Quote-Driven vs Order-Driven Markets

Quote-driven and order-driven markets are two answers to one question: when you send an order, who is on the other side of it? In an order-driven market any participant's resting order can be that other side, and a central book ranks and matches them. In a quote-driven market a dealer publishes the price at which it will buy and the price at which it will sell, and the dealer is your counterparty on every fill. A third structure sits alongside those two: the brokered or request-for-quote market, where an intermediary goes and finds the other side of one specific trade. The taxonomy is clean on paper. Real markets are hybrids, and which structure runs which part of a market decides how much you are allowed to see before you commit.

What is an order-driven market?

An order-driven market runs on a central limit order book, often shortened to CLOB. Participants post limit orders, a price and a size they are willing to trade at, the book stacks them by price, and an arriving order that crosses the best price on the other side executes against whatever rests there. Nobody is obliged to be the counterparty. What decides who gets filled first at a given price is the matching rule: US equity exchanges mostly run price-time priority, where the earliest order at the best price stands at the front of the queue, while several derivatives venues split incoming size pro rata across resting orders. Our guide to price-time priority vs pro rata matching works through both.

The structural point is what the book publishes. Across US exchanges the best bid and offer are consolidated into one public quote, the NBBO, and the price and the size standing behind it are visible to everyone at once. The panel below takes a mid-morning hour of the June 16, 2026 session and measures the two things a book shows at the touch: how wide the quote is, and how much size waits there.

WITH quotes AS
(
    SELECT
        ticker,
        toUInt64(sequence_number)           AS weight,
        toFloat64(bid_price)                AS bid,
        toFloat64(ask_price)                AS ask,
        toFloat64(bid_size + ask_size) / 2  AS touch_lots
    FROM global_markets.cache_stocks_quotes
    WHERE ticker IN ('SPY', 'AAPL', 'MSFT', 'PG', 'KO')
      AND sip_timestamp >= '2026-06-16 14:30:00'
      AND sip_timestamp <  '2026-06-16 15:30:00'
      AND bid_price > 0
      AND ask_price > bid_price
)
SELECT
    ticker                                                                                       AS symbol,
    round(quantileDeterministic(0.5)(10000 * (ask - bid) / ((ask + bid) / 2), weight), 2)        AS quoted_spread_bps,
    round(quantileDeterministic(0.5)(touch_lots, weight), 1)                                     AS touch_size_lots
FROM quotes
GROUP BY ticker
ORDER BY quoted_spread_bps

Sorted tightest first, SPY held a median quoted spread of 0.27 basis points (one basis point is a hundredth of a percent) with a median 180 round lots showing at the touch. At the other end, PG quoted 3.29 basis points on 350 lots. Nobody had to ask permission for either number. Measuring that width is its own topic, covered in what a bid-ask spread is.

What is a quote-driven market?

In a quote-driven market, also called a dealer market, there is no shared book to join. A dealer quotes a bid and an offer, and when you accept one of them the dealer takes the other side onto its own balance sheet. Corporate bonds trade this way. So does most of the foreign exchange market, and so did Nasdaq for its first two decades, when competing dealers posted quotes instead of orders meeting in one book.

Two things follow from the dealer being the counterparty. A dealer showing two sides has to buy while sellers keep arriving and sell while buyers keep arriving, whatever position it would prefer to hold. And you see the quote sent to you, with no view of the quote sent to the account next door, or of the size behind it.

What is a brokered or RFQ market?

When an instrument trades rarely, or the size on offer dwarfs anything a book displays, the search itself becomes the service. In a brokered market an intermediary works a network of contacts to locate the other side. The electronic version is the request for quote, or RFQ: you name the instrument and the exact size, ask several liquidity providers at once, collect their responses, and trade with one of them. Those quotes are private, they are good for the size you named, and they expire in seconds.

Why that structure exists is visible in one month of US equity data. A continuous book needs two-sided interest arriving all day. Most symbols never get it.

