In simple terms — The bid is the best displayed price from a buyer; the ask is the best displayed price from a seller. Their difference is the quoted spread. The price actually received may be better or worse than those quotes, so the cost of an execution cannot be inferred from the spread alone.
The bid-ask spread is the distance between the best buying price (bid) and the best selling price (ask or offer) available for an instrument in a specific market at a specific time. Its elementary definition is:
quoted spread = best ask − best bid
The calculation is meaningful only when both quotes belong to the same informational scope: the same instrument, currency, venue or set of venues, timestamp, and trading conditions. In fragmented markets, the best bid and best ask may come from different venues. In an over-the-counter market, they may depend on the dealer, requested quantity, and client relationship.
Bid, ask, and midpoint
The midpoint is the arithmetic mean of the best quotes:
midpoint = (best bid + best ask) / 2
If the bid is 99.98 and the ask is 100.02, the midpoint is 100.00 and the quoted spread is 0.04. Relative to the midpoint, the spread is 4 basis points:
relative spread in bps = spread / midpoint × 10,000
The normalization in bps here assumes a positive midpoint.
A market buy normally trades against the ask side; a market sell trades against the bid side. This does not mean that the full order quantity is available at the best quote. If size exceeds displayed depth, the order may cross several levels of the order book and receive an average price different from the top of book.
A market may also be locked, with bid equal to ask, or crossed, with bid above ask, because of market structure, fragmentation, feed delays, or temporary conditions. Such observations require explicit data-handling rules; they are not negative spreads that should automatically be treated as executable opportunities.
Three measures, three questions
The quoted spread is the distance between the best displayed quotes at a timestamp. The effective spread compares the execution price with a reference midpoint and doubles the one-way difference. The realized spread uses a midpoint observed after a specified interval measured from execution.
Let P be the execution price, m_ref the reference midpoint and m_τ the midpoint observed τ after execution:
E = 2q(P − m_ref); R_τ = 2q(P − m_τ)
The side belongs to the order being evaluated: q = +1 for a buy and q = −1 for a sell. Do not take the absolute value. The one-way price cost against the benchmark is E/2.
Realized spread helps examine subsequent price movement and possible adverse selection against the liquidity provider. It is not the trader’s profit, excludes commissions and hedging costs, and depends on the horizon.
SEC scope checked on 8 September 2026. Rule 605 requires monthly execution-quality reports on covered orders in NMS stocks from market centers and broker-dealers subject to the rule, rather than all orders or financial instruments. The 2024 amendments have a compliance date of 1 August 2026. The benchmark is the national best bid and offer (NBBO) midpoint at receipt or first executability as prescribed for the order type, not automatically at the fill. Realized-spread horizons are 50 ms, 1 s, 15 s, 1 min and 5 min after execution. The staff FAQs of 1 April 2026 clarify application; they are not the rule itself. These reporting requirements do not automatically extend to futures, FX, cryptoassets or OTC instruments.
Example: quoted spread and execution price
Using a 99.98 bid and 100.02 ask:
- the quoted spread is 0.04, or 4 bps relative to the 100.00 midpoint;
- a buy executed at 100.01 receives 0.01 of improvement over the ask;
- its effective spread is 2 × (100.01 − 100.00) = 0.02, or 2 bps;
- its one-way cost relative to the midpoint is 0.01, or 1 bp.
For a separate sale at 99.99, improvement over the bid is also 0.01 and effective spread is 2 × (100.00 − 99.99) = 0.02. Assume, in each scenario, a midpoint of 100.03 after 15 seconds:
| Order evaluated | Effective spread | Realized spread at 15 s |
|---|---|---|
| Buy at 100.01 | 0.02 | 2 × (100.01 − 100.03) = −0.04 |
| Sell at 99.99 | 0.02 | 2 × (100.03 − 99.99) = 0.08 |
Amounts are per unit before commissions. The later move is adverse to the provider who sold in the first case and favorable to the provider who bought in the second; it does not measure an actual closing trade.
The “half-spread” rule of thumb holds only when an order executes exactly at the best quote and the midpoint is the symmetric mean of bid and ask. An immediate buy at the ask followed by an immediate sell at the bid would instead cross the full quoted spread, before fees, market movement, and available-quantity effects. Price improvement, multi-level fills, and quote changes during routing alter the result.
This distinction also prevents double counting. If slippage is measured against the midpoint, crossing the spread is already included in that deviation. Adding the spread again as a separate cost would overstate the loss.
Why the spread changes
The spread compensates for several costs and risks involved in supplying liquidity:
- the risk that price moves while a quote remains exposed;
- the risk of trading against a better-informed order;
- inventory and hedging costs borne by a market maker;
- volatility, uncertainty, and the speed at which information changes;
- tick size and other market rules;
- competition among liquidity providers;
- depth, order size, and time of day;
- venue-specific costs, incentives, and fees.
Saying that a market maker “earns the spread” is therefore incomplete. The participant may capture a difference between purchases and sales but also bears inventory risk, adverse selection, hedging, technology, and fees. The realized spread may be much smaller than the quoted spread or even negative.
A tight spread often indicates good tightness—a low cost to trade a small quantity near the midpoint. It does not establish that the market has enough depth for a large order or that liquidity will recover quickly after a shock. Those are separate dimensions of liquidity.
Advanced: data limitations
The value shown by a platform may represent the latest quote received, only the selected venue, or an indicative price. A Level 1 feed does not show depth beyond the best quotes. Hidden liquidity, iceberg orders, and venues excluded from the feed remain outside the snapshot. The size associated with bid and ask may also change before an order arrives.
Comparing spreads across instruments or periods requires at least:
- stating whether spread is absolute, percentage, or basis points;
- identifying the venue and feed;
- defining the session, time window, and sampling frequency;
- specifying the quantity to which the measure refers;
- documenting the treatment of locked, crossed, or missing quotes;
- reporting means, medians, and percentiles rather than one snapshot alone.
Comparing a stock quoted in dollars with a futures contract quoted in index points is not useful when only absolute units are used. To estimate the spread's contribution to transaction costs for a specific size, quotes must be combined with depth, fills, and actual fees.
Common mistake — Treating the displayed spread as a certain total cost. It measures quotes within a defined scope; the outcome also depends on size, order type, latency, depth, venue, and chosen benchmark.
Sources
- SEC, Investor.gov — Bid Price/Ask Price
- SEC — Release No. 34-99679, Disclosure of Order Execution Information, 6 March 2024, §III.B.4 and Rules 600(b)(8)–(13), 605: spreads, benchmarks and statistical aggregation.
- SEC — Release No. 34-104147, Extension of Compliance Date, 30 September 2025, §I, p. 4 and note 11: collection from 1 August 2026, first reports due by the end of September.
- SEC staff — Frequently Asked Questions: Rule 605 of Regulation NMS, 1 April 2026, preamble and note 10: application from 1 August 2026 and five realized-spread horizons.
- BIS, Committee on the Global Financial System — Structural Aspects of Market Liquidity from a Financial Stability Perspective
Sources consulted on 8 September 2026. The single-execution examples are Cyclepedia teaching calculations, not a reproduction of a monthly Rule 605 report.