Why the US Has So Many Stock Exchanges

Why does the US have so many stock exchanges? A share of a listed company is identical wherever it changes hands, and federal rules oblige every broker and every rival venue to respect a new exchange's best quote from the day it opens. A venue that attracts even a sliver of volume is viable on those terms, and the companies that already run one exchange keep opening more.

What that looks like on the tape is fragmentation: one company's shares printing across a long list of venues in a single session, with one consolidated best bid and offer stitched together from all of their books. The panels below count that spread on a real session and show where the incentives sit.

Why the US has so many stock exchanges: four forces

1. Fungibility. A share is a claim on the company, not on the venue where you bought it. Every US equity trade clears through the same central counterparty, so a share bought on one exchange settles against a share sold on another. No venue needs the listing exchange's permission to host trading in a listed name.
2. Order protection and fair access. Regulation NMS, adopted by the SEC in 2005 and phased in through 2007, carries two rules that matter here. Rule 611, the order protection rule, bars an execution at a worse price than the best automated quote displayed by a protected venue: if a new exchange shows the best price, orders elsewhere have to honor it or route to it. Rule 610, the access rule, requires venues to make those quotes reachable on fair terms and caps the fee for taking them. That cap stood at $0.003 per share from 2005, and the SEC adopted a lower one in 2024. A new exchange opens with its quote already inside everyone else's routing logic.
3. Different economics on an identical product. Venues cannot differentiate the product. They differentiate the pricing instead. Under maker taker, the venue pays a rebate to the order that rests on the book and charges the order that removes it. Under taker maker, often called inverted pricing, the payment runs the other way. Other books charge a flat fee on both sides or sell connectivity and membership rather than per share economics.
4. One owner, several licenses. Each registered exchange is its own self-regulatory organization with its own rulebook, fee schedule and market identifier. Nothing prevents one company from operating several at once, which is how a single operator offers maker taker pricing on one book and inverted pricing on another.

MEMX and MIAX Pearl Equities both opened for equities trading in 2020, each starting from zero volume with a displayed quote that counted from its first session.

How many venues does one stock print on in a day?

Pick an ordinary session and count. The panel below splits Apple's share volume for June 10, 2026 by the venue that reported each print, with the venue codes carried on the tape matched to the exchange registry.

WITH registry AS
(
    SELECT
        toInt64(id)                   AS venue_id,
        argMax(name, _ingest_time)    AS venue_name,
        argMax(acronym, _ingest_time) AS venue_acronym,
        argMax(type, _ingest_time)    AS venue_type
    FROM global_markets.stocks_exchanges
    WHERE asset_class = 'stocks'
    GROUP BY id
)
SELECT
    multiIf(r.venue_acronym != '', r.venue_acronym,
            r.venue_name    != '', r.venue_name,
            concat('code ', toString(t.exchange)))                         AS venue,
    multiIf(lower(r.venue_type) = 'exchange', 'lit exchange',
            r.venue_type != '',               'off-exchange report',
            'unmapped code')                                               AS venue_kind,
    round(100 * sum(t.size) / sum(sum(t.size)) OVER (), 2)                 AS pct_of_shares,
    round(sum(t.size) / count(), 0)                                        AS avg_trade_size
FROM global_markets.stocks_trades AS t
LEFT JOIN registry AS r ON r.venue_id = toInt64(t.exchange)
WHERE t.ticker = 'AAPL'
  AND t.sip_timestamp >= toDateTime('2026-06-10 04:00:00')
  AND t.sip_timestamp <  toDateTime('2026-06-11 04:00:00')
GROUP BY venue, venue_kind
ORDER BY pct_of_shares DESC
LIMIT 25

FINRA Alternative Display Facility took the largest slice at 38.93% of the day's shares, at an average print size of 47 shares. The list runs 18 venues deep, and the smallest entry printed 0% of the total. A venue that far down is a rounding error on the day's volume, and its displayed quote is still protected on the same terms as the one at the top. That is what keeps a small book worth operating.

Rows tagged off-exchange report were never matched on an exchange book. They are prints sent to a FINRA trade reporting facility: wholesalers filling retail orders, and alternative trading systems, commonly called dark pools. Those trades reach the consolidated tape after they happen and display no quote beforehand. Everything tagged lit exchange published a quote first, and the rest of the market had to respect it. Who must respect what is the subject of what the NBBO is and how it gets built.

Why does one operator run several exchanges?

Every venue carries a Market Identifier Code, the four letter ISO code that names a market in trade reporting (XNYS for the New York Stock Exchange). Venues under common ownership share an operating MIC, so the ownership map falls straight out of the registry.

WITH registry AS
(
    SELECT
        id,
        argMax(mic, _ingest_time)           AS mic,
        argMax(operating_mic, _ingest_time) AS operating_mic,
        argMax(type, _ingest_time)          AS venue_type
    FROM global_markets.stocks_exchanges
    WHERE asset_class = 'stocks'
    GROUP BY id
)
SELECT
    operating_mic                                           AS operator,
    countDistinct(mic)                                      AS venue_medallions,
    arrayStringConcat(arraySort(groupUniqArray(mic)), ', ') AS mic_codes
FROM registry
WHERE lower(venue_type) = 'exchange'
  AND operating_mic != ''
  AND mic != ''
GROUP BY operating_mic
ORDER BY venue_medallions DESC, operator ASC

The registry sorts the US equity exchanges into 9 operating families. The largest, XNYS, carries 5 separate exchange licenses, known in industry shorthand as medallions (ARCX, XASE, XCHI, XCIS, XNYS).

