Liquidity and spreads

A market shows two prices. The bid is the highest price anyone is currently willing to pay; the ask is the lowest price anyone is currently willing to accept. You buy at the ask and sell at the bid, so the two are the prices available to you in each direction, and the distance between them is the spread.

With a bid of 60¢ and an ask of 64¢, the spread is four points and the midpoint is 62¢. The midpoint is the number to treat as the market’s probability, because it is the only one that does not include the cost of transacting immediately. Neither the bid nor the ask is a probability estimate; each is a probability estimate plus or minus the price of being in a hurry.

Everything else on the screen is secondary to those two numbers. The last traded price is history and may sit anywhere relative to the current book. A single “price” field is a derived convenience. If you want to know what a trade will cost you, the ask is the only honest answer.

Buy at 64¢ and sell straight back at 60¢ and you are down four cents per contract without the market having moved at all. On the 64¢ you committed, that is a 6.25% round-trip cost — before any explicit fee. Half of it, two points, is what you pay relative to the midpoint on each single crossing.

This is the cost people most reliably forget, because it never appears on a statement as a fee. It shows up as having bought slightly higher and sold slightly lower than the price you would have quoted as “the market”. On a strategy that trades often it compounds much faster than any fee schedule.

The alternative is not to cross. Posting a limit order at or inside the bid means you are the one being quoted: you might get filled at 61¢ instead of paying 64¢, and you might not get filled at all. That trade-off — a better price against the risk of no position — is the entire choice between taking and making, and it is the same choice on every order book in any market.

The spread tells you the price of the first contract. Depth tells you the price of the hundredth. A book showing an ask of 62¢ with thirty contracts behind it, then 64¢ with fifty, then 67¢ with a hundred, is a book where a hundred-contract order does not cost 62¢ a contract.

Work it through: thirty at 62¢ is $18.60, fifty at 64¢ is $32.00, and the remaining twenty at 67¢ is $13.40. Total $64.00 for a hundred contracts, an average of 64¢ — two points worse than the price displayed. That difference is slippage, and it is a function of your size against the book rather than of anything the venue charges.

So a one-point spread with almost nothing behind it is a worse market than a three-point spread with real size behind it, for anybody trading more than a token amount. Spread is what a tourist sees; depth is what determines the fill.

On a market with a bid of 55¢ and an ask of 65¢, the midpoint is 60¢ and nobody has agreed to anything. The nearest buyer and the nearest seller are ten points apart. Calling 60¢ “the market’s probability” describes a transaction that does not exist.

Our own pipeline draws that line at eight points: wider than that, a quote is flagged illiquid and is not used in any cross-platform comparison. The price is still displayed, with its spread alongside it, because withholding it entirely would be its own distortion — but it does not get to participate in a claim about where two venues disagree. Eight is a judgement; the principle that a very wide midpoint is not a consensus is not.

The same reasoning explains why an apparent gap between a liquid venue and an illiquid one is so often nothing. If one side of the comparison carries a twelve-point spread, six of those points are consumed just by getting in, and a gap of five points against a cost of six is not an opportunity. Cross-venue differences only become interesting when both sides are tight.