Midpoint Liquidity Pools

Every displayed quote has two prices: the bid a buyer will pay and the offer a seller will accept, and the gap between them is pure cost to whoever crosses it. A retail investor market-buying 100 shares pays the offer; a seller gets only the bid. Midpoint liquidity pools exist to remove that toll for the subset of traders willing to give up guaranteed, immediate execution in exchange for a better price: instead of trading at the bid or the offer, both sides meet exactly in the middle.

The idea: splitting the difference

Think of a used-car sale where the buyer wants to pay $18,000 and the seller wants $20,000. A dealer forcing an instant sale takes one side or the other. A patient matchmaker instead waits for a buyer and seller who are both willing to settle at $19,000 — the midpoint — and neither party pays the dealer's spread. Midpoint pools, mostly dark pools and exchange midpoint order types, do exactly this for stocks: an order to buy and an order to sell are matched at the midpoint of the public best bid and offer (the NBBO) at the moment they cross, with no price improvement negotiation needed because the reference price is defined mechanically.

How it prices and what shows up there

If the NBBO is $50.00 bid / $50.04 offer, the midpoint is

$$\text{Midpoint} = \frac{50.00 + 50.04}{2} = 50.02 .$$

A buyer who would otherwise have paid $50.04 saves $0.02 per share, and a seller who would have received $50.00 also gains $0.02 per share — together capturing the full spread that would otherwise have gone to a market maker or been paid as a transaction cost. Midpoint orders are typically non-displayed: they don't add to visible depth on the exchange's order book, they simply rest and wait for an opposing midpoint order or an aggressive contra order to arrive.

The catch is that midpoint fills are a biased sample of order flow. Uninformed, patient traders — index rebalancers, portfolio managers with no urgency — are happy to wait at the midpoint. Informed traders racing a fast-moving price generally are not; they need certainty of execution now, so they cross the spread at the lit offer or bid instead. That means volume executed at the midpoint tends to be less toxic, on average, than volume executed by aggressively taking displayed liquidity — but "tends to be" is doing real work, because a large midpoint fill just before news still happens and still moves the estimate.

Worked example: comparing execution costs

A fund needs to buy 50,000 shares of a stock quoting $50.00 / $50.04.

• Lit market order: crosses the spread immediately, paying $50.04 × 50,000 = $2,502,000.
• Midpoint order, fully filled: pays $50.02 × 50,000 = $2,501,000 — a saving of $1,000, or 2 cents a share.

But the midpoint order isn't guaranteed to fill. If only 30,000 of the 50,000 shares find a midpoint counterparty over the course of the day, the desk must finish the remaining 20,000 shares in the lit market, likely at a worse price if the stock has drifted up while waiting. The midpoint saving is real but conditional on the stock staying still long enough to get filled — which is exactly the condition that makes the volume less informative about where the price is headed next.

Drag the mean and standard deviation above to see the intuition transferred to trade prices: midpoint fills cluster tightly around the NBBO midpoint by construction, while lit-market fills spread out to the bid and offer edges — a narrower, more central distribution is the visible signature of patient, less time-pressured flow.

What this means in practice

Traders route to midpoint pools specifically to avoid paying the spread, and execution algorithms routinely "ping" several midpoint venues simultaneously before resorting to a lit order. For researchers building order-flow signals, midpoint fills need to be handled separately from lit trades: they're a real, executable price but they don't reveal an aggressor's willingness to pay up, so a trade-classification rule built for lit markets will mislabel them.