GS vs. EVR: Which Investment Bank Stock Offers Greater Upside?

Goldman Sachs' investment-banking fees rose 52% year over year to $6.2 billion in 1H26, and its backlog reached a five-year high as global M&A value hit a record $2.8 trillion. The article favors GS over Evercore, citing its leadership in advisory, underwriting and leveraged lending, a more diversified business mix, a 2.1% dividend yield, and upward 2026-27 estimate revisions. Evercore trades more cheaply at 11.5x forward P/E versus GS at 13.1x and has stronger projected 2027 earnings growth, but its 97.6% revenue reliance on investment banking and equities raises sensitivity to any dealmaking slowdown.

Analysis

The relevant distinction is not advisory exposure versus diversification in isolation; it is incremental operating leverage at this point in the cycle. GS can convert a sustained reopening in equity issuance, leveraged finance and sponsor activity into earnings across banking, trading and financing simultaneously, but this also means its already-positive estimate revisions leave less room for a simple backlog-driven surprise. The more durable rerating catalyst is whether recent alternative-asset and ETF acquisitions demonstrate positive fee-related earnings growth without meaningful integration costs or balance-sheet deployment.

EVR is the higher-beta expression of a broadening M&A cycle: its fixed senior-compensation base means advisory revenue growth should produce disproportionate margin expansion once fee conversion accelerates. Its weak relative performance and lower multiple create a cleaner 6-18 month catch-up case if announced mega-deals close and the 2027 fee pipeline holds; however, its expanded senior headcount raises the revenue-per-MD hurdle and makes a soft advisory quarter materially more damaging than for GS. Lazard (LAZ) and Moelis (MC) are useful read-throughs: broad independent-advisory strength would validate sector demand, while EVR-specific underperformance versus both would point to execution or compensation pressure.

Consensus likely overweights the apparent safety of GS and underweights the possibility that its diversified model dilutes upside if capital-markets activity merely normalizes rather than accelerates. Conversely, EVR's discount is not automatically mispricing: advisory backlogs have long conversion periods and are vulnerable to financing volatility, antitrust delay, and board-level risk aversion. Over the next 1-3 months, underwriting league tables, completed-deal fees and compensation guidance matter more than announced deal value; over 6-18 months, lower rates and a stable regulatory regime would favor EVR's earnings torque.

Key Decisions for Investors

• Initiate a 6-12 month long EVR / short GS pair at equal dollar exposure only after EVR shows relative stabilization versus GS and LAZ; target 15-20% EVR relative outperformance from valuation mean reversion and advisory operating leverage. Exit if EVR guides compensation ratio higher, advisory backlog declines sequentially, or the pair breaks 10% against entry.
• Maintain GS as the preferred absolute long into the next earnings report, but size modestly after estimate revisions; use a 5-8% post-results upside target versus a 4-5% stop. Add only if banking fees and equities trading beat while AWM margins or net inflows confirm that acquisitions are accretive rather than simply AUM additive.
• For a more diversified deal-cycle basket, long EVR and LAZ versus short KRE over 6 months: independents benefit from strategic advisory fees while regional banks remain more exposed to funding costs and credit normalization. Do not initiate if rate volatility spikes or high-yield spreads widen materially, as that would impair deal financing and erase the differentiation.
• Set an event alert around completed, rather than announced, large-cap M&A and quarterly revenue-per-senior-MD at EVR. A sustained decline in high-yield issuance or a meaningful widening in credit spreads is the key macro falsifier for both longs and should prompt reduction before backlog conversion disappoints.