Reg NMS
∞structuralRegulation National Market System (SEC 2005). Its Order Protection Rule requires execution at the best displayed price across all venues, and by linking fragmented venues into one grid, it created the conditions for cross-venue latency arbitrage.
The idea
Reference figure. This concept is explained in prose and diagram; the interactive widgets live on the flagship pages it links to under Where this fits.
What is Reg NMS?
Regulation NMS is a set of SEC rules adopted in 2005 (and phased in through 2007) that governs how US equities trade across the many venues that make up the "national market system". Its goal was a fairer, more connected, more competitive market; its effect was to formalise competition between venues and tie them together with price-protection and consolidated-data rules.
The intuition first: before Reg NMS, US equity trading was consolidating but contested. Multiple exchanges and ECNs competed, and the rules tying them together were patchy. Reg NMS set the terms of that competition. It said venues may compete freely, but they must respect each other's best displayed prices, and the public must get a consolidated view of the best price. (Source: SEC, Regulation NMS, Release No. 34-51808, 9 June 2005.)
Reg NMS has four pillars; the first two are what matter for microstructure and strategy. Rule 611, the Order Protection Rule (the "trade-through" rule): you cannot execute at a price worse than the best displayed price available on another protected venue (detailed below). Rule 610, the Access Rule: governs access to quotations and caps the access fee a venue can charge to take liquidity, the rule that underpins maker-taker (detailed below). Rule 612, the Sub-Penny Rule: prohibits displaying quotes in increments finer than \$0.01 for stocks priced at \$1 or more (the tick, now being revisited in the 2024 reforms). And the Market Data Rules: govern consolidation and distribution of quote and trade data, namely the SIP (the public consolidated tape) and how its revenue is shared.
The honest framing: Reg NMS is genuinely pro-competition and pro-investor in intent and in much of its effect: tighter spreads, more venue choice. But its structural consequence (a fragmented market tied by a consolidated-but-lagging tape) is also exactly what created the speed game. The rule and its unintended children are inseparable.
The Order Protection Rule (Rule 611) and the NBBO
Rule 611 forbids a "trade-through", executing an order at a worse price than the best displayed price (the "protected quote") on another venue. To enforce it, the market must continuously know the best bid and offer across all venues: the National Best Bid and Offer (NBBO). Rule 611 is therefore the rule that requires a consolidated, cross-venue best price.
One-line gloss: if a better price is showing somewhere else, you must not execute through it, so everyone needs a single, agreed "best price" to check against. The mechanics: a protected quote is the best displayed bid or offer on an automated, accessible venue; the NBBO is the highest protected bid and lowest protected offer across all such venues. When you send a marketable order, it must not execute at a price inferior to the NBBO; if a better price is displayed elsewhere, the order must be routed there (or the venue must satisfy a narrow exception). This is what makes a single instrument's fragmented order books behave as one connected market.
Why this is the structural rule: by mandating price protection across venues, Rule 611 made it both safe and attractive for many venues to compete for the same stock: a new venue's quotes are protected and must be honoured. Fragmentation is not a market failure under Reg NMS; it is the designed equilibrium. The US went from a few dominant exchanges to a dozen-plus lit venues plus dark pools, all tied by the NBBO.
And here is the catch that creates the speed game: enforcing the NBBO requires building the NBBO, and building it means consolidating data from every venue, which takes time. That consolidation step is the SIP, and its lag is the next section.
The SIP, the NBBO, and the latency gap
The Securities Information Processor (SIP) is the public consolidated feed that assembles the NBBO from every venue's quotes. Because consolidation is an extra hop, the SIP's NBBO arrives a beat behind each venue's own direct feed. A firm reading the fast direct feeds knows the true price before the slow SIP does, and that gap is the engine of latency arbitrage.
The intuition: the SIP is the official, public scoreboard. But scoreboards lag the game: someone has to read each venue, normalise it, and publish the consolidated view. The direct feeds are the game, in real time. Whoever watches the game beats whoever watches the scoreboard.
The structural consequence is the whole reason this page links so tightly to the strategies guides. A resting quote that was fair against the SIP NBBO can be stale against the direct-feed reality: the price has already moved on the fast feeds, but the SIP (and anyone pricing off it) has not caught up. A faster participant trades against that stale quote before it updates, classic latency arbitrage, and a large part of why colocation and direct feeds are worth paying for (colocation & FPGA).
