Skip to content

Brokerage and Sales Credit: What a Trade Actually Costs to Execute

Think of a real estate agent's commission. One agent charges 2% of the sale price. Another, selling an identical flat next door, charges a flat ₹25,000 regardless of price. A third works for a big brokerage that quotes the fee in "basis points of deal value" because the client is a fund buying a hundred flats at once, not a family buying one. All three are paid for doing the same job — getting a transaction done — priced three completely different ways.

Brokerage in capital markets works the same way. There is no one formula. What you're charged depends on which of four pricing models your broker uses, which asset class you're trading, and — once you're inside a bank rather than a retail app — a second, entirely internal number called sales credit that decides who at the bank gets paid for your trade.

What brokerage actually is — and isn't

Brokerage is the fee your broker charges for executing your order. It is not the only line on your contract note. In India, a single trade also carries Securities Transaction Tax (STT), exchange transaction charges, SEBI turnover fees, stamp duty, and GST on the brokerage itself — none of which is brokerage, all of which get lumped into "what I paid to trade" in a client's head. The same split exists in the US: a trade can carry an SEC fee and a FINRA Trading Activity Fee alongside whatever the broker itself charges. Brokerage is specifically the broker's own fee for the service of getting the order filled — everything else is a statutory pass-through the broker collects but doesn't keep.

The four ways brokerage gets calculated

Percentage of trade value. Brokerage = Trade Value × Rate. A full-service Indian broker like ICICI Direct or HDFC Securities has historically charged something like 0.30% on a delivery trade and 0.03% on an intraday one — buy ₹1,00,000 of shares for delivery and the brokerage alone is ₹300. Internationally, a traditional full-service US wealth manager pricing a trade as a percentage of order value is the same model, just a legacy of an era before discount brokers existed at all.

Flat fee per order. A fixed rupee or dollar amount, independent of trade size. Zerodha's ₹20-or-0.03%-whichever-is-lower on intraday and F&O trades is the model that built India's discount-broking boom. Its zero-commission US counterpart is arguably no fee at all — which is its own story, since "free" there is usually funded by payment for order flow instead rather than by the client paying nothing to anyone.

Per-share or per-contract. Brokerage = Quantity × Rate per unit. This is how US institutional equities and listed options are typically priced — Interactive Brokers, for instance, has charged institutional clients around $0.005 per share with a small per-order minimum, and around $0.65 per options contract. Buy 10,000 shares at $0.005/share and the brokerage is $50, regardless of the stock's price. India's equivalent shows up in the derivatives segment: a flat fee "per lot" on F&O orders is the same per-unit logic, just denominated in lots instead of shares.

Basis points (bps) of notional. Brokerage = Notional Value × (bps ÷ 10,000). This is the institutional and investment-banking convention — the unit small enough to price a fee that has to work whether the trade is ₹10 lakh or ₹1,000 crore. An institutional cash-equities desk, Indian or international, pricing a large block trade at 5bps means: on a ₹10 crore order, brokerage is ₹10,00,00,000 × (5 ÷ 10,000) = ₹5,00,000.

The same ₹1 crore trade priced under each model lands nowhere near the same number: 0.30% is ₹30,000; a ₹20 flat fee is ₹20; 5bps is ₹5,000. None of these numbers is "the" brokerage rate — each is correct for the model it belongs to, which is exactly why comparing two brokers' rates without checking which model they quote in is comparing nothing at all.

Most flat-fee brokers actually run a hybrid: brokerage = minimum of (percentage × trade value) and the flat cap. Zerodha's "₹20 or 0.03%, whichever is lower" is that rule stated explicitly — on a small trade, 0.03% might be less than ₹20, so the client pays the smaller number; on a large one, the flat ₹20 wins. It's a small formula, but it's exactly the kind of boundary condition that breaks a poorly tested commission engine: what happens at the exact value where both sides are equal, and does the system round to the client's benefit or the broker's?

How investment banks actually price it, asset class by asset class

This is where "brokerage" stops being one number and starts depending entirely on what's being traded — because not every asset class even carries an explicit commission.

