An overview of how stock and options market structure work, what they force participants to do, and what onchain markets should take from it.
Retail Flow Is Valuable
Payment for order flow starts with adverse selection.
When a market maker posts a bid and an offer on a public exchange, it does not know who is about to hit it. The buyer might be a retail investor buying 50 shares of Apple on her lunch break. Or it might be a quantitative fund acting on a signal that Apple is about to move. The first trade is profitable for the market maker, the retail investor is not trading on information, and the market maker captures the spread. The second trade is not, the quant fund knows something the market maker doesn’t, and the market maker is about to be on the wrong side.
On a public exchange, both types of orders arrive in the same queue. The market maker prices the spread to cover its expected losses to the informed flow. That spread is a tax on everyone, including the uninformed retail trader who is subsidizing the market maker’s losses to smarter counterparties.
Retail flow is valuable precisely because it is, in aggregate, less toxic. The SEC’s proposed Order Competition release described individual-investor orders as “low-cost” flow that imposes lower adverse-selection costs on liquidity providers. PFOF is the mechanism that shares some of the resulting surplus with the broker who controls where the order goes.
Stock PFOF: Off-Exchange Internalization Against a Strong Benchmark
A retail customer submits a market order through Robinhood. The broker does not send it to NYSE or Nasdaq. It routes it to a wholesale market maker: Citadel Securities, Virtu Financial, or G1 Execution Services (Susquehanna). These three firms handle over 80% of U.S. retail equity market orders.
The wholesaler executes as principal: it takes the other side from its own inventory, filling at or better than the National Best Bid and Offer (NBBO). Typically it improves the NBBO by a fraction of a cent, sub-penny price improvement. The trade prints on a Trade Reporting Facility, not on a lit exchange. The wholesaler hedges or unloads the position on public venues afterward.
The wholesaler pays the broker for the privilege. In aggregate, about 34% of total U.S. equity volume now executes at wholesalers off-exchange. Over 90% of retail marketable orders specifically are routed to wholesalers rather than to lit venues. The execution quality evidence is generally strong. Dyhrberg, Shkilko, and Werner (2025) in the Journal of Financial Economics found that wholesalers price-improve 65.7% of orders versus only 9.5% on exchanges, saving retail investors roughly $1 billion per month in aggregate.
Although PFOF generally improves execution quality, the picture is not uniformly clean, as it is up to the broker to monitor and to send order flow to dealers who attempts best execution. Schwarz, Barber, Huang, Jorion, and Odean (2025) in the Journal of Finance ran a randomized controlled trial, 85,000 simultaneous market orders across five brokers, and found round-trip execution costs ranging from 0.07% to 0.46%. A 6.5× dispersion for the same trades at the same time.
In any case, the stock wholesaler’s optimization problem is narrow and they are incentivized to scale. The can price tightly around the NBBO as equity spreads are penny-constrained in liquid names. The largest wholesalers offer the lowest execution costs due to economies of netting, technology, and flow diversity. This is a natural-monopoly-adjacent business.
The stock broker’s optimization problem is also narrow. They do not make real-time per-order routing decisions but instead set up routing arrangements and then flow is sent automatically. The broker’s obligation is to monitor execution quality after the fact. FINRA Rule 5310 requires “reasonable diligence” to find the best market, with “regular and rigorous” reviews of execution quality.
Those reviews must compare price improvement, speed, fill rates, and transaction costs on a security-by-security and order-type basis, and explicitly must consider “the existence of internalization or payment for order flow arrangements.” Rule 606 requires quarterly public disclosure of routing destinations and material PFOF terms.
The conflict is real but bounded. The NBBO provides a hard, observable floor. Equity execution quality reduces to a single clean dimension: net price relative to the quote. A compliance department can audit this with reasonable confidence. The gap between “defensible” and “optimal” is relatively small.
Options PFOF: On-Exchange Privilege Inside a Weak Benchmark
Listed options must trade on registered exchanges and clear through OCC. There is no off-exchange internalization. But the market maker’s advantage is recreated inside exchange architecture.
Cboe’s Automated Improvement Mechanism (AIM) lets an order flow provider submit a customer order paired with an initiating contra order, triggering a 100-millisecond auction. Others can respond with competing prices. But the initiating market maker retains a guaranteed participation right: 40% of the fill if two or more competitors match the best price, 50% if only one matches. Nasdaq ISE’s PIM and BOX’s PIP work similarly.
Options spreads are dramatically wider than equity spreads, often 5–20% of the option’s value, because liquidity is fragmented across hundreds of strike-expiry combinations and market makers must price gamma, vega, skew, and term-structure risk into every quote.
The money follows the spread. Optiver’s analysis of 2020 Rule 606 data found that 61% of all PFOF, $1.5 billion out of $2.5 billion, came from options, not equities. Citadel Securities channels approximately 71% of its PFOF spending to options and only 29% to equities. The SEC’s 2000 Special Study found that options specialists paid for 78% of retail options orders examined, and those payments were “generally not passed through to customers.”