WITH
daily AS
(
    SELECT
        ticker,
        date,
        max(toFloat64(volume)) AS shares
    FROM global_markets.stocks_daily_aggs
    WHERE date >= '2026-05-01'
      AND date <  '2026-06-01'
      AND ticker NOT IN ('SPCX')
    GROUP BY ticker, date
),
per_symbol AS
(
    SELECT
        ticker,
        avg(shares) AS adv,
        sum(shares) AS month_shares
    FROM daily
    GROUP BY ticker
    HAVING adv > 0
),
buckets AS
(
    SELECT
        multiIf(adv <    10000, 'under 10k',
                adv <   100000, '10k to 100k',
                adv <  1000000, '100k to 1M',
                adv < 10000000, '1M to 10M',
                                'over 10M') AS adv_bucket,
        min(adv)                            AS bucket_floor,
        count()                             AS listings,
        sum(month_shares)                   AS bucket_shares
    FROM per_symbol
    GROUP BY adv_bucket
),
totals AS
(
    SELECT
        count()           AS all_listings,
        sum(month_shares) AS all_shares
    FROM per_symbol
)
SELECT
    b.adv_bucket                                        AS adv_bucket,
    b.listings                                          AS listing_count,
    round(100 * b.listings / t.all_listings, 1)          AS share_of_listings_pct,
    round(100 * b.bucket_shares / t.all_shares, 1)       AS share_of_volume_pct
FROM buckets AS b
CROSS JOIN totals AS t
ORDER BY b.bucket_floor

Across every US symbol with a daily bar in May 2026, the thinnest group, those averaging under 10k shares a day, made up 22.8% of all symbols and 0.1% of the month's share volume. The heaviest group, averaging over 10M shares a day, was 2.7% of symbols and 50.6% of the volume, spread across 346 names. A book is a good machine for that top group. For the long tail, someone has to go looking.

Size splits the same way inside a liquid name. Options make it plain: the exchanges run order books, and the largest trades are still negotiated privately at one price, then printed to the tape. Block trades follow the same path in equities.

WITH
session_trades AS
(
    SELECT toUInt32(size) AS contracts
    FROM global_markets.options_trades
    WHERE underlying_symbol = 'AAPL'
      AND sip_timestamp >= '2026-06-16 13:00:00'
      AND sip_timestamp <  '2026-06-16 21:00:00'
      AND size > 0
),
totals AS
(
    SELECT
        count()        AS all_trades,
        sum(contracts) AS all_contracts
    FROM session_trades
)
SELECT
    b.size_bucket                                            AS size_bucket,
    b.trades                                                 AS trade_count,
    round(100 * b.trades / t.all_trades, 1)                  AS share_of_trades_pct,
    round(100 * b.bucket_contracts / t.all_contracts, 1)     AS share_of_contracts_pct
FROM
(
    SELECT
        multiIf(contracts = 1,    '1 contract',
                contracts <= 10,  '2 to 10',
                contracts <= 100, '11 to 100',
                contracts <= 500, '101 to 500',
                                  'over 500') AS size_bucket,
        min(contracts)                        AS bucket_floor,
        count()                               AS trades,
        sum(contracts)                        AS bucket_contracts
    FROM session_trades
    GROUP BY size_bucket
) AS b
CROSS JOIN totals AS t
ORDER BY b.bucket_floor

On that session, AAPL option prints of 1 contract were 52.3% of all trades and 10.8% of the contracts that changed hands. Prints of over 500 contracts ran 0% of trades while carrying 3.5% of the contracts. The tape shows the print. The negotiation that produced it never appeared in any book.

Is any market purely quote-driven or order-driven?

No major market today is one or the other. US equities are order-driven at the exchanges, and a large share of retail marketable orders never touches an exchange book: a wholesaler fills them at or inside the NBBO and reports the print to a FINRA trade reporting facility. Institutional orders cross inside alternative trading systems, the subject of dark pool trading, while exchange fee schedules pay rebates to orders that do rest in the book, covered in maker-taker fees and rebates. The panel below measures how much of each session's volume in three household names prints away from the exchanges.

