Payment for Order Flow Is Not Why Your Trade Costs More

When you tap Buy on a commission-free trading app, your order does not go to the stock exchange.

Not first.

It goes to a private firm that will sell you the share out of its own inventory — and that paid your broker for the right to see your order before anyone else did.

This arrangement is called payment for order flow. It is legal in the United States, banned in the European Union, and it has been the subject of more retail outrage than any other piece of market plumbing in the last decade.

The outrage is aimed at the wrong target.

We know this because five academics went and measured it.

The experiment nobody reported

In 2022, Christopher Schwarz, Brad Barber, Xing Huang, Philippe Jorion and Terrance Odean did something researchers almost never do: instead of inferring execution quality from public trade data, they opened six real brokerage accounts at five brokers and traded their own money.

Then they placed 85,000 simultaneous market orders. Same stocks. Same seconds. Real fills.

This design matters more than it sounds. Most market-structure research has to guess whether a trade was a buy or a sell, because the public tape does not say. Schwarz and his co-authors knew — they had placed every order themselves. For the first time, someone could say with certainty what the same trade cost at five different brokers at the same moment.

The paper was published in the Journal of Finance in 2025. It should have been front-page financial news. It was not.

What they found

Across the six accounts, the average round-trip cost of a trade ranged from 0.07% to 0.46%.

Same stocks. Same seconds. A sixfold difference, depending entirely on which app you happened to have installed.

Measured against the quoted spread — the more honest yardstick — it is starker still:

Broker	Price improvement captured	What a round trip actually cost
TD Ameritrade	47.2% of the spread	5.6% of the spread
Robinhood	26.8%	46%
Interactive Brokers Pro	18.8%	62%

The authors’ own summary of that gap: “over 10 times more than TD.”

Sit with that. Two investors, buying the same share in the same second, with the same order type. One hands over 5.6% of the spread. The other hands over 62%.

Now the part that ruins the story everyone wants to tell

The obvious explanation is payment for order flow. The broker sold you out. Follow the money.

Except the researchers checked. And the money does not go where the story says it does.

“We find that PFOF explains almost none of the cross-broker variation in execution prices.”

— Schwarz, Barber, Huang, Jorion & Odean, The ‘Actual Retail Price’ of Equity Trades

The payments themselves are tiny: $0.001 to $0.002 per share. A tenth of a cent. Against a price-improvement gap measured in whole cents per share, PFOF is not the mechanism — it is a rounding error attached to the mechanism.

And the paper contains a detail that ought to end the argument on its own: two of the brokers received no payment for order flow at all, and still delivered worse execution than a broker that did.

If PFOF were the cause, that is impossible.

So something else is setting your price.

What actually sets your price: who else uses your broker

Here is the mechanism, and it is not hidden. It is just never explained to the people it applies to.

A wholesale market maker fills orders from its own inventory. Its entire business is the spread — the gap between what it pays and what it charges. It makes that spread thousands of times a second, and it loses money whenever it trades against someone who knows something it does not.

So it does the only rational thing: it prices each broker’s flow according to how dangerous that flow is.

Flow from a broker whose customers are mostly professionals — people running screens, reacting to filings, trading on information — is toxic. Fill those orders carelessly and you get picked off. So you quote carefully, and you keep less.

Flow from a broker whose customers are mostly retail investors buying what they read about this morning is benign. It is, in aggregate, uninformed. You can quote a wider spread and keep more of it, because on average you will not be run over.

Which produces a conclusion most people find genuinely strange the first time they hear it:

You do not pay for your trades. You pay for who else uses your broker.

Your execution quality is set by the average sophistication of strangers who happen to have downloaded the same app. You cannot negotiate it, you cannot improve it by trading better, and no disclosure document will ever put it in those words.

This is not a conspiracy. Nobody is cheating. It is a market maker doing exactly what a market maker is supposed to do — pricing risk — and the risk it is pricing is you, in aggregate, as a category.

While we are here: dark pools are not the problem either

The most common thing retail investors believe about market structure is also wrong, so let us dispose of it.

The viral version: your orders vanish into “dark pools” where institutions trade against you in secret.

The reality: dark pools are not where your order goes.

Dark pools are private venues built for the opposite problem — a pension fund that needs to sell a very large block of shares without the act of selling crashing the price against itself. They exist to hide size, not to hide you.

Your retail order goes somewhere else entirely: to a wholesale market maker, which fills it from inventory. Different venue. Different participants. Different business model. Different problem.

Conflating the two is the single most common mistake in this entire debate — and it has the unfortunate effect of pointing people’s anger at a venue they never touch, while the thing that actually costs them money goes unexamined.

