Choosing a Brokerage Account in 2026: Fees, Execution, Protection, Cash

Commissions went to zero in 2019 and the comparison industry kept ranking brokers as if price still varied. It does, everywhere except the sticker: in what your idle cash earns, in how your orders are routed, in what margin costs, in which account type wraps the whole thing. This page is the grid for reading the economics of a brokerage account in 2026, line by line.

1. The wrapper before the storefront

The first decision is not which broker; it is which account, and the industry’s storefronts are built to make you skip it. The same index fund held in a taxable account, a traditional IRA or a Roth produces three different after-tax outcomes, and the gap between wrappers routinely exceeds any gap between brokerages. The choice between tax-advantaged and taxable wrappers is therefore the entry point of this grid, not an appendix to it.

The wrapper layer has its own internal decisions, each with a documented answer. Whether the employer plan or the individual account takes priority is a question of costs and menus, mapped in employer wrapper versus individual wrapper. When the tax is paid, now or at withdrawal, is the entire difference between the two account families, and the two clocks of Roth and traditional accounts shows why the answer depends on tax rates you can only estimate. The funding order is the one near-consensus in the sequence: the match-first funding convention exists because an employer match reprices every other dollar in the chain. And the exit rules are the fine print that decides liquidity: early withdrawal penalties and their exceptions is what makes a retirement dollar different from a brokerage dollar long before returns enter the picture.

The wrapper layer also sets the stakes for everything below. Fees inside a taxable account reduce returns; fees inside a retirement account reduce returns that decades of compounding were supposed to multiply, which is why an identical cost line weighs more in the account you will hold longest. Everything below assumes the layer is settled and reads the account itself: the brokerage as an economic machine.

2. The evaluation grid: six lines that replace the sticker price

The account feels free; it is monetized. Since the sticker price is zero, the grid moves to where the money actually flows. Six lines: residual commissions and fees (options contracts, mutual fund transactions, transfer-out charges); execution quality and order routing; the cash sweep rate; margin pricing; protection and custody; and the account menu itself, which section 1 covered. Each line is observable in a public document, which is the quiet advantage of the US disclosure regime: the grid can be filled without trusting anyone’s marketing.

Zero commission is a price tag, not a cost. The business model did not shrink when commissions vanished; it relocated. A brokerage earns on the spread between what your idle cash yields it and what it pays you, on routing revenue from market makers, on margin interest, on securities lending, on fund fees where it manages the funds. None of this is scandalous; all of it is measurable; and the dispersion across brokers on each line is wider than the commissions ever were.

The residual fee schedule is the easiest line to read and the least decisive. Options still price per contract, typically $0.65 at the large houses (published schedules, 2026); transaction-fee mutual funds carry charges around $50 to $75 per purchase; wires, paper statements and outgoing transfers each have a line. What the schedule reveals is less the amounts than the shape of the clientele the broker prices for: a house that waives fund fees but charges for broker-assisted trades is telling you who it wants. The instructive exercise is reading the schedule backwards, from the revenue side: for each line, ask what behavior of yours generates it. An investor who holds index ETFs, trades rarely, keeps no idle cash and borrows nothing generates almost no revenue anywhere, which is precisely why the industry’s economics concentrate on the customers who do one of those four things without noticing.

A reading order helps, because the six lines do not weigh equally for a given holder. For a long-horizon index investor the sweep and the fund lineup dominate; for an options trader, routing and per-contract fees; for anyone borrowing, margin terms; for everyone, the wrapper. The grid is filled once per broker but weighted once per person, and the weighting is the part no comparison site can do, since it requires knowing how much idle cash, trading volume and leverage the account will actually see. What follows takes the lines in the order their costs are usually discovered: last.

3. Order execution and payment for order flow

A zero-commission retail order is routed, most often, to a wholesale market maker that pays the broker for the flow. Whether the customer loses anything depends on execution quality: the fill relative to the NBBO, the consolidated national benchmark quote that regulation defines for every listed security. The regime makes this inspectable rather than debatable: SEC Rule 606 requires quarterly reports disclosing where each broker routes its orders and what payment it receives, and Rule 605 reports from execution venues quantify price improvement. Reading a 606 report takes ten minutes and replaces an hour of forum opinion; the report sits on the broker’s own site, filed quarterly, because filing it is the law. Reading it also reveals how little of that flow ever reaches an exchange, which is how much trading now happens away from the visible book.

The economics are worth stating neutrally. Payment for order flow transfers part of the bid-ask spread from the market maker to the broker; brokers that refuse PFOF typically charge elsewhere or route for price improvement they can advertise. For an investor placing a few market orders a month in liquid large caps, measured execution differences amount to cents per share; for an active options trader, routing economics scale into real money, since options PFOF per contract runs far higher than equity PFOF per share. The practice moved from footnote to headline in early 2021, when meme-stock volatility put retail routing in front of Congress and made 606 reports briefly famous; the regulatory outcome was more disclosure rather than prohibition, which is why the reading skill matters more than the opinion. The scale asymmetry between products is the detail worth retaining: equity PFOF is measured in hundredths of a cent per share, options PFOF in tens of cents per contract, so the routing economics of an options-heavy account differ in kind, not degree, from an index investor’s. The grid line is not “PFOF bad” but “PFOF disclosed”: a broker whose 606 report shows concentrated routing to a single paying venue, on order types where price improvement statistics lag, has answered the question the marketing avoids. The practical test fits in one sitting: pull the current quarter’s 606 for the brokers on your shortlist, compare the share of flow routed to paying venues and the disclosed payment per hundred shares, and note whether the numbers vary more across brokers than your annual trading volume can make material. For most index-fund households, they do not, which is itself a finding: the routing line matters in proportion to order flow, and the grid weighs it accordingly.