Asset classHow the fee is actually chargedTypical conventionIndian exampleInternational example
Cash equitiesExplicit commission, in bps of notional~3–10 bps on institutional agency tradesAn institutional desk executing an FPI's block order in Reliance shares for a few bpsA US bulge-bracket agency equities desk pricing a large program trade in single-digit bps
Fixed income (bonds)No standalone commission — the dealer trades as principal and the fee is embedded in the bid-ask spreadSpread of roughly 5–25bps, wider for illiquid or long-tenor paperA primary dealer quoting a spread on a Government of India security via NDS-OMA Wall Street dealer quoting a wider spread on a high-yield corporate bond than on a US Treasury
FXSpread (in pips) built into the quoted rate, or an explicit small commission on agency e-FX execution1–3 pips on major pairs; ~0.1–0.5bps agency commission on institutional e-FX flowA corporate treasury getting a USD/INR quote from its relationship bank, spread already inside the rateA US corporate executing EUR/USD on an e-FX platform, paying spread plus an optional bps commission
Listed futures & optionsPer-contract or per-lot fee, plus exchange and clearing fees passed straight through, unmarked upFlat fee per lot/contractAn NSE F&O broker's flat per-lot charge plus SEBI turnover fee and exchange transaction chargesA CME futures broker's per-contract round-turn fee plus exchange and clearing fees
OTC derivatives (interest rate & currency swaps)No brokerage line at all — the bank's margin is baked into the fixed rate it quotesSpread embedded in the quoted rate, not itemizedAn Indian corporate's INR interest rate swap, priced with the bank's margin folded into the fixed legA US corporate's USD interest rate swap, same mechanic, different currency

The pattern underneath that table: wherever a market is exchange-traded and agency-executed (equities, listed derivatives), the broker charges an explicit, itemized fee. Wherever a bank trades as principal against its own book (bonds, FX, OTC swaps), there usually isn't a visible "brokerage" line at all — the bank's profit is inside the price it quotes you, not added on top of it. Knowing which of those two worlds an asset class lives in is the actual answer to "how do investment banks calculate brokerage by asset class" — the honest answer for half the table is that they don't calculate a brokerage fee at all; they calculate a spread instead, and the two aren't the same thing wearing different names.

One asset class isn't one price — in commodities, and everywhere else

"Commodities" in that table hides as much variation as it reveals, and metals make the easiest place to see it: sit inside an international bank's commodities desk for a day and gold, silver, copper, and platinum — four metals on the same desk, sometimes quoted by the same salesperson — get priced on four different logics, because liquidity and market structure differ metal by metal, not desk by desk.

MetalWhere it actually tradesHow the bank prices itWhy it's priced that way
GoldCOMEX (part of CME Group) internationally; MCX in IndiaTightest of the group — roughly 1–2bps institutional, or a few cents-per-ounce spread on OTC bullionThe single most liquid metal traded anywhere — the deepest order book gets the narrowest spread
SilverCOMEX; MCXWider than gold — roughly 3–5bps, or a proportionally wider $/oz spreadLess liquid and more volatile than gold, so the bank needs a bigger cushion to hedge the position it just took on
CopperLondon Metal Exchange (LME)Not bps at all — a flat commission per lot (25 tonnes), a structural holdover from LME's ring-dealing conventionsLME's entire fee architecture predates a common electronic bps standard — the mechanic is historical, not liquidity-driven
Platinum / PalladiumNYMEX; LBMA OTCWidest of the four — easily 8–15bps equivalentThinnest market of the group: demand concentrates in auto-catalyst industrial use, so one large order can move the price by itself

Put a number on it: a bank's commodities sales desk quoting $1,000,000 of notional in each metal, same day, same client tier, might land at $150 on gold (1.5bps), $400 on silver (4bps), and $1,000 on platinum (10bps) — nearly seven times the cost on the identical notional, purely because of which metal it is. Nobody's charging platinum more out of preference; the bank is pricing in the real cost of hedging a position in a market that thin.

Copper is the outlier that proves the point a different way. Because LME still commissions per lot rather than per bps, a $1,000,000 copper trade doesn't translate into a clean bps figure at all — the fee depends on how many 25-tonne lots that notional works out to, not on the dollar value itself. India's version of the same instrument, MCX copper, trades in 2,500kg lots and is usually folded into the same flat per-lot brokerage plan an Indian broker already applies to gold and silver — a simplification international banks can't make, because LME's lot-based mechanic and COMEX's bps-based one are genuinely different systems, not two ways of describing the same fee.

