Payment for order flow (PFOF) is the practice in which a retail broker (Robinhood, Schwab/TD Ameritrade, Webull, E\Trade) is paid a small per-share or per-contract rebate to route its customers' marketable orders to a wholesale market maker — chiefly Citadel Securities, Virtu Financial, and a handful of others — rather than to a public exchange. The wholesaler then internalizes the order: it takes the other side itself out of inventory, profiting from the bid-ask spread. The core tension is that this arrangement is what funds "commission-free" trading, yet it creates a structural conflict of interest — the broker is paid by the very firm whose execution quality it is legally obligated to police on the customer's behalf. The same internalization machinery also concentrates retail order flow in identifiable streams, which gives rise to a separate idea: that retail positioning* (where the crowd is buying and selling) can itself be measured and used as a signal.

How it works (the mechanics)

1. A retail customer submits a market or marketable limit order in a stock or option. 2. The broker, instead of sending it to NYSE/Nasdaq, routes it to a wholesaler under a standing arrangement. The broker receives PFOF — typically fractions of a penny per equity share, and meaningfully more per options contract. 3. The wholesaler executes the order against its own book at, or better than, the National Best Bid and Offer (NBBO), pocketing some of the spread and any short-term inventory edge. The portion of the spread it returns to the customer is called price improvement.

Two SEC disclosure rules govern transparency: Rule 606 requires brokers to disclose their order-routing arrangements and PFOF received, and Rule 605 requires execution-quality statistics. The SEC adopted a major Rule 605 overhaul in March 2024 that broadens covered order types and reporting metrics — a direct response to long-standing complaints that the old metrics let brokers obscure execution quality. The compliance date, originally December 14, 2025, was extended by the SEC in October 2025 to August 1, 2026 (SEC FAQ / Federal Register).

Concentration is extreme. Industry reporting (The TRADE; Global Trading, citing 606 data) has put Citadel Securities at roughly 41%, Virtu at ~26%, and G1 Execution at ~16% of U.S. retail equity marketable order flow — i.e., the top three handle over 80%. Citadel Securities alone reportedly paid on the order of $2.6 billion in PFOF across 2020–2021, the majority of it in options, where rebates are far richer than in equities.

How it's used in practice

PFOF matters to a markets analyst in two distinct ways.

As a cost/conflict lens on execution. Sophisticated traders treat PFOF as a reason to check execution quality, not assume it. For small marketable retail orders in liquid names, internalization usually delivers price improvement versus the displayed NBBO. For larger orders, illiquid names, or options, the picture is weaker and the conflict more consequential.

As a positioning signal. Because internalized retail flow can be inferred from public tape data, researchers and quant desks reconstruct retail order imbalance — net retail buying vs. selling per stock. Vendors and dashboards (e.g., Nope/retail-flow trackers, and the academic BJZZ algorithm) attempt the same. This is the empirical heart of "retail positioning": treating the retail crowd's aggregate tilt as information, whether to follow it or fade it.

Adoption, debate & evidence

PFOF is ubiquitous in the U.S. retail brokerage business and is the financial mechanism that makes zero-commission trading viable. It is also one of the most contested practices in market structure.

The harm case. In December 2020 the SEC charged Robinhood and it settled for $65 million, finding the firm misled customers about PFOF being its largest revenue source and failed its duty of best execution; the SEC estimated customers were deprived of over $34 million in price improvement (Oct 2016–June 2019) even after accounting for commission savings. This is the canonical evidence that the conflict can bite.

The "it's fine" case. Multiple studies and the SEC's own DERA staff find that PFOF is, on net, associated with price improvement for retail orders and does not clearly harm execution. A widely cited point: PFOF rebates of roughly a tenth-to-fifth of a cent per share (fractions of a cent — e.g. NBER working-paper figures around $0.0005–$0.0015/share) are a tiny fraction of the observed variation in price improvement, suggesting they don't drive execution quality much either way (research summarized by Wharton WIFPR; SEC DERA staff analysis).

The honest synthesis is heterogeneity. Execution quality varies a lot across brokers — research finds brokers earning more PFOF (Robinhood was the example) provided negligible improvement, while lower-PFOF brokers (TD Ameritrade) provided economically meaningful improvement. And it varies across asset classes: equity wholesalers often beat exchange spreads, but the options market is far less competitive — Ernst & Spatt argue exchange rules (DMM internalization privileges) create a barrier to entry that protects wholesaler profits and sustains the much higher options PFOF rates, so the retail benefit is weaker there (Ernst & Spatt, Review of Financial Studies, advance article 2026).