4. The cash sweep: the largest hidden number in the account

The sweep line deserves the grid’s longest section because it combines the largest amounts with the least attention. Uninvested cash does not rest; it is swept nightly into a destination the broker chooses, and that choice is the single widest pricing dispersion in retail finance in 2026. Default sweep destinations observed this year range from affiliated-bank programs paying 0.01% to 0.05% (E*TRADE, Schwab, per published 2026 schedules and comparison surveys) to money market sweeps paying roughly 3.3% to 3.6% (Fidelity’s core position near 3.3%, Vanguard’s settlement fund near 3.6%, per 2026 comparisons). On $100,000 of idle cash, the gap exceeds $3,000 a year, every year, for an identical service.

The dispersion is young as a visible problem: at the zero rates of the 2010s every sweep paid nothing and the line was moot, which is why account holders who opened their accounts then have often never looked. The mechanism is the business model made visible: a broker sweeping to its affiliated bank captures the spread between the policy rate, 3.50-3.75% at the June 2026 FOMC, and the 0.05% it credits. The pattern drew SEC scrutiny and a wave of class actions against several large firms in 2024, and some rates moved; the dispersion survived. The line extends beyond idle trading cash. Managed products embed cash allocations: the most prominent robo program held 6% to 30% of client portfolios in cash at the program’s sweep rate, a design that produced a $187 million SEC settlement in 2022 over how that monetization was disclosed. A portfolio of $300,000 carrying a structural 10% cash sleeve at a 0.05% sweep, against the same sleeve at 3.5%, gives up roughly $1,000 a year to an allocation decision the holder never explicitly made. Two mitigations exist and both cost attention: manually parking cash in a money market fund inside the low-sweep account, or choosing a broker whose default already is one. Where that cash earns most after tax is its own comparison, run in where idle cash earns after tax. The grid line is a single question with a published answer: what does the default sweep pay, today, in writing. The follow-up question is behavioral rather than contractual, and only the holder can answer it: how much cash does the account structurally hold, between contributions, after dividends, around rebalancing. The product of those two numbers, rate gap times average balance, is the line’s annual cost, and it is the only line on the grid the holder computes from their own statement rather than from the broker’s documents.

5. Margin, securities lending, and the price of leverage

Margin is the account’s credit card, and its pricing is as dispersed as the sweep: published schedules in early 2026 put all-in borrowing costs between roughly 5% at the price-leader discount brokers and 7% at full-service firms for comparable draws, spreads quoted over SOFR and negotiable at size. An investor who never borrows can skip the line; for anyone who might, it is worth reading before it is needed, because margin terms are set when the account opens, not when the loan is taken. The symmetrical line is securities lending: fully paid shares can be lent out by the broker, generating revenue that is shared with the customer at some houses and retained at others, at splits that range in published programs from half to nothing, with a side effect worth knowing: lent shares receive cash-in-lieu instead of qualified dividends, a small tax cost the enrollment screen does not mention.

Margin mechanics deserve two sentences of sobriety, because the price is only half the line. A margin loan is collateralized by the portfolio and marked continuously: under Regulation T and house maintenance rules, a falling account triggers a demand for cash or a forced sale, at the broker’s timing, in the market conditions that caused the fall. The cost of leverage is therefore the published rate in calm markets and the liquidation terms in stressed ones; the second number is in the margin agreement, and nobody quotes it in comparisons. An account intended to carry leverage is chosen on both numbers, not the advertised one.

6. Protection: what SIPC covers, and what it cannot

Protection is the grid’s most misread line, because the acronym carries more reassurance than its statute. SIPC coverage protects up to $500,000 per customer per capacity, of which $250,000 can be cash, against the failure of the brokerage itself: it restores custody of missing securities when a broker collapses or misappropriates assets. It does not, and was never designed to, protect against market losses, and cash swept into a program bank sits under FDIC rules instead, with different limits and mechanics. The regime has been tested at scale exactly once in recent memory, in the 2008 Lehman brokerage liquidation, where customer property was recovered and transferred under SIPA process; the precedent is why custody failure ranks low on the worry list. The distinction sorts real risks: the probability that a large US brokerage fails with custody gaps is remote; the certainty that markets fluctuate is total. Reading SIPC as an investment guarantee is the most common protection misreading in retail finance, and it inverts the actual hierarchy: custody risk is the one already handled.