The same liquidity-driven spread shows up outside commodities too — just swap "metal" for "liquidity tier":

Asset classCheapest tier to tradeMost expensive tier to trade
EquitiesA liquid large-cap — a Nifty 50 or S&P 500 constituent — institutional ~3–5bpsAn illiquid small-cap or micro-cap stock — 15–25bps or more
Fixed incomeSovereign debt — Government of India securities, US Treasuries — a few bpsHigh-yield or distressed corporate credit — 50–100bps or more
FXG10 major pairs — EUR/USD, USD/JPY — under a pip, institutionalEmerging-market or exotic pairs — USD/TRY, USD/ZAR — several pips wide
Listed derivativesA liquid index future — Nifty 50, S&P 500 e-mini — a near-flat per-lot feeAn illiquid, far-dated or deep out-of-the-money single-stock option

Every row across both tables is answering the same question with a different label stuck on it: how hard would it be for the bank to get out of this exact position if it had to, right now. That's the real variable behind "asset class," "instrument," and "liquidity tier" — three different names for the same thing.

Sales credit — the number the client never sees

Brokerage is what the client pays. Sales credit is a completely separate, internal number: how the bank splits that revenue between the people who serve the client relationship and the desk that actually executed or warehoused the risk.

Say an institutional client's equity order generates $10,000 in commission. That $10,000 is booked once, as firm revenue. But two different teams get measured on it, so the bank runs an internal "sales credit" split — a pre-agreed grid, negotiated per desk and often per client tier — that might attribute, say, 40% ($4,000) to the salesperson who owns the client relationship and 60% ($6,000) to the trading desk that filled the order. Neither number changes what the client was billed. Both numbers exist purely to feed two different performance scorecards and, eventually, two different bonus pools from the same $10,000.

Sales credit gets genuinely complicated the moment a client trades across multiple products with one coverage salesperson: a relationship manager covering a client's equities, FX, and rates business gets credited across all three, even though three different desks actually executed the trades — because the grid is measuring who owns the relationship, not who pressed the button.

India runs the same underlying split, just under a different name and a simpler structure. Say an HNI client's relationship with a full-service Indian broker generates ₹5,00,000 in brokerage over a quarter. The RM who owns that relationship doesn't keep the brokerage — the firm books all of it as revenue — but a payout grid might credit the RM 15% of what their book generates, roughly ₹75,000, as variable payout, with the rest retained by the firm to cover the dealing desk, research, and back-office infrastructure that actually got the trades filled and settled. Same mechanic as the Wall Street desk's 40/60 split — revenue generated by a relationship, divided between whoever owns the client and whoever runs the machinery behind the trade — it just isn't called "sales credit," because most Indian retail and HNI broking still bundles the sales and execution roles into one RM rather than splitting them into a separate coverage salesperson and a separate trading desk the way a bulge-bracket institutional bank does.

Institutional sales credit also intersects with a mechanic that has no real Indian parallel yet: the commission-sharing arrangement, or CSA. A fund's commission doesn't always go entirely to the broker that filled the order — part of it can be contractually earmarked, at the client's instruction, to pay a completely different firm for investment research, so the executing desk never had the whole commission to begin with, before sales credit even gets carved out of what's left. A step-out trade is the mechanical version of the same idea: one broker executes, but "steps out" part of the commission to a second broker the client designated, usually to settle that same research obligation. The EU's MiFID II rules, effective January 2018, went further and forced funds to pay for research and execution as two separate, unbundled line items rather than one blended commission — a split India's research-and-broking market has never been required to make, since SEBI hasn't mandated the same unbundling.

Why this matters for a Business Analyst

Think of two ledgers that must never touch

A restaurant's cash register and its staff tip-pooling spreadsheet both reference the same bill. The customer only ever sees one number on the receipt. The tip split behind the counter can change entirely — a different formula, a different week, a different staff roster — without the customer's bill moving by a rupee.

Brokerage and sales credit are exactly that pair of ledgers, and a trade-booking or commission-management system has to keep them that way: one client-facing, one strictly internal, computed from the same trade but never allowed to alter what the client was actually billed.

