Block Trade Discounts and Bought Deals

A block trade discount is the gap between the price a large seller accepts and the price on the screen, quoted either as a percentage off the last sale or off a reference VWAP. The seller is buying immediacy: one price for the entire position, agreed with one counterparty before the market reopens. The bank on the other side owns the stock before it has found a single buyer, and the discount is the fee for carrying it.

What is a block trade discount, exactly?

A block is a single negotiated trade far larger than the visible order book can absorb, arranged away from the continuous market and reported afterwards. Our primer on what counts as a block trade covers the size conventions and the venues.

The discount is always struck against a stated reference, and the reference matters as much as the percentage.

• Last sale: the discount comes off the closing print, or off the last trade before the deal was agreed. Six percent off a $50.00 close is $47.00.
• Reference VWAP: the discount comes off the volume weighted average price over a named window, often the prior five sessions. A single odd closing print has far less pull on the number.

Two desks quoting the same six point discount can mean two different cash prices when one is working off the close and the other off a five day VWAP. The reference window belongs in the term sheet, right next to the percentage.

Bought deal or accelerated bookbuild: who owns the stock overnight?

These are the two mechanisms readers conflate most often. The difference is whose balance sheet holds the shares between the handshake and the resale.

In a bought deal, also called a block purchase or a risk trade, the bank commits its own capital at a struck price, usually after the close, and becomes the owner of the shares. If the resale clears below the struck price, the loss belongs to the bank. It is long an outsized position in one name with no guaranteed buyer at any price.

In an agented deal, also called an accelerated bookbuild, the bank commits no capital. It canvasses institutions for a few hours, sets the price once the book is covered, and earns a fee on what it places. The seller keeps the price risk and can end up with a smaller deal, or no deal.

A bought deal looks more expensive on paper. What the seller gets for the extra points is a known cash number before the market reopens. In an agented deal the proceeds are discovered rather than guaranteed.

What determines the size of the block trade discount?

Four things do most of the work. Free float, meaning the shares actually available to trade rather than locked up with insiders and strategic holders. Average daily volume, which sets how long an unwind takes; average daily volume explains how that figure is measured. Borrow availability, which decides whether the position can be hedged at all. And how informed the seller looks: a founder selling a slice reads differently from an index fund rebalancing, and a book prices that difference.

The most portable of the four is volume. The panel below prices a hypothetical $250 million sale against the real dollar turnover of seven large US listings over the summer of 2026.

SELECT
    ticker,
    round(avg(toFloat64(close) * volume) / 1e6, 1)         AS daily_dollar_volume_millions,
    round(100 * 250e6 / avg(toFloat64(close) * volume), 1) AS block_share_pct
FROM
(
    SELECT
        ticker,
        date,
        max(close)  AS close,
        max(volume) AS volume
    FROM global_markets.stocks_daily_aggs
    WHERE ticker IN ('NVDA', 'AAPL', 'MSFT', 'WMT', 'PG', 'KO', 'F')
      AND date >= '2026-06-01'
      AND date <  '2026-09-01'
    GROUP BY ticker, date
)
GROUP BY ticker
ORDER BY block_share_pct ASC

For NVDA, $250 million is 0.9% of a single session's dollar volume, against 29248.5 million dollars of turnover a day. At the far end of the panel the same cash is 34.2% of a session in F. Identical size, a completely different unwind, and the discount tracks the unwind rather than the dollar amount.

Borrow sits alongside volume. A buyer who wants the shares without the market exposure has to short something against them, and the names where that is already crowded show up in the short interest data.

SELECT
    ticker,
    round(argMax(toFloat64(short_interest), settlement_date) / 1e6, 1) AS short_interest_millions,
    round(argMax(toFloat64(days_to_cover), settlement_date), 2)        AS days_to_cover_ratio,
    formatDateTime(max(settlement_date), '%b %d, %Y')                  AS reported_for
FROM global_markets.stocks_short_interest
WHERE ticker IN ('NVDA', 'AAPL', 'MSFT', 'WMT', 'PG', 'KO', 'F')
  AND settlement_date >= '2026-07-01'
GROUP BY ticker
ORDER BY days_to_cover_ratio DESC

As of Sep 15, 2026, WMT carried the panel's highest days to cover at 3.89, on short interest of 80.6 million shares. Days to cover is short interest divided by average daily volume: the number of ordinary sessions a full unwind of the short side would occupy. Where that figure sits high, borrowable shares are scarcer and the hedge costs the buyer more, which turns up in the price that buyer will bid for a block.

The arithmetic of the underwriter's overnight risk

Put numbers on the risk trade. Take a hypothetical deal: $250 million struck at a six percent discount leaves the bank a cushion of $15 million. That cushion is the entire buffer between the struck price and a realised loss, and the bank wears every price move from the moment it signs.

The first move it wears is the overnight gap, close to next open. The panel measures that for the same names over the twelve months to October 2026, as an absolute percentage so direction does not cancel out.

