What liquidity is

The same order, twice 

A price on a screen is an offer to deal a quantity. It is natural to read it as one number that applies to any amount, because that is how a price behaves in a shop, where the label holds whether one item is bought or twenty. A market price holds only for the size resting behind it. Once that size is taken, the next price is whatever the next willing party is quoting.

So one instruction can produce two different outcomes on two afternoons. In the first, more size was resting near the price than the order needed, and the fill landed within a fraction of the level requested. In the second, the resting size ran out part way through, and the remainder reached further out to find the rest of what it needed. The instrument was the same. The conditions were not.

Size available at a price, not volume traded 

The word is most often misread as a synonym for volume. Volume records how much changed hands over a period that has already finished. Liquidity is a statement about capacity now: how much can be dealt, at or near the price currently shown, before the price moves. The two are related, because an instrument that trades heavily usually attracts many parties willing to quote it, but they routinely disagree. Volume is often at its highest in the minutes when liquidity is thinnest, because a burst of transactions and a shortage of resting interest are two descriptions of the same disorderly moment.

A working definition, then. Liquidity is the size that can be dealt within a tolerance of the current price, together with the speed at which that size is replaced once taken. Two quantities are hiding in that sentence, and a third sits beside them, which is why practitioners rarely use a single number.

Three quantities, not one number 

• Tightness. The distance between the best price to buy and the best price to sell. It is the most visible measure and the only one published continuously, which is why it stands in for the others. It describes the top of the market and nothing behind it.
• Depth. The quantity resting at the best price and at each level behind it. A market can be tight and shallow at once: a narrow quote, for a small size, with very little standing behind it.
• Resilience. How quickly the size an order consumed is replaced. Two markets can look identical in the instant before an order arrives and behave completely differently in the second after it, because one refilled and the other did not.

No published figure combines the three, and any figure that is published describes one of them. That matters to anyone comparing instruments or venues on a quoted spread alone. The tightest quote in a set is not necessarily the one that will absorb a given size, and a typical spread figure says nothing about the depth or the resilience behind it.

The narrower quote produced the worse average price, and nothing unusual happened to make it do so. That is the argument for treating tightness and depth as separate facts rather than one property with two names.

What it costs when it is not there 

Nobody charges for liquidity. It is the condition that decides the size of two costs a reader has already met, and it moves both at once. A quote widens when the parties standing behind it become less willing to hold the other side of a position, which is the same event as liquidity thinning. An order larger than the size resting at the best level is filled across several levels, so its average price sits further from the level requested, which is the same event seen from the other end. The two are not independent and do not arrive one at a time.

The instrument, the size and the instruction were identical in both conditions. The only variable was how much rested behind the price, and it accounts for the entire difference in what the order cost. That is the sense in which liquidity is expensive when it is missing, though free when it is there.

When it reliably thins 

Thin conditions are not random, and several recur on a schedule published in advance. The mechanism is the same in each case. The parties who quote a price are either unwilling to hold inventory they cannot value or unable to lay it off elsewhere, so they widen their quotes, show less size, or step away until the uncertainty passes.

• The turn of the trading day. Around the daily rollover, when the accounting day changes and financing is applied, quoting parties in many instruments reduce or withdraw size for a short window, and quotes widen before normalising again.
• The hours between the major centres. Activity passes from Asian to European to North American hours, and in the stretch where one region has finished and the next has not begun, fewer parties are quoting. The same order meets less resting interest than it would at the busiest hour.
• The moments around a scheduled release. In the seconds before an announced number or a policy decision, resting interest is pulled, because the price about to exist is unknown to everybody. It returns afterwards, though not instantly and not always to the same depth.
• Public holidays and the days around them. A market that stays open while its principal centre is closed runs on a fraction of its usual participation, as do many instruments across the final week of December.
• Unscheduled news. There is no calendar entry for this one, which is what distinguishes it. Resting interest disappears faster than it can be replaced, and the effect can last minutes or hours.
• The edges of the week. Instruments that trade around the clock during the week still stop and restart, and the first minutes after a restart carry everything that happened while the market was shut.

Why it differs across the five asset classes 

Liquidity is decided by how many parties are willing to quote an instrument and by what they can do with the position once they hold it. Both are facts about market structure rather than popularity, which is why the differences between the classes follow from the structure.

• Currency pairs are dealt over the counter, continuously through the week, across many venues at once. The pairs joining the largest economies carry the most quoting parties and are the most consistently deep instruments most readers will meet. A pair involving a smaller economy is quoted by far fewer, and behaves differently in identical conditions.
• Indices derive from a basket of listed shares, so the interest behind them concentrates into the hours when those shares are trading. A contract on an index can be quoted while the exchange underneath it is closed, and the size behind that quote outside exchange hours is not the size behind it during them.
• Commodities and metals price from futures markets with their own sessions, contract months and delivery mechanics. Interest concentrates in the contract nearest delivery and thins as attention rolls toward the next, so depth changes across the month for reasons unconnected to the commodity itself.
• Shares carry the widest range of outcomes of any class. A large listed company is quoted by many parties throughout the exchange session; a small one may have very little size resting at any moment. Both are shares, and treating the class as uniform is the most common error here.
• Exchange traded funds inherit the liquidity of what they hold rather than depending on their own turnover alone. A fund holding widely traded constituents can be dealt in size even when the fund itself changes hands rarely, because the parties quoting it can lay off the position in the holdings underneath. A fund holding thinly traded assets has no such route.

One conclusion cuts across all five. Liquidity belongs to an instrument at a moment, not to a class as a category. The class explains the baseline and the shape of the day. It does not settle what a particular instrument looks like at a particular hour.

Where practitioners disagree 

There is no agreed way to measure any of this, and the disagreement is substantive rather than academic. One tradition treats the quoted spread as the working measure, because it is the only figure available continuously, in the same form, everywhere. The standing objection is that it describes the top of the market only, and that a quote can be narrow for a size almost nobody deals in.

A second measures depth from the resting interest a venue displays, which is closer to what an order meets. The objection there is that a display shows only what has been shown: interest held back is invisible to it, and displayed interest can be withdrawn faster than it can be dealt on. A third measures after the event, comparing fills against the price at the moment each order was sent, which describes real orders rather than intentions and only ever describes the past.

Each answers a different question and none answers all three. What follows is modest but useful. Any single liquidity figure is a description of one dimension at one moment, and carrying it across to a different size, a different instrument or a different hour is an assumption rather than a reading.