Gulf public holidays and thin liquidity

Two kinds of Gulf holiday 

The Gulf public holiday calendar has two halves that behave completely differently. One half is fixed to the Gregorian calendar and can be entered into a system once: the Gregorian new year, and each state's own national and commemorative days, which fall on the same date every year and differ from country to country. The other half is fixed to the lunar Hijri calendar, which is shorter than a solar year, so those holidays arrive earlier in the Gregorian year each time and cannot be scheduled permanently.

The lunar half includes the Islamic new year, the birthday of the Prophet, the day of Arafat, and the two Eid holidays that follow Ramadan and the pilgrimage. Their start is confirmed by observation of the new crescent rather than by calculation, so the announcement can arrive the evening before, the length can be extended after it is announced, and neighbouring states can begin the same holiday on different days.

No page can publish that calendar reliably, and this one deliberately does not try. Holiday dates are set by each government, exchange closures are published by each venue, and any firm's own coverage over a holiday is a fact about the firm. All three are read from their own sources.

What a Gulf holiday actually closes 

• Regional exchanges. A listing at a closed venue does not trade, and an instrument written on that listing does not quote while the venue is shut.
• Local banks and payment infrastructure. Domestic transfers do not settle on a closed day, and anything that depends on a local business day is deferred until the next one.
• Local business generally, including the offices of any regional counterparty, service provider or authority.

What a Gulf holiday does not close is the global market. Turnover in the major currency pairs, in the widely quoted indices and in the international commodity benchmarks is overwhelmingly transacted in London, New York and the Asian centres, and none of those is observing a Gulf holiday. Quoted spreads in those instruments do not widen because the Gulf is not working, and scheduled economic releases elsewhere are published on their own calendars regardless.

The practical consequence for a reader in the region is an asymmetry worth stating once, plainly. The markets that a regional position is exposed to are open on days when the local infrastructure around it is not, and closed on days, principally the western weekend, when the local infrastructure is working. Neither of those is a fault. They are two calendars that were never designed to agree.

Which closures genuinely thin the market 

A market thins when the firms that make prices in it are absent, so the holidays that matter are the ones observed by the centres that carry the turnover. A public holiday in the United States or the United Kingdom removes a large share of global dealing capacity for a day. A Japanese holiday does the same to the Asian session, and Japan observes a considerable number of them, including a cluster in spring during which several fall in the same week. Continental European holidays thin the European session, and because they are not uniform across countries a given day can be a holiday for some European desks and not others.

Several dates are thin without being holidays at all. Major markets schedule shortened sessions around some national holidays, so the venue is open but closes early. The last stretch of the calendar year is conventionally quiet across most centres, and the first working day after any long closure carries the accumulated adjustment of everything that happened while the market was shut.

What a thin book does mechanically 

Liquidity is not a mood. It is the quantity of size resting at and near the current price, and everything commonly described as a thin market follows arithmetically from having less of it. Fewer firms are quoting, each quotes for a smaller amount, and the distance between the best bid and the best offer widens because the firms making prices are compensating for the greater difficulty of laying off what they take on.

An order that asks for immediate execution consumes the resting size at the best price, then the size at the next price, and so on until it is filled. The average price it achieves is therefore a weighted average of the levels it consumed, and that average sits further from the price displayed at the moment the order was sent when there is less size at each level. That difference between the price seen and the average price achieved is what slippage measures, and it is a property of the book rather than of the order.

The second mechanical effect is the gap. When a market is closed and information arrives, there is no market to trade the difference through, so it appears at the reopen as a step rather than as a path. An instruction resting at a price inside that step is not filled at its price, because that price never traded; it is filled at the first price actually available past it. The same thing happens inside a thin session on a large release, where the step is small enough to be called a jump and behaves identically.

Reading the calendar from the Gulf 

Putting the two halves together gives a reader in the region four distinct kinds of day, and the distinctions are worth holding because they look alike on a calendar and behave nothing alike.

1. A Gulf holiday on a global working day. Regional venues and banks are closed, the global market is normal, and any exposure to it is unchanged and unwatched locally.
2. A major centre holiday on a Gulf working day. Regional offices are open, the global market is thin, and quoted spreads and fills in the affected instruments reflect that.
3. A day that is a holiday in both, which is genuinely quiet everywhere and is also the case in which the fewest people are available anywhere to resolve anything.
4. A reopen after any of the above, which carries the accumulated adjustment of everything that arrived while the market was shut, concentrated into the first prices quoted.

Where practitioners disagree 

One argument is whether thin conditions are riskier or merely different. The case that they are riskier rests on the observation that the same order moves the price further and that protective instructions are less likely to be filled near their level. The case that they are merely different rests on the observation that a thin market is also a quiet one, that many thin days produce almost no movement at all, and that the risk is concentrated into rare events rather than raised uniformly. Both are describing the same distribution and disagreeing about which part of it to name.

A second argument is whether electronic market making has removed the holiday effect. One side points out that automated systems do not take holidays, so quoting continues around the clock and the old picture of an empty dealing room is obsolete. The other points out that automated systems are calibrated to expected conditions and are frequently configured to quote smaller and wider when those conditions are absent, so the capacity is present but deliberately reduced. The measurable outcome, a wider quote and less resting size, is the same under either explanation.