SELECT
    ticker,
    round(avg(abs(gap_pct)), 2) AS avg_overnight_move_pct,
    round(max(abs(gap_pct)), 2) AS largest_overnight_move_pct
FROM
(
    SELECT
        ticker,
        100 * (toFloat64(open) / prev_close - 1) AS gap_pct
    FROM
    (
        SELECT
            ticker,
            date,
            open,
            lagInFrame(toFloat64(close)) OVER (PARTITION BY ticker ORDER BY date ASC
                ROWS BETWEEN 1 PRECEDING AND CURRENT ROW) AS prev_close
        FROM
        (
            SELECT
                ticker,
                date,
                max(open)  AS open,
                max(close) AS close
            FROM global_markets.stocks_daily_aggs
            WHERE ticker IN ('NVDA', 'AAPL', 'MSFT', 'WMT', 'PG', 'KO', 'F')
              AND date >= '2025-10-01'
              AND date <  '2026-10-01'
            GROUP BY ticker, date
        )
    )
    WHERE prev_close > 0
)
GROUP BY ticker
ORDER BY avg_overnight_move_pct DESC

NVDA moved an average of 1.09% between one close and the next open, with a widest single gap of 6.3% over the window. The calmest name in the panel, KO, averaged 0.46%. Six points of cushion covers the average overnight move in any of these names several times over.

The gap is the easy part. The harder part is impact: what the bank's own selling does to the price while it works out of a position that size. A hedge offsets the market move and leaves the impact, and the discount is where a desk charges for both.

Registered stock, or Rule 144?

Whether the shares are registered changes who is allowed to buy them, and the eligible buyer pool sets the clearing price. Freely tradable registered stock can be resold to anyone, which puts the whole institutional market in the book. Restricted or control stock sold under Rule 144 carries a holding period, volume caps, manner of sale conditions and a filing obligation above certain sizes, and the deal gets placed with a narrower set of buyers, often privately, at a wider discount for the illiquidity the buyer takes on. The mechanics are in Rule 144 and restricted stock sales.

The same logic runs in reverse when the issuer is the buyer rather than the seller. See how companies execute buybacks for the rules a company works under when it takes size out of its own float.

How the block print shows up on the tape

A block agreed overnight still has to print, and it prints as a trade like any other: one price, one size, reported inside the window the rules allow. How block trades print on the tape goes through the condition codes and the reporting delays. To see why a single block print stands out so sharply, look at how an ordinary session is actually distributed.

SELECT
    b.print_size                                 AS print_size,
    b.print_count                                AS print_count,
    round(100 * b.tier_shares / t.day_shares, 2) AS share_of_volume_pct
FROM
(
    SELECT
        multiIf(size <   100, 'under 100 shares',
                size <  1000, '100 to 999 shares',
                size < 10000, '1,000 to 9,999 shares',
                              '10,000 shares and up') AS print_size,
        multiIf(size < 100, 1, size < 1000, 2, size < 10000, 3, 4) AS tier_rank,
        count()                                       AS print_count,
        toFloat64(sum(size))                          AS tier_shares
    FROM global_markets.stocks_trades
    WHERE ticker = 'AAPL'
      AND sip_timestamp >= toDateTime('2026-06-17 08:00:00', 'UTC')
      AND sip_timestamp <  toDateTime('2026-06-18 02:00:00', 'UTC')
    GROUP BY print_size, tier_rank
) AS b
CROSS JOIN
(
    SELECT toFloat64(sum(size)) AS day_shares
    FROM global_markets.stocks_trades
    WHERE ticker = 'AAPL'
      AND sip_timestamp >= toDateTime('2026-06-17 08:00:00', 'UTC')
      AND sip_timestamp <  toDateTime('2026-06-18 02:00:00', 'UTC')
) AS t
ORDER BY b.tier_rank

On 17 June 2026, prints of 10,000 shares and up made up 47.92% of the AAPL shares that changed hands, across 77 separate prints. The smallest tier, prints under 100 shares, accounted for 22.84% of the volume across 643334 trades. Ordinary trading is assembled out of small pieces. A block arrives as one line.

Why the next day's open often references the deal price

Once a deal prices, the struck price is public, and every institution that saw the book knows the level at which a very large holder transacted. The next session's bids tend to cluster around that level rather than around the prior close, and the opening auction is where the clustering becomes visible as a single price. The matching logic is the one described in the closing auction, run at the other end of the day.

The same anchoring appears on a first day of trading, where an offer price set the night before frames the first print. How the IPO opening price is set walks through that auction. In both cases the deal price is the most recent level at which real size cleared, which is the level the next auction has to clear against.

FAQ

Why do block trades price below the screen price?

The screen price is the price for a normal sized order. A seller with size the book cannot absorb is paying for immediacy, and the discount is that fee. The buyer, in a bought deal usually a bank, takes ownership of shares it then has to work out of over days, and the discount prices that inventory risk.

What is the difference between a bought deal and an accelerated bookbuild?

In a bought deal the bank buys the block with its own capital at a struck price and carries the risk of the resale. In an accelerated bookbuild the bank solicits demand first and prices once the book is covered, and the seller keeps the price risk while the bank earns a placement fee.

How big is a block trade discount?

It is negotiated deal by deal and quoted against a named reference, either the last sale or a reference VWAP, and there is no single number. The two measures that travel across deals are the ones in the panels above: the block as a share of one session's dollar volume, and the stock's own overnight move.

Does a block trade discount mean the stock is overvalued?

No. A discount is a statement about one holder's size and the time that size takes to absorb, not a statement about the company. A thinly traded name can print a wide discount in a week when nothing about the business changed.

Every panel here carries the exact SQL beneath it, so the counting is open to inspection. To size a block against a given name's own volume and overnight behaviour, ask the question in plain English on the Strasmore terminal.