The options market maker’s problem is fundamentally different from the stock wholesaler’s. Because the rent per order is large, the MM’s priority is securing high quality infrastructure on routing and exchange to capture flow (you want to direct good flow to exchanges where you are top of book, direct bad flow to exchanges where other people are top of book, as well as being able to intercept the good from others while avoiding bad flow), auction-initiation rights, and directed-order arrangements.
Since each of the 16 different U.S. options exchange has different fee schedules, allocation rules, and auction mechanics. The MM’s venue strategy is itself a competitive variable. This is a market-structure-control problem.
The battleground is slightly less who can price and execute better but more who controls the structural position and technology to exploit exchange rules better. Separately, it also matters as well who can warehouse multidimensional risk better. Stock wholesalers hedge in one dimension. Options market makers manage delta, gamma, vega, theta, and skew across a surface of strikes and expirations.
The broker still has to exercise reasonable due diligence to represent the best interests of the customer, but the difference is that the periodic review is much harder because execution quality is multidimensional (do you just want to optimize for price improvement, or also fill rate, or execution time since underlying prices can also move in 100ms). The gap between “defensible” and “optimal” is relatively large, and brokers may be incentivized to route flow to dealers who pay more and executes worse especially since retail investors themselves don’t monitor execution quality.
There is a second inherent conflict of interest that has no equivalent in equities: because options PFOF is so much larger, brokers may have incentives to encourage customers to trade options rather than stocks. This is not a routing distortion but an asset-class steering distortion. The market structure shapes not just where orders go but what retail is nudged to trade.
Why Each Structure Fits Its Asset (But Not the Other)
Stock-style bilateral internalization works in equities because the public benchmark is strong. Deep lit NBBO, penny ticks, abundant hedging liquidity, one-dimensional risk. The wholesaler can improve a tight quote, hedge cheaply, and the public market still anchors price discovery even with retail flow siphoned off.
That same structure fits options badly. The benchmark is weak: fragmented across hundreds of strike-expiry pairs, thin displayed depth, valuation driven by an entire volatility surface, and wide spreads. Pushing options into pure bilateral off-exchange internalization would weaken an already fragile benchmark further.
Options-style on-exchange auctions with participation entitlements make more sense in options because options genuinely need a formal mechanism to surface counterparties around a thin, dispersed book. But that same structure makes less sense in equities, where the spread is too tight to justify auction time.
Ambiguity Is Where Rent Extraction Lives
The deepest difference between stock and options PFOF is how easy the execution is to evaluate. In equities, execution quality is almost entirely one-dimensional: net price relative to the NBBO. Did the customer get filled better than the displayed quote? By how much? Fill rate at wholesalers is effectively 100% for normal retail size. Speed is fast and roughly uniform across venues. A compliance department can audit this quarterly with reasonable confidence. The benchmark is hard, the metric is clean, and the gap between “compliant” and “good” is narrow.
In options, every dimension fractures. Price improvement against what? The displayed quote is often 5 to 10 cents wide. “Improving” a 10-cent spread by 2 cents is nominally 20% price improvement, but that 10-cent spread may itself be stale, strategically wide, or uncompetitive. When the benchmark is soft, improvement relative to it means less.
Fill rate trades off against price. In equities, the wholesaler fills nearly everything. In options, fill is not guaranteed, particularly for larger size, less liquid strikes, or wider-spread names. A market maker can show tight price improvement on the orders it chooses to fill while passing on the ones where it would lose money. A 50% fill rate at 3 cents of improvement is not obviously better than a 95% fill rate at 1 cent.
100 milliseconds is a long time. CBOE AIM runs a 100ms auction. In that window, the underlying stock can move several pennies, enough to shift the option’s theoretical value by several cents through delta and gamma. The initiating market maker prices first, at a moment it chose, using its own fair-value model. Competing responders must observe the auction, compute their own fair value, and decide whether to bid, all within the same 100ms.
Greeks create hidden costs that no disclosure captures. In equities, a worse price is straightforwardly a worse fill. In options, execution quality also depends on where the fill sits on the volatility surface. A market maker can offer seemingly competitive pricing on a specific contract while knowing the implied vol it’s selling at is cheap relative to its surface model (or indeed, lift or offer the vol surface prior to larger orders that it is expecting).
In equities, an exploitative routing arrangement is hard to sustain because the evaluation is simple enough to audit. In options, multidimensionality creates genuine ambiguity, and ambiguity is where exploitative relationships survive.
A market maker offers better fill rates on wide-spread names but worse price improvement on tight names. Or it fills fast but at prices already behind fair value. Or it provides excellent execution on high-volume series and poor execution on illiquid ones the broker will never scrutinize. There is almost always a way to construct a narrative that the current arrangement satisfies Rule 5310 even when it is meaningfully suboptimal for the customer.
The evaluation surface is rich enough that an arrangement can be simultaneously defensible under regulation and meaningfully suboptimal for the customer. The wider the spread, the more complex the instrument, and the more dimensions of quality that matter, the larger the gap between “compliant” and “good.”