Running several is a pricing strategy. One book pays a rebate for resting liquidity while a sibling book pays the side that takes it, and a broker's router sorts among them on the net cost of a fill rather than on the displayed price alone. The operator collects either way, and the fee schedules are public and mechanical: how maker taker fees and rebates work walks through the per share arithmetic.

What fragmentation costs, and what it buys

The cost starts with visibility. Displayed size for one stock sits scattered across many books, so the top of any single book understates what is available. The consolidated quote is the repair: a securities information processor collects the best bid and best offer from every protected venue and publishes the National Best Bid and Offer, one price pair stitched from many books. Firms that want the same picture sooner buy each venue's own feed and assemble the book themselves, the tradeoff covered in SIP versus direct exchange feeds. The consolidated quote also carries an inclusion rule, which is why odd lots do not set the NBBO even when they carry the best price on the tape.

The other cost is routing complexity. A broker choosing where to send an order weighs the fee schedule alongside the price, and the venue showing the best quote is not always the cheapest fill after rebates. What fragmentation buys is competition on access pricing and on matching engine design, plus depth that is real in aggregate even where any one venue looks thin.

Participation moves through the session as well. This panel buckets the same Apple day into half hours of Eastern clock time and counts the venues printing in each.

WITH registry AS
(
    SELECT
        toInt64(id)                AS venue_id,
        argMax(type, _ingest_time) AS venue_type
    FROM global_markets.stocks_exchanges
    WHERE asset_class = 'stocks'
    GROUP BY id
)
SELECT
    formatDateTime(
        toStartOfInterval(toTimeZone(t.sip_timestamp, 'America/New_York'), INTERVAL 30 MINUTE),
        '%H:%i')                                                                  AS et_time,
    countDistinct(t.exchange)                                                     AS venue_count,
    round(100 * sumIf(t.size, lower(r.venue_type) != 'exchange') / sum(t.size), 2) AS off_exchange_pct
FROM global_markets.stocks_trades AS t
LEFT JOIN registry AS r ON r.venue_id = toInt64(t.exchange)
WHERE t.ticker = 'AAPL'
  AND t.sip_timestamp >= toDateTime('2026-06-10 04:00:00')
  AND t.sip_timestamp <  toDateTime('2026-06-11 04:00:00')
GROUP BY et_time
HAVING sum(t.size) > 0
ORDER BY et_time ASC

The first bucket with prints, 04:00 ET, carried 7 venues, with 82.71% of its shares reported away from a lit book. The last, 19:30 ET, carried 7 venues at 59.8%. The shape between those two ends is the useful part, and the chart draws it.

Does every stock spread the same way?

No. This panel takes five household names on that same session and counts, for each, how many venues printed it and how much of its volume landed on the single busiest one.

WITH by_venue AS
(
    SELECT
        ticker,
        exchange,
        sum(size) AS shares
    FROM global_markets.stocks_trades
    WHERE ticker IN ('AAPL', 'MSFT', 'NVDA', 'SPY', 'KO')
      AND sip_timestamp >= toDateTime('2026-06-10 04:00:00')
      AND sip_timestamp <  toDateTime('2026-06-11 04:00:00')
    GROUP BY ticker, exchange
)
SELECT
    ticker,
    countDistinct(exchange)                                   AS venue_count,
    round(100 * max(shares) / sum(shares), 2)                 AS top_venue_pct,
    round(100 * (sum(shares) - max(shares)) / sum(shares), 2) AS rest_of_market_pct
FROM by_venue
GROUP BY ticker
ORDER BY venue_count DESC, ticker ASC

AAPL spread across the most venues, 18 of them, with 38.93% of its shares on the busiest. KO used the fewest, 17, at 43.02% on its busiest venue. A name with fewer prints has less volume to scatter, and the same rulebook covers both.

The lit venues in these panels are order driven books, matching resting orders against incoming ones by price and time. A wholesaler filling a retail order off exchange works closer to the quote driven model, a contrast drawn out in quote driven versus order driven markets.

FAQ

How many stock exchanges are there in the US?

The count is not fixed: exchanges register, merge and close. The registry behind this post sorts the current set of US equity exchanges into 9 operating families, the largest of which runs 5 licenses on its own.

Why do so many exchanges trade the same stock?

Shares are fungible and clear centrally, so a trade in one name can be matched anywhere. Order protection under Regulation NMS then requires the rest of the market to respect a venue's displayed best quote, which makes even a small new book usable from its first session.

What is market fragmentation?

Fragmentation describes one security trading on many venues at once, with no single book holding all of the liquidity. The consolidated quote reassembles a top of book view across every protected venue.

Are dark pools stock exchanges?

No. A dark pool is an alternative trading system: it matches orders without displaying a quote and reports its prints through a FINRA facility. Those prints land on the consolidated tape next to exchange trades, which is why off-exchange volume appears in the panels above.

Does fragmentation change what an ordinary investor pays?

Any effect shows up order by order rather than in general. The measurable version of the question compares a single fill against the NBBO at the moment it printed, which is an audit of one order rather than a property of the whole market.

Every panel here carries the SQL that produced it, one expander under each table. To count venues on a different session or a different name, ask for it in plain English on the Strasmore terminal.