Odd Lots: Why Under 100 Shares Fills Differently

You bought 7 shares of a $600 stock on a Monday morning. The confirmation came back at $599.9950, a price that appeared on neither side of the quote you were staring at, and the print never showed up on the tape in front of you. Nothing broke. You placed an odd lot, and odd lots live outside the quote system that publishes the national best bid and offer.

That single fact — your order cannot be part of the public quote — explains nearly everything readers find strange about small trades. It changes how the order is advertised, who sees it, and whether anyone routes toward it. It does not change what the spread costs you as a percentage of what you spent, which is the part most people get backwards.

The definition is smaller than you think

Fewer than 100 shares of a stock whose standard trading unit is 100 shares. That is an odd lot, and that is the whole rule.

One hundred shares is a round lot. Two hundred, seven hundred, three thousand — all round. Anything in between carries a remainder: 137 shares is a mixed lot, one round lot riding along with a 37-share odd piece, and only the 100-share portion is eligible to appear in the public quote. Your 7 shares of the $600 stock are pure odd lot, no round portion at all.

The 100-share unit is a convention, not physics. Quote infrastructure was built around a standard unit and the market kept using it. FINRA’s plain-language overview of how stocks trade covers the surrounding mechanics for readers who want the wider frame.

Why your 7 shares can’t set the best bid

Say the $600 stock is quoted $599.98 bid, $600.02 offer, and instead of buying at the offer you post a limit to buy 7 shares at $599.99. You are now bidding a penny more than anyone else in the country. The national best bid stays $599.98.

Your order is real. It rests on the book at whichever exchange your broker sent it to, and it can trade. What it cannot do is announce itself. The protected quote — the one every broker’s routing logic reads before deciding where to send flow — only accepts round lots, so your improved price is invisible to everyone who isn’t already looking at that specific venue’s depth feed.

A round lot at $599.99 gets broadcast nationally the instant it rests. Your odd lot is a note left on one desk. Someone has to walk past that desk.

This is the structural difference in full, and it produces a consequence worth being precise about: the penalty is a lower chance of a passive fill, not a worse price when you do fill. Suppose nobody ever hits your $599.99 bid and after twenty minutes you give up and buy at the $600.02 offer. You paid three cents more than your limit. Twenty-one cents in total, on a $4,200 trade. A round-lot buyer who did the same thing paid the same three cents per share.

What the spread actually costs

Here is the arithmetic on that $600 fill, and on eleven other price levels, so the shape of the answer is visible.

Every figure below is my own calculation. The assumptions, stated plainly: you buy at the offer and would have to sell at the bid, so half the spread is the one-way cost measured against the midpoint. Spread widths are illustrative, picked as plausible for each price band, not pulled from a live feed. Commission is zero, standard at US retail brokers on stocks. No price improvement is assumed, which makes these figures a ceiling rather than an expectation.

Read the last column only. It does not climb or fall cleanly with price, because it isn’t really about price — it’s about spread width relative to price. Your $600 stock lands at 0.0008% because a heavily traded name can hold a penny spread even at that level, so three and a half cents of cost sits on top of $4,200. The $1,000 and $2,500 rows are worse in percentage terms despite being more expensive stocks, because thin high-priced names quote wider.

The genuinely bad row is $2.00. Three cents on a two-dollar stock is 0.75% one way, 1.5% round trip, and that has nothing to do with order size — a 5,000-share buyer pays the identical percentage. Compare it to your $600 fill and the ratio is roughly 450 to one. The stock’s price and liquidity set that gap, not your share count.

The $4.00 row worked line by line, since one worked example beats twelve asserted ones. Spread of three cents means a bid of $3.985 and an offer of $4.015, midpoint $4.000. You buy at $4.015. Cost per share is $4.015 − $4.000 = $0.015. Seven shares brings that to $0.105. Trade value at the midpoint is $28.00. Divide: $0.105 ÷ $28.00 = 0.00375, or 0.375%.

Size changes the dollars, not the percentage

Return to your $600 purchase and imagine you’d bought 70 shares instead of 7. The spread cost goes from $0.035 to $0.35 — ten times more money, on ten times more stock. Percentage unchanged.

That relationship holds everywhere, and it’s the single most useful thing in this article:

Every column is flat top to bottom. Buying 3 shares is not proportionally more expensive than buying 99, or 300. In this model the percentage is determined by the spread relative to the price and by nothing else at all.

What size does change is whether you fill at your price and whether a market maker finds your order worth improving.

The tape used to leave you out entirely

Odd-lot executions were not printed to the consolidated tape until late 2013. Before that, a trade like yours happened, cleared, and settled in the ordinary way, but it never appeared in the public trade stream and never counted toward reported consolidated volume. Anyone measuring activity from tape volume was working from a picture with a piece cut out of it.

Since December 2013 the prints show up and the volume counts. The quote restriction survived the change, which is why your 7-share bid still can’t set the NBBO.

There’s a scale effect buried in that history worth spelling out. Odd lots move volume statistics only slightly and trade-count statistics enormously, because it takes an enormous number of them to add up. At an average size of 5 shares, it takes 2,000 prints to equal a single 10,000-share institutional block, and 200,000 prints to reach a million shares. At 25 shares average, those figures are 400 and 40,000. So when you see a headline “number of trades” figure on an expensive stock, understand what’s inside it: a great many people buying 2 and 5 and 7 shares, exactly like you did.

Settlement and fees

Your Monday buy settles Tuesday. So does a 7,000-share buy placed the same morning by someone else.