This is ultimately two overlapping problems. It is a principal-agent problem: the broker routes on behalf of the customer, is paid by the market maker, and faces a conflict between maximizing PFOF revenue and maximizing customer execution. This is the standard PFOF critique and it applies in both equities and options.
But it is also an information asymmetry problem: the market maker knows far more about the true quality of the execution than either the broker or the customer. It knows its own fair-value model, its fill selection criteria, its surface positioning, and the real economics of each trade. The broker sees reported price improvement, fill rate, and speed, summary statistics that the market maker can optimize for while leaving the deeper economics favorable to itself.
In equities, the principal-agent problem exists but the information asymmetry is small, the NBBO is a sufficient statistic for execution quality, and the market maker cannot easily hide behind it. In options, both problems exist simultaneously and reinforce each other.
The agent (broker) has a conflict of interest and cannot observe the thing it would need to observe to resolve it. The principal (customer) is two layers removed and sees even less. That double bind, misaligned incentives plus asymmetric information, in a setting where the benchmark is weak and the evaluation surface is high-dimensional, is the structural reason options PFOF extracts more rent than equity PFOF.
What Onchain Markets Should Learn (And Why It’s Hard)
Onchain markets inherit both halves of this problem. The principal-agent problem reappears whenever a front end, wallet, aggregator, or intent router controls where a user’s order goes. The user delegates execution to an intermediary. The intermediary may be compensated by the counterparty, through solver kickbacks, exclusive fill arrangements, MEV rebates, or builder-side payments. The user sees a fill price and maybe a slippage estimate. The intermediary sees the full economics. This is PFOF with different labels.
The information asymmetry problem reappears in proportion to the complexity of the asset and the execution. For a BTC-USDT swap on a CLOB, execution quality is roughly one-dimensional, price against a visible book, and the user can evaluate it. For a long-tail low-float cross-chain swap routed through a solver that splits the order across three AMMs, bridges one leg, and pays priority fees to avoid sandwich attacks, execution quality is a bundle of price, slippage, MEV exposure, gas cost, fill probability, route risk, and bridge settlement time. The user cannot decompose that bundle. Neither, realistically, can the front end.
The tempting conclusion is “just add transparency.” But the options analogy tells you why that’s insufficient. The problem is not that information is hidden. Options exchanges publish auction results, price improvement statistics, and fill rates. Rule 606 requires disclosure of PFOF arrangements. Rule 5310 mandates best-execution reviews. The information exists. The problem is that when the evaluation surface is high-dimensional, disclosure does not resolve the asymmetry.
Giving the user the raw onchain data their solver left behind does not help if the user cannot independently assess whether the execution across six simultaneous dimensions was good or bad. For complex, wide-spread, thin-liquidity assets, the ones where the rent pool is largest and the exploitation risk is highest, the evaluation problem is genuinely hard.
For intent-based execution, frontends should therefore probably have the option to let the user be their own broker. Expose the parameters that matter, maximum slippage tolerance, minimum acceptable fill probability, time limit for solver competition, acceptable gas overhead, and let the user set them, or at least choose a preset that matches their priorities. Then expose the intent to open solver competition under those constraints.
This is the onchain equivalent of a broker’s best-execution policy, except the user defines the policy instead of delegating it to an intermediary with a conflict of interest. No single smart contract parameterization will be “best execution” for everyone, instant fill at any slippage is not the same objective as tight slippage with a willingness to wait, but making the tradeoffs explicit and user-controlled, then letting solvers compete within those bounds, is the closest you get to solving the principal-agent problem and the information-asymmetry problem simultaneously.
That said, if you are trading weird, thin, complex instruments onchain, you need to be sophisticated enough to set those parameters. There is no way around this. The whole point of the options PFOF analysis is that complexity creates evaluative fog, and evaluative fog creates rent.
Outsourcing execution of exotic swaps to a single platform that handles everything behind a “swap” button is the onchain equivalent of a retail broker routing options flow to a market maker, it looks simple, it might even look cheap, but you cannot tell whether the execution is good because you have given up every lever that would let you check (for how egregious execution quality sometimes is, I have personally seen calendars from retail print a negative price when executed by a MM).
For liquid majors on a CLOB, that is fine, the evaluation is one-dimensional and the book is visible. For everything else, the user who does not understand execution parameters probably should not be trading the instrument.
Disclaimer
The information provided on TheLogbook (the “Substack”) is strictly for informational and educational purposes only and should not be considered as investment or financial advice. The author is not a licensed financial advisor or tax professional and is not offering any professional services through this Substack. Investing in financial markets involves substantial risk, including possible loss of principal. Past performance is not indicative of future results. The author makes no representations or warranties about the completeness, accuracy, reliability, suitability, or availability of the information provided.
This Substack may contain links to external websites not affiliated with the author, and the accuracy of information on these sites is not guaranteed. Nothing contained in this Substack constitutes a solicitation, recommendation, endorsement, or offer to buy or sell any securities or other financial instruments. Always seek the advice of a qualified financial advisor before making any investment decisions.