The Whole Buy-Side vs Sell-Side Map: Who Sits Where and What Each Side Actually Does

Goldman Sachs booked record Global Banking & Markets net revenues of $15.52bn in the second quarter of 2026, a 53% increase on the same quarter a year earlier. Almost none of that revenue came from the firm being right about market direction.

In the same six-month window, Point72 returned 14.5% year to date, Citadel’s tactical trading fund returned 14.3%, and Millennium returned 10.5% on roughly $89bn of assets. Almost all of that came from being right about market direction.

Two sets of institutions, operating in identical instruments, clearing through the same venues, recruiting from the same doctoral programmes, running on entirely opposed revenue physics. That divergence is the subject of this piece.

Most people are taught the buy-side and sell-side split as a vocabulary exercise. Banks sell securities and funds buy them. The definition survives about four minutes of contact with a live desk.

For daily Quant articles and market breakdowns:

It fails because Citadel Securities and Citadel share a founder while sitting on opposite sides of the line. It fails because prime brokerage is a sell-side business whose entire purpose is to lever buy-side balance sheets. It fails because a bank’s central risk book warehouses positions all day inside a regulatory perimeter that formally prohibits proprietary trading.

What follows is the map that holds up under pressure, the seat-by-seat breakdown of what each side actually does, and the compensation architecture that falls out of the structure.

The correct question is whose capital absorbs the loss, and who the client is.

Sell-side firms intermediate. They take the other side of a client’s trade, hold the resulting exposure for the shortest period they can manage, hedge whatever they cannot immediately offload, and collect the spread alongside the fee. Inventory is a cost of doing business rather than the source of return.

Sell-side revenue therefore scales with client activity and with volatility. It scales in a rising market and in a falling one, because the bank is being paid for capacity and access rather than for a view.

Buy-side firms allocate. They deploy capital belonging to external investors or to the firm’s own principals, they hold positions because they expect those positions to pay, and they earn management fees on the assets alongside performance fees on the outcome. Inventory is the entire business.

If you are mapping which of these seats your background actually qualifies for, QuantFinanceWiki.com breaks the route down by role requirement, interview format, and realistic entry point across firm type.

Every structural difference between the two sides descends from that single distinction. Compensation design, risk limits, hiring bar, technology budget, the shape of a twenty-year career, and what happens to you personally in a bad year all follow mechanically from whose money is at risk.

Goldman’s second quarter shows the sell-side version cleanly. Combined FICC and equities financing revenues reached $4.5bn, up 62%, and accounted for 37% of total trading revenues.

That is money earned by lending balance sheet and supplying execution capacity to other people’s positions. Assets under supervision hit a record $4.04trn after $91bn of long-term net inflows during the quarter, producing fees on flow, fees on custody, and fees on leverage.

A bulge-bracket markets division separates into functions that outsiders routinely collapse into a single word. Each of them has a distinct revenue mechanism, a distinct risk profile, and a distinct compensation ceiling.

Sales owns the client relationship. The desk distributes ideas, axes, and inventory to asset managers, hedge funds, corporate treasuries, and central banks. Compensation runs off a credited revenue share from the accounts covered, which makes the seat closer to institutional relationship management than to trading.

Flow trading makes prices in liquid instruments and manages the book that results. Position sizing is constrained by desk risk limits rather than by conviction, and a trader who wants to hold a view for three weeks generally cannot. Goldman’s equities revenue climbed 72% to a record $7.42bn on the back of derivatives, cash equities, and prime financing activity.

Structuring builds bespoke payoffs for clients who want exposure the listed market does not offer. Revenue is priced at inception, booked as a day-one mark, and then hedged across the life of the trade, which is why structuring desks are staffed with pricing quants rather than with risk-takers.

Strats and quantitative research price derivatives, calibrate models, build execution algorithms, and run valuation adjustment and model validation work. This is the principal sell-side quant seat, and it functions as support to a client-facing revenue line rather than as a revenue line of its own.

Prime brokerage finances, clears, and custodies hedge fund positions. The buy side cannot run meaningful leverage without it, which makes prime the single most informative window any bank has into hedge fund positioning.

Investment banking coverage handles advisory and underwriting. Goldman’s investment banking fees rose 55% to $3.40bn, supported by advisory alongside record debt underwriting, with the backlog reaching its highest level in five years.

Equity research publishes coverage and maintains corporate access. This function deserves separate treatment, because it is the part of the sell side that is contracting while everything around it expands.

Data on analyst movement shows 736 mid-tier analysts exiting the market between 2023 and 2026, leaving a leaner sell-side structure built around a small group of senior analysts supported by more junior teams.

The global equity research market is worth roughly $16.5bn, headcount continues to shrink, and approximately 3,000 listed companies now carry no analyst coverage at all. Analyst counts at the largest banks have fallen by more than 30% over the past decade, with the deepest cuts landing in Europe and Asia and concentrated at entry level.

Inflation-adjusted total compensation for the function has declined by around 30% from pre-2008 levels. For a role that still demands sixty to seventy hours a week, that is a meaningful repricing.

The cause is mechanical rather than cyclical. MiFID II unbundled research from execution commissions, which forced asset managers to write an explicit cheque for something they had previously received as a rebate on trading volume.

Once the cheque became visible on a line item, it got cut. 80% of asset managers running more than $150bn now expect research budgets to rise only marginally over the next two years, and what they do spend is being reallocated toward a narrow set of tenured, differentiated analysts as routine coverage becomes automated.

A sell-side function whose output can be substantially reproduced at marginal cost loses its price. The functions that survived the last decade are the ones requiring balance sheet, a licence, or physical inventory, and none of those can be replicated by a competitor with a GPU cluster and a data subscription.

For anyone weighing a research seat against a strats seat or a pod seat, that asymmetry is the entire decision.

The clean binary fails at exactly one point, and it happens to be the most important point for anyone building a quant career in 2026.

Jane Street reported $39.6bn in net trading revenue for 2025, close to double the $20.5bn it earned in 2024. In the first quarter of 2026 alone the firm produced $16.1bn of trading revenue and $10.3bn of net income, more than doubling the same quarter a year earlier.

A single quarter delivering more than 40% of the prior full year tells you how closely those earnings track realised volatility. The firm accounts for more than 10% of North American equity trading and holds the leading share in exchange-traded funds, with roughly 3,000 employees across five offices.

Citadel Securities closed 2025 with $12.2bn in net trading revenue, a 25% increase on the prior record of $9.7bn set in 2024. The firm then posted a record $4.3bn in the first quarter of 2026, up 28% year on year, converting that into $1.9bn of net income.

Hudson River Trading followed with $6.4bn in the first quarter of 2026, more than half its entire 2025 result. Across the three firms, first-quarter trading revenue reached roughly $27bn.

Structurally these firms belong on the sell side. They provide liquidity, they earn spread, they hold inventory only for as long as hedging requires, and they manage no external client capital.