The honest 2026 note: the SIP has been modernised (faster processing, and reforms expanding the data it carries: round lots, odd lots, depth), narrowing the direct-feed gap. But consolidation is inherently an extra step, so the gap narrows rather than closes, and the cross-venue fragmentation that fuels routing-level latency edges is permanent. The form of the edge has shifted; the structural opening has not vanished (as of 2026).
The Access Rule (Rule 610), access fees and maker-taker
Rule 610 governs access to quotes and caps the fee a venue may charge to take displayed liquidity (historically \$0.0030 per share). That cap underpins the maker-taker model: venues charge takers up to the cap and pay makers a rebate to attract quotes. It is the rule behind the rebate economics that drive a whole class of HFT behaviour.
One-line gloss: the rule that limits what a venue can charge you to hit a quote, which is what makes "pay makers, charge takers" economically possible. The mechanics and consequences: under maker-taker, a venue pays a rebate to the party that posts (makes) liquidity and charges a fee to the party that takes it, keeping the net within the Rule 610 cap. This is the economic engine behind rebate capture and shapes the maker-vs-taker decision on every order. The cap creates a measurable incentive (posting can be paid, taking costs), so strategies optimise to be on the rebate-earning side, and execution algorithms weigh the rebate against fill risk.
The 2024 reform: the SEC's 2024 equity-market-structure rules reduced the access-fee cap (to \$0.0010 per share for the most actively traded stocks) and tied it to finer tick sizes, directly reshaping rebate economics. This is live, parameter-level change to the rule that governs a whole strategy family; verify the current cap against the SEC's latest release (as of 2026).
The tick-size and access-fee debates (the 2024 reforms)
A tick is the minimum price increment a stock can be quoted in (Rule 612: \$0.01 for stocks priced at \$1 or more). For "tick-constrained" stocks (those whose natural spread is finer than a penny) the penny tick is too coarse, fattening spreads and lengthening queues. The SEC's 2024 reforms introduced finer increments for such names, and cut the access-fee cap alongside.
Why the tick matters to a trader: the tick sets the minimum spread and therefore the value of queue position. A coarse tick relative to true value means a wide spread, a long queue, and outsized rewards to being first in line, a structural feature market makers and rebate-capturers exploit. A finer tick narrows the spread and shortens the queue, shifting where the edge is.
The 2024 changes (verify against the SEC's final rules, as of 2026): finer minimum pricing increments for tick-constrained, actively traded stocks; a reduced access-fee cap (Rule 610) for those names; and related changes to round-lot definitions and SIP data content. The net effect is to compress the artificial spread the penny tick was creating in liquid names, and to reshuffle the rebate-capture and queue-value economics built on top of it.
The standing debate: tick size is a genuine policy trade-off. Too coarse and spreads are artificially wide (bad for investors, good for queue-priority strategies); too fine and quotes flicker, depth thins and queue priority loses meaning. There is no free optimum, which is why the parameter keeps being revisited. This is regulation tuning the microstructure dial in real time, and strategies adjusting in response. (See continuous-vs-auction and maker-taker for the downstream effects.)
Worked example
Tracing the causal chain with concrete, illustrative numbers (as of 2026; verify rule parameters against the SEC's current releases). The rule in action: stock XYZ shows a best offer of \$50.00 on Venue A and \$50.01 on Venue B. Under Rule 611, a marketable buy order cannot execute at \$50.01 on Venue B while \$50.00 is displayed and protected on Venue A. It must access the \$50.00 protected quote first. That is the trade-through prohibition that ties the venues together.
The NBBO and the SIP lag: suppose new selling hits Venue A and its offer drops to \$49.98 on the direct feed at time . The SIP, consolidating across venues, publishes the new \$49.98 NBBO at , where δ is the consolidation and transmission lag, historically on the order of hundreds of microseconds to low milliseconds, narrowed by SIP modernisation but non-zero. In the window , anyone pricing off the SIP still believes the best offer is \$50.00.
The latency-arb pick-off: a firm reading Venue A's direct feed knows the offer is \$49.98 at . It sees a resting bid on Venue B still priced for the old \$50.00 world (a quote now too high) and lifts that stale bid before the SIP-driven quoter updates at . The edge is exactly the SIP lag δ, captured many times a day across thousands of names, the structural reason colocation and direct feeds pay for themselves (latency arbitrage, colocation & FPGA).
Small per trade, the basis of rebate capture at scale. Every figure is illustrative; the SIP lag, access-fee cap and tick increments are specific, dated parameters, so reverify against the SEC's current rules before relying on them (as of 2026). This is educational only and not investment advice.