That separation is where the real requirements live. A rate table has to apply the correct formula by instrument type — charging equity bps logic to a bond trade, or vice versa, silently over- or under-bills a client. The "lower of percentage or flat cap" rule needs explicit acceptance criteria for the boundary itself: what happens at the exact trade value where both sides compute equal, and which way does rounding go. And a sales-credit engine has to reconcile back to the same trade ID as the client-facing brokerage ledger without ever feeding forward into it — a system that lets an internal allocation influence what a client is charged has broken the one wall this whole model depends on. India's exchange bye-laws have long capped member brokerage at a ceiling — 2.5% of trade value under NSE and BSE rules — a limit that matters less today only because competition already pushed real rates far below it; the US never had an equivalent cap once the SEC abolished fixed minimum commissions on May 1, 1975 — "May Day" — the single event that made discount brokerage possible there in the first place, decades before it caught on in India.

The reconciliation break that isn't actually a break

This is exactly the check Middle Office runs every day — the same "matched and controlled" clock that verifies price, account, and position limits also verifies that the commission charged on a trade matches what the bank's own rate card says it should be, before that trade is allowed to feed into settlement and, eventually, Post-Settlement's books-and-records. But that recon only works if the engine running it knows, trade by trade, which of the formulas above actually applies — because as every table in this piece shows, there's no single "expected commission" formula that covers a whole asset class, let alone the whole business.

Feed a gold trade through a recon built for bps and it works cleanly: expected 1.5bps, actual 1.6bps, within tolerance, auto-matched. Feed a copper trade through that same bps-only recon and every single one of those trades throws a break — not because anything is actually wrong, but because LME's per-lot commission was never going to resolve into a clean bps figure to begin with. That's a false break: a real analyst's morning spent chasing a difference that doesn't exist, because the system applied equities-shaped logic to an LME instrument that runs on a different mechanic entirely.

A genuine break looks different from that, and it's the one worth catching: a rate card entry that's gone stale, a trade booked under the wrong instrument code — a silver ticket tagged as gold — or a lot-size error on a copper trade that throws off the entire expected-commission calculation. Those real breaks get buried at exactly the rate the false ones pile up, because a badly modeled rate table dumps both into the same exception queue with no way to tell them apart on sight.

Sales credit carries its own separate reconciliation, on a completely different ledger. Where the brokerage recon compares the bank's number against the broker's or client's confirm, the sales-credit recon checks something purely internal: that every trade's allocated sales credit sums back to exactly 100% of the gross commission booked against that trade ID — no more, which would mean two teams are both getting paid off the same revenue, and no less, which would mean some of it has quietly gone missing from both scorecards. Same discipline as the brokerage recon, just run on the other side of the wall this entire model depends on keeping intact.

Lighthouse Insight

Go back to the three real estate agents — the same flat, three unrelated formulas, all technically correct.

A trading commission isn't one number waiting to be looked up. It's a formula chosen from four possible shapes, applied to an asset class that may or may not even show a fee on the surface, then split a second time — invisibly, internally — between whoever owns the client and whoever took the risk. The contract note shows you the first number. The bank's own compensation grid is where the second one lives, and it was never meant for you to see it at all.

Continue the system

A curated path through the next concept, so one essay becomes a map.

Related essays

Capital Markets

Front Office, Middle Office, Back Office: Three Clocks on One Trade

Front Office, Middle Office and Back Office aren't three departments arguing about the same trade — they're three different, equally correct answers to the question "is it done?" Why the split exists, what each clock actually measures, and why settlement cycles getting shorter made the middle one non-negotiable.

Front Office, Middle Office and Back Office aren't three departments arguing about the same trade — they're three different, equally correct answers to the question "is it done?" Why the split exists, what each clock actually measures, and why settlement cycles getting shorter made the middle one non-negotiable.

Surya · 8 min read

Capital Markets

What a Trade Lifecycle Actually Looks Like

A trade is not done when two people agree on price.

Six pillars, one lifecycle — what actually happens between an order and a settled trade, and where a business analyst sits at each stage.

Surya · 4 min read

Capital Markets

Margin, Options: Two Ways the Same Capital Disappears

Nine out of ten individual traders in India's F&O market lost money last year. They didn't all lose it the same way.

F&O and margin trading don't produce one kind of loss — they produce two, and they work by opposite logic. One comes from leverage: a small cushion of capital controlling a much larger position, closed out by force the moment that cushion runs thin. The other comes from time: a premium that decays every single day regardless of which way the price moves. What each mechanism actually is, why India's SEBI and Europe's ESMA both found leveraged retail products losing money for the large majority of traders, and why those two losses aren't the same failure wearing different clothes — with paired Indian and international examples throughout.

Surya · 11 min read