot reconcile to a $17M/yr model assumption. At 8-9x P/E, ≈$2.00-$2.25 of fair-value pressure from this single line revision.

Scenario math

FY2027 GAAP EPS by segment build-up. Probabilities incorporate cross-ticker confirmation (Four idio) and the new PPA drag.

EV @ $35.88: +5.7% over 18-24mo (≈3.3% annualized; Sharpe ≈0.08 at σ_idio 43% — below threshold). EV @ $31.00: +22% (≈12% annualized; Sharpe ≈0.28 — investable). EV @ $28.00: +34% (≈18% annualized; Sharpe ≈0.42 — attractive).

The cliff is real because at $36 our base case (-10pp vs market-implied) and bull tail (+13pp vs market-implied) cancel on EV. At $31, both insights compound into asymmetric upside.

Stephens at $48 PT is roughly our bull case (9-10x of $5-5.50 EPS). They got Four right and the 10-Q dropped same day as their initiation — the PPA correction is unlikely in their model. Their target survives if Four sustains 30%+ EBT margin AND PPA stays at $17M/yr. The first is plausible. The second will not survive Q2.

Falsifiers with magnitudes

1. Q2 60+DPD ≥ 13% — vintage thesis dead. Bull P 35% → 15%, fair value $28 (-22% from spot). pred-mblyv6 (60% Four 60+DPD ≤8% Q3 2026) breaks first.
2. AFRM mid-May 60+DPD ≥ 3% — peer-level cross-ticker invalidation. Bull P 35% → 22%, fair value $32. pred-zrj28f (75%) breaks first.
3. Four EBT margin Q2 < 20% — operating leverage broken. Bull P 35% → 18%, fair value $30. The driver itself reverses.

The PL "stable credit" framing is weaker than the headline. Q1 2026 PL EBT +6.9% YoY includes a $6.5M one-time gain on charged-off receivable sale. Strip it and adjusted PL EBT is $45.5M vs $48.6M prior year — -6.4% YoY. The base case scenario already incorporates this; the headline does not.

Catalysts

LR derivation

Pre-filing prior: cumulative LR 0.32 (P_bull ≈ 24%). Filing evidence — collapsing correlated bullish into the strongest single signal (max LR 1.8, Four cross-ticker confirmed) and correlated bearish into the strongest single signal (max LR 0.6, PPA correction):

Net LR ≈ 1.8 × 0.6 = 1.08
Posterior odds: 0.32 × 1.08 ≈ 0.35
P_bull: ≈24% → ≈26%

Two findings cancel mostly on direction. The value is in path shape, not the bet. Memo LR: 1.1.

Evidence