WITH
facility AS
(
    SELECT
        ticker,
        date,
        max(toFloat64(total_volume)) AS facility_shares
    FROM global_markets.stocks_short_volume
    WHERE date >= '2025-08-01'
      AND date <  '2026-08-01'
      AND ticker IN ('AAPL', 'KO', 'SPY')
    GROUP BY ticker, date
),
tape AS
(
    SELECT
        ticker,
        date,
        max(toFloat64(volume)) AS consolidated_shares
    FROM global_markets.stocks_daily_aggs
    WHERE date >= '2025-08-01'
      AND date <  '2026-08-01'
      AND ticker IN ('AAPL', 'KO', 'SPY')
    GROUP BY ticker, date
),
joined AS
(
    SELECT
        f.ticker              AS ticker,
        f.date                AS date,
        f.facility_shares     AS facility_shares,
        t.consolidated_shares AS consolidated_shares
    FROM facility AS f
    INNER JOIN tape AS t ON f.ticker = t.ticker AND f.date = t.date
)
SELECT
    formatDateTime(date, '%Y-%m') AS month,
    formatDateTime(date, '%b %Y') AS month_label,
    round(100 * sumIf(facility_shares, ticker = 'AAPL') / sumIf(consolidated_shares, ticker = 'AAPL'), 1) AS aapl_pct,
    round(100 * sumIf(facility_shares, ticker = 'KO')   / sumIf(consolidated_shares, ticker = 'KO'), 1)   AS ko_pct,
    round(100 * sumIf(facility_shares, ticker = 'SPY')  / sumIf(consolidated_shares, ticker = 'SPY'), 1)  AS spy_pct
FROM joined
GROUP BY month, month_label
HAVING countIf(ticker = 'AAPL') > 0
   AND countIf(ticker = 'KO') > 0
   AND countIf(ticker = 'SPY') > 0
ORDER BY month

In Jul 2026, 33.9% of AAPL's consolidated share volume was reported through those off-exchange facilities. KO measured 31.9% and SPY 32.9% over the same month. The level holds across the window: AAPL sat at 37.3% back in Aug 2025. Anyone watching only the exchange book is watching part of the market, and the part out of view is largely dealer flow.

The blending runs the other way too. Corporate bonds remain dealer markets with an electronic RFQ layer on top, reporting trades to a public feed after the fact rather than before. ETF blocks get negotiated with a market maker who then works the creation and redemption mechanism against the underlying basket.

What the spread pays for

Textbook taxonomies usually skip the part that makes dealing hard: inventory risk. A dealer quoting both sides accumulates positions it never sought, buying through a morning of sellers and ending the session long a book it now has to hedge or work out of. Adverse selection sits on top of that. Some of the flow arriving at a dealer's quote knows something the dealer does not, and those are the trades that tend to move against it immediately after the fill. The spread is payment for carrying both risks, alongside the capital and technology behind the quote. How market makers make money lays out that economics in full.

What each structure lets you see

Transparency follows from the structure. An order book shows the resting queue: prices, sizes, and how quickly the book thins out away from the touch, which is what traders mean by depth. A dealer market shows you the quote you were sent. There is no depth to inspect, and a competing dealer's price stays invisible until you ask for it. An RFQ shows the handful of responses you solicited, and nothing about the ones you skipped. Before trading an unfamiliar instrument, the first useful question is which of those pictures you are looking at.

FAQ

What is the difference between a quote-driven and an order-driven market?

In an order-driven market, orders from any participant rest in a central book and trade against each other under a published matching rule. In a quote-driven market, a dealer posts a two-sided quote, stands as the counterparty to every trade, and carries the resulting position itself.

Is the US stock market order-driven or quote-driven?

Both. The exchanges run order books under price-time priority, while a large share of retail marketable orders is filled off-exchange by wholesalers who report the print to a FINRA facility. The monthly panel above puts a measured number on that off-exchange share.

Are bond markets order-driven?

Corporate and municipal bonds trade mainly in dealer markets, with electronic request-for-quote platforms layered on top. A single bond issue trades far less often than a listed stock, which is the setting where a search-based structure fits better than a continuous book.

What does RFQ mean in trading?

RFQ is short for request for quote. A participant names the instrument and size, asks several liquidity providers to respond with a price, then chooses among the responses. It is standard for bonds and for large ETF or options trades.

Every panel here carries the exact SQL beneath it, so any figure traces back to the query that produced it. The same questions can be asked of any symbol in plain English on the Strasmore terminal.