Three regulators looked at the same plumbing and disagreed completely

Within roughly two years, three of the world’s major financial regulators examined this exact system and reached three incompatible conclusions.

Europe banned it. Article 39a of MiFIR, introduced by Regulation (EU) 2024/791, prohibits investment firms from receiving any fee, commission or non-monetary benefit from a third party for routing retail orders to a particular venue. It entered into force on 28 March 2024 and applied to cross-border business immediately.

Member states were allowed to exempt purely domestic business — but only until 30 June 2026, and only if their firms were already doing it.

Exactly one member state asked: Germany. Which is not a coincidence. The German neo-broker generation — Trade Republic, Scalable Capital, Smartbroker — was built on this revenue model, and it needed time to rebuild the business around something else.

That exemption expired on 30 June 2026. As of today, there is no corner of the European Union where a broker may be paid to route your order.

America looked at the same thing and walked away. Under Chair Gary Gensler the SEC proposed two rules that would have gutted the model: the Order Competition Rule, which would have forced retail orders into a competitive auction before a wholesaler could internalise them, and Regulation Best Execution. Then the administration changed. On 12 June 2025, under new Chair Paul Atkins, the SEC formally withdrew both — part of a withdrawal of fourteen proposals — and stated it did not intend to issue final rules.

Payment for order flow remains entirely legal in the United States.

Britain, having banned it back in 2012, is now reconsidering — reviewing whether the prohibition should be loosened, at precisely the moment the EU tightened its own.

Three regulators. One piece of plumbing. Three opposite directions. If you were waiting for the grown-ups to agree on whether this is harmful, stop waiting.

The twist: what happens when the broker becomes the market maker

Here is the part of the story that the ban did not anticipate.

Article 39a forbids a broker from being paid by a market maker for routing your order.

It does not forbid the broker from being the market maker.

Scalable Capital launched its own regulated venue — the European Investor Exchange — in December 2024, and acts as one of the market makers on it. Trade Republic obtained a BaFin licence to run its own multilateral trading facility, and is now its own market maker on its own venue.

Look at the architecture before and after:

• Before the ban: the market maker captures the spread, and pays a slice of it back to the broker. That slice is PFOF, and it is now illegal.
• After the ban: the broker is the market maker. It captures the entire spread. It pays no one — because there is no one to pay.

The payment is gone. The economics did not go anywhere. They moved inside the same company, where no disclosure regime is currently pointed at them.

Whether this is functionally identical to the practice the EU just outlawed is a question European regulators have not yet ruled on. It is, at minimum, an awkward one.

So what does this cost you?

This is the part where most coverage of this topic loses its nerve and implies that you are being fleeced. You are probably not. Here is the honest version, split by the only thing that determines the answer: what you actually do.

If you buy an index fund every month and never sell

This is close to irrelevant to you, and you should feel free to stop thinking about it.

The spread is a one-time cost paid at the moment of purchase. You pay it once per contribution, on a broad, deeply liquid instrument where spreads are thin to begin with. Over an investing lifetime it is a small number — and it is comprehensively outweighed by two things that are not small: the fund’s ongoing charge, which you pay every year on the entire balance, and your own behaviour in a bad market.

If you came to this article worried, the correct response is relief. Structure matters more than venue. Keep buying.

If you trade single stocks, actively

Then this is not a rounding error. It is a toll, and you pay it on every decision.

A sixfold difference in execution cost, applied to every position you open and close, across a year of trading, in names that are less liquid than the mega-caps the study measured — that is real money, and it is money you lose before your investment thesis has been tested at all.

Broker choice is one of the very few decisions in investing that you make once and get paid for every year afterwards. It deserves an afternoon.

The three questions

If you want to know where you stand, ask your broker these — and pay attention to which one they answer vaguely.

1. Do you publish execution quality data? Rule 605 and Rule 606 reports in the US; RTS reports in Europe. Actual numbers, not a marketing page that uses the words “best execution” as a slogan.

2. Where do my orders actually go? A public exchange? A wholesale market maker? A venue you own yourself? All three are legal. Only one of them is what most customers assume.

3. On that venue, is there one market maker — or a competitive auction? A single market maker on a venue owned by your broker is not a market. It is a counterparty with a monopoly on your flow.

The thing worth remembering

The debate got captured by the wrong villain.

Payment for order flow is visible, it has a sinister name, it involves Citadel, and it fits neatly into a paragraph. So that is what everyone argued about — while regulators banned it, brokers restructured around it, and the actual driver of what your trade costs went almost entirely unexamined.

The actual driver is not a payment. It is a judgement about you, made in microseconds, by a firm you will never speak to, based on the observed behaviour of everyone else who uses your broker.

The system is not broken. It is working exactly as designed.

It was simply never designed with you in mind.