Block Trades And The Upstairs Market
How very large trades get negotiated privately between counterparties away from the public order book, and why this 'upstairs' route exists alongside ordinary exchange trading.
Prerequisites: Sourcing Dark Liquidity, The Grossman-Miller Model
Some orders are simply too large for the ordinary order book to absorb without doing serious damage to the price. A pension fund liquidating a $200 million position in one stock can't slice that into small enough child orders to avoid signaling its intent — at some point the sheer size overwhelms whatever an execution algorithm can quietly work into normal trading. For orders like this, markets have a separate channel entirely: the upstairs market, where large trades are negotiated privately rather than worked through the public book.
Trading away from the crowd
A block trade is a large trade — often defined above some minimum size threshold — negotiated directly between two counterparties, or between a client and a broker-dealer acting as principal, rather than executed as a stream of small orders on a lit exchange. This negotiation typically happens "upstairs," through a broker's sales and trading desk rather than the exchange's central limit order book, which is why the channel is called the upstairs market. A broker facilitating a block trade often takes the other side onto their own book and agrees to a price — sometimes at a discount to compensate for the risk of unwinding that position — or works to find a natural counterparty before the trade prints.
The core trade-off versus algorithmic execution is speed and certainty against information cost: a block trade transfers a huge position in one transaction, but requires disclosing size and intention to the broker, carrying its own leakage risk before the trade is finalized — a different flavor of the leakage problem algorithmic execution manages through anonymity instead.
Worked example: pricing a block versus an algorithmic schedule
A fund needs to sell 2 million shares of a mid-cap stock trading at $40, where average daily volume is only 3 million shares — this order is nearly two-thirds of a normal day's volume. Working it algorithmically over several days might achieve an average price near $39.85 (37 basis points below $40), but exposes the fund to multiple days of market risk. A broker instead offers to take the block upstairs at $39.70 (75 basis points below) in a single negotiated transaction, immediately transferring all position risk to the broker. The fund gives up roughly 38 extra basis points versus the algorithmic estimate, but eliminates days of market exposure and operational risk — worthwhile precisely when the fund's urgency is high enough to value that certainty over the algorithmic route's lower expected cost.
What this means in practice
Block trading and algorithmic execution aren't rivals so much as different tools for different situations, and large desks routinely use both: a trade list might send liquid, moderate-size names through algorithms while routing its most illiquid or oversized names to the upstairs market. Post-trade, most exchanges require block trades to be reported within a set window, so the market eventually sees the size — the upstairs route controls when and how that information becomes public, not whether it does.
Block trades move very large positions in a single negotiated transaction through the upstairs market rather than working them algorithmically through the public order book, trading a worse average price and upfront size disclosure for speed, certainty, and reduced exposure to ongoing market risk.
A rough rule of thumb desks use: as an order's size relative to average daily volume climbs past what a reasonable multi-day algorithmic schedule could absorb without excessive impact, the upstairs market becomes worth pricing out as an alternative.
Related concepts
Practice in interviews
Further reading
- Harris, Trading and Exchanges, ch. 18