The SEC’s T+1 settlement rule cut the standard cycle from two business days to one, effective 28 May 2024, and odd lots ride the same schedule as everything else. There is no small-order exception, no delayed handling, no separate queue.

Fees are slightly more interesting. Exchanges do sometimes price odd-lot liquidity differently inside their maker-taker schedules, but at a zero-commission US retail brokerage none of that reaches your statement. What reaches you is the fill price and, on sales, small regulatory fees assessed on dollar proceeds rather than share count. That distinction cuts in your favour when you’re small: a per-dollar sale fee on $28 of proceeds is a rounding error, while the same rate on $28,000 is a real number. It does not scale against you for being small.

The saving-up mistake

“Small orders are expensive, so I’ll wait until I can buy 100 shares.” This is the costliest idea in the whole subject, and the flat table above disposes of it.

One share and 100 shares of a $200 stock carry the identical 0.005% spread cost. To avoid that cost on the round lot you’d need to accumulate $20,000, and 0.005% of $20,000 is one dollar. Meanwhile the cash sits uninvested for however many months it takes to get there. Idle cash waiting for a round lot has an opportunity cost measured in hundreds of basis points against a saving of half of one. The waiting is the expensive part.

When these figures are wrong

Every number above assumes a stable spread of a penny or slightly wider and a fill at the quoted offer. Several ordinary situations break that assumption hard.

Thin stocks. A lightly traded name might quote $18.40 bid, $18.85 offer — a 45-cent spread, 2.4% one way. Nothing in my table goes near that. Look at the real quote before assuming a penny.

The open and the close. Spreads at 9:30:05 run several times wider than at 11:00, and a market order for 7 shares can fill somewhere you did not expect. The percentage cost is indifferent to the fact that you only bought 7 shares.

Wide ticks on expensive stocks. My $2,500 row uses a 50-cent spread. If the actual spread is $2.00, the half-spread on 7 shares is $7.00 rather than $1.75 — multiply my figure by four.

Fractional shares. A 0.4-share purchase is not an odd lot in the exchange sense at all. It is an internal broker allocation, priced according to that broker’s own execution and rounding policy, which sits entirely outside this model.

Halts and auctions. During a volatility halt and the reopening auction there is no continuous quote, so a half-spread model has nothing to attach itself to.

What this does not tell you

A spread model is not an execution-quality model, and the gap between them is where most of the real variation lives.

It does not capture price improvement. A large share of retail marketable orders fill between the bid and the offer rather than at it, and when that happens your true cost falls below every figure in the tables, sometimes by half.

It does not capture adverse selection. If you buy at the offer and the offer drops two cents a second later, you are worse off than the half-spread suggests — but that is market movement, not a transaction cost, and I’ve made no attempt to model it.

It does not distinguish between brokers. Two firms routing identical 7-share orders can produce measurably different average fills across hundreds of orders. A single fill tells you nothing about that; you’d need a large sample and the execution-quality reports to go with it.

It uses invented spreads. Every spread width in every table is one I judged plausible for that price band. None came from a live feed. Substitute the actual spread on the stock you care about and the arithmetic still holds, but the outputs move.

And it says nothing about whether any given purchase is sensible. Transaction cost is one input among many.

FAQ

Is 50 shares an odd lot?

Yes. Anything from 1 to 99 shares qualifies. Fifty shares of a $50 stock at a one-cent spread carries a one-way spread cost of $0.25 on $2,500 of value, which is 0.010% — the same percentage you’d pay buying 5 shares or 500.

Do odd lot orders take longer to fill?

Marketable odd lots generally fill right away, because a market maker will take the other side. Passive odd-lot limit orders can sit longer, since they never appear in the national quote and so attract less routed flow. That is a fill-probability effect, not a queue-position penalty — price-then-time priority at a venue treats your 7 shares the same as anyone’s 700.

Are odd lot trades reported to the tape?

Yes, since December 2013. Before that they executed and settled normally but never printed publicly, which meant reported consolidated volume understated real trading. They still cannot set the protected national best bid or offer.

How much does it cost to buy 3 shares instead of 100?

The same percentage, exactly. On a $200 stock with a two-cent spread, 3 shares costs $0.03 one way against $600 of value — 0.005%. One hundred shares costs $1.00 against $20,000, also 0.005%. Dollars scale with size; the percentage does not.

Do odd lots settle on a different schedule?

No. Every US equity trade settles one business day after trade date under the T+1 standard effective 28 May 2024. A 7-share trade on Monday and a 7,000-share trade on Monday both settle Tuesday.

Why did my odd lot fill at a price I never saw quoted?

Almost certainly price improvement — you filled between the displayed bid and offer instead of at the offer. On a stock quoted $99.98 by $100.02, a buy filling at $100.00 beats the offer by two cents, which is $0.14 across 7 shares. Small in dollars, but it explains why the printed price matches neither side of the quote you were watching.

What to check against your own orders

Pull the live spread on the specific stock before sending a market order, particularly outside the middle hours of the session. The penny spread in most of my rows is common on liquid names and rare on thin ones.

Compare each fill price to the bid and offer at the moment of execution. Fills landing consistently right at the offer on buys mean you are getting no price improvement, and that is worth knowing about your broker.

Then ask what a round lot would have changed. In the percentage tables, nothing. If your own numbers disagree, the difference is coming from spread width or timing — not from the odd-lot label itself.

This article is general information, not financial advice. See our disclaimer.

More in Market Mechanics · all topics · calculators · how this was checked