The practical protection questions sit elsewhere: whether the broker carries excess-SIPC private insurance (most large ones do), how client assets are segregated, and, at the account level, the security features that prevent the failure mode that actually occurs, account takeover. None of these vary enough among the major houses to decide the choice alone; all of them belong on the grid as pass-fail checks. One nuance earns its sentence: sweep destination changes the protection regime. Cash in a money market fund is an investment covered by SIPC custody rules; cash in a program bank is a deposit under FDIC rules, spread across partner banks with per-bank limits. The higher-yielding arrangement and the more conventionally insured one are not always the same, and the account agreement states which applies, in the same paragraph nobody reads that publishes the sweep rate.

7. The cost stack, compounded

The simulator below builds the same contribution plan under two generic account profiles: one where fund fees run high and idle cash earns a near-zero sweep, one where both lines are cut to the low end of observed 2026 pricing. Same deposits, same 5% gross return assumption on invested assets, declared and identical for both branches: what diverges is the cost stack. The asymmetry is the point: the costs are certain, the return is an assumption. Vary the deposits and the horizon; the gap scales with both, because every dollar of cost is a dollar that stops compounding the day it leaves.

8. The grid read through the regime

One line of the grid is regime-dependent, and it is the sweep. At the 2010s’ zero rates, sweep dispersion was invisible: 0.01% and 0.5% round to the same nothing. At 3.50-3.75% policy rates (FOMC, June 2026), the same structural choice by the broker is worth three thousand dollars a year per hundred thousand of cash. The current environment, a transition reading on the current regime assessment, mixed signals, no clear direction (Eco3min classifier, June 2026), is precisely the kind in which investors hold more cash while waiting, which makes the sweep line temporarily the most expensive one on the grid. What each configuration historically did to cash, duration and equity sits in regime-by-regime asset behavior, and aligning vehicles with the regime extends the reading from asset classes to accounts, the level where this page operates. A cutting cycle would compress the sweep gap mechanically; nothing about the account needs to change for its costs to change. The same regime dependence runs through margin, priced over SOFR and repriced with every policy move, and through securities-lending revenue, which swells when short interest and rates are high. An account, in other words, is not a fixed-cost object: it is a bundle of spreads whose width the macro cycle sets, which is why a grid filled in 2021 needed refilling by 2023, and why this one carries its dates.

The tax line has its own clock, independent of the macro one: realized gains are priced by holding period, and lock-in effects when the tax clock meets a drawdown documents the interaction the grid cannot show, between when you need liquidity and when the code makes liquidity cheap.

9. FAQ

What does payment for order flow change for a retail order?

It changes who pays the broker. The order routes to a market maker that pays for the flow and fills at or inside the NBBO benchmark. Execution quality relative to that benchmark is measurable: Rule 606 reports disclose routing and payments quarterly, and venue-level Rule 605 statistics quantify price improvement. For small orders in liquid stocks the measured differences are cents; for active options flow they compound. What a commission-free trade actually costs and who collects it is settled at that moment, long before any commission line appears on a statement.

How does SIPC protection work and what does it exclude?

SIPC restores custody of securities and cash, up to $500,000 per customer including $250,000 in cash, if the brokerage fails or client assets go missing. It excludes market losses entirely, and cash swept to a program bank falls under FDIC coverage instead. Most large brokers add private excess-SIPC insurance above the statutory ceiling.

How do cash sweep rates differ across brokerages?

By two orders of magnitude in 2026: published default rates run from 0.01% to 0.05% at brokers sweeping into affiliated banks to roughly 3.3% to 3.6% at those sweeping into money market funds. The dispersion drew SEC attention and class actions in 2024. The rate is published; checking it takes one page of the account agreement.

Which costs remain in a zero-commission account?

Options contract fees, mutual fund transaction fees, transfer-out and wire charges, margin interest, the sweep spread on idle cash, fund expense ratios, and the execution spread embedded in fills. Each is disclosed somewhere: fee schedules, 606 reports, fund prospectuses, the sweep paragraph of the account agreement. Zero commission moved the account’s economics off the sticker, not off the statement.

How does an ACATS transfer work when switching brokers?

The receiving broker initiates the transfer through the automated ACATS system; whole positions move in kind, typically within about a week, without selling, so no tax event occurs. The delivering broker usually charges an outgoing transfer fee, disclosed in its schedule, and the receiving one often reimburses it. Proprietary funds that only exist at the old broker cannot move; they are sold or left behind before the transfer.

10. From account to allocation

A brokerage account is plumbing, and plumbing is judged on what it silently takes, not on what it advertises. Essential, priced, comparable, and still upstream of every decision that determines outcomes: what the account holds, in what proportions, held through which conditions, the layer treated in allocation architecture before account choice. The comparison of the firms behind the accounts, platforms, menus, service, is a separate exercise with its own grid, run in comparing the brokers behind the accounts, and what the account ultimately holds is the subject of the vehicle panorama the account will hold. What this page fixes is narrower and prior: the six lines where an account that advertises zero actually charges, each with a public document attached. An account read this way cannot surprise you. That is the entire ambition of the grid. The figures carry their dates, July 2026 throughout, and the January revision will replace them; the six lines will not move, because they are where the business model lives, whatever the rates are doing.

Last updated — 23 July 2026