Regulatory trajectory. The SEC under Gary Gensler proposed a 2022 "order competition rule" forcing certain retail orders into open auctions; it was withdrawn in 2025 under Chair Paul Atkins. The EU went the other way: under MiFIR Article 39a, PFOF is banned, with Germany's transitional exemption sunsetting June 30, 2026 — so the U.S. keeps PFOF while the EU eliminates it.

Does retail positioning predict returns? Here the evidence is nuanced and easy to misread. Boehmer, Jones, Zhang & Zhang (Journal of Finance, 2021) found that stocks with net retail buying outperformed net-selling stocks by ~10 bps over the following week — i.e., short-horizon retail imbalance is positively correlated with future returns, the opposite of the lazy "dumb money, fade it" story. But Barber, Lin & Odean ("Resolving a Paradox," JFQA, 2024) show the catch: retail order imbalance positively predicts equal-weighted returns, yet retail investors still lose money on average — because their purchases concentrate in attention-grabbing stocks that subsequently underperform (a long-short imbalance strategy earned roughly -15% annualized in heavily-traded names vs. +6-7% elsewhere, per the paper). Crucially, later work (arXiv, 2024) finds the BJZZ signal weakened or reversed in more recent samples, so this is not a stable, bankable edge.

Strengths & limitations

When the concept is useful: as a discipline to verify execution rather than trust it; and, in the positioning sense, as one contextual input — extreme retail crowding in a name (especially with options skew) flags conditions where short squeezes, gamma effects, and reversals cluster (the 2021 GameStop episode being the archetype).

When it fails / the #1 misuse: treating "retail = dumb money to fade" as a mechanical rule. The peer-reviewed evidence does not support that; the short-horizon imbalance signal has historically been positive, and it has decayed over time. Equally, assuming PFOF always means you got a bad fill is wrong — for small liquid equity orders it usually doesn't. The deeper limitation is inference quality: retail-flow reconstruction from the tape (the BJZZ subpenny method) is an estimate with material classification error — a 2024 Journal of Finance validation (Barber et al.) found it mis-signs roughly a quarter of the trades it labels retail — and it captures only the internalized marketable slice, not limit orders or dark activity.

Sources

• SEC, SEC Charges Robinhood Financial With Misleading Customers... (press release 2020-321) — $65M settlement; "over $34 million" lost price improvement (Oct 2016–June 2019).
• SEC Rule 605 amendments (press release 2024-32; compliance-date extension to Aug 1, 2026 per Federal Register 2025-19316 / SEC Rule 605 FAQ).
• SEC DERA staff analysis on PFOF and price improvement (as summarized in policy/academic literature); CRS IF12332 on the SEC's ~$1.5B estimated annual savings from the proposed auction model.
• NBER w29883 (Ernst & Spatt, PFOF and Asset Choice) — per-share price-improvement figures (~$0.0005–$0.0015/share).
• Wharton WIFPR, Research Spotlight: Payment for Order Flow and Price Improvement.
• Ernst & Spatt, Payment for Order Flow and Option Internalization, Review of Financial Studies (advance article, 2026) — options internalization is imperfectly competitive; weaker retail benefit, higher PFOF.
• Boehmer, Jones, Zhang & Zhang, Tracking Retail Investor Activity, Journal of Finance (2021) — ~10 bps/week.
• Barber, Lin & Odean, Resolving a Paradox: Retail Trades Positively Predict Returns but Are Not Profitable, JFQA 59(6) (2024).
• Revisiting Boehmer et al. (2021): Recent Period, Alternative Method, Different Conclusions (arXiv 2403.17095, 2024; pub. Financial Markets and Portfolio Management, 2025) — ROI predictive power weakens, no longer profitable for large caps in 2016–2021.
• Barber et al., A (Sub)penny for Your Thoughts: Tracking Retail Investor Activity in TAQ, Journal of Finance (2024) — BJZZ classification error (mis-signs ~28% of identified trades).
• The TRADE; Global Trading — wholesaler market-share and Citadel ~$2.6B PFOF figures (606-based reporting).
• Congressional Research Service IF12332 / IF12594; ESMA MiFIR Article 39a; Sidley/Cleary notes on 2024 Reg NMS amendments and the 2025 withdrawal of the order competition rule.

Disputes flagged: (1) whether PFOF on net harms or helps retail execution is genuinely contested — evidence is broker- and asset-class-dependent, not uniform; (2) the retail-positioning predictive signal is real in-sample but modest and unstable out-of-sample. Market-share and Citadel PFOF dollar figures are from industry/press reporting of 606 disclosures, not a single audited source — treat as approximate.