Volatility around events is a risk topic first

What happens in the minutes around a release 

The participants who quote continuously are quoting into an unknown at a known time. They do not have the number either, and the cost of being on the wrong side of it is concentrated into a single instant that everybody can read off a calendar. The rational response for a quoting participant is to show less and charge more for it: fewer units behind each price, a greater distance between the price at which they will buy and the price at which they will sell, and in the final seconds, on some venues, no quote at all. The book thins, and it thins in advance.

Then the figure prints. Quotes return, but struck against a different consensus, so for a short period price moves in jumps rather than steps: at several levels nothing traded. The window closes as quoting capacity comes back, and its length varies by instrument and by how far the figure landed from what was expected, which is the part nobody can schedule.

Every effect in the list below is a consequence of the same thinning, and every one of them is an execution or cost effect rather than a directional one. They occur around scheduled releases with enough regularity to be treated as a property of the window rather than as an unusual event within it.

• The spread widens, often several times over, and stays wide until quoting capacity returns.
• The size available at each price falls, so an order large enough to consume a level reaches worse levels behind it.
• Market orders fill away from the price on screen, in either direction, more often and by more than in a calm window.
• Resting orders trigger and fill at prices further apart than usual, because the two are separate events and the interval between them is where the market is moving fastest.
• Prices gap, meaning consecutive prints are separated by levels at which nothing traded, so an order resting inside the gap is filled beyond it rather than at it.
• Requotes and rejections become more frequent on venues that use them, because the price a request was built on expired before the request arrived.
• Margin requirements on the affected instruments may be raised in advance of the event, changing what an already open position requires.

The spread is the first thing to move 

A spread is the price of immediacy, and it is set by the state of the book rather than by a schedule of charges. That distinction matters most in exactly this window. Nothing is being repriced by anyone as a policy; the distance between the best bid and the best offer is simply wider because the participants standing between them require more compensation for standing there. A firm quoting a client around a release is quoting off the same thinned book, and the widening passes through.

The consequence is arithmetic and it is the same for a position in either direction. Crossing a wider spread costs more, both to open a position and to close one, and a position opened inside the window starts further from break even than the identical position opened outside it. The calculation below holds size constant and changes only the spread.

A stop is least reliable in the window it seems most needed 

A stop order names a level at which it becomes live, not a level at which a position closes. Those are two events, and everything about a scheduled release stretches the distance between them. The trigger price trades, the instruction becomes an order to close at the next available price, and the next available price around a release can be several pips away from the level named, because the levels in between were passed without a trade occurring at them.

The uncomfortable part is that the two things move together. The conditions that make a stop feel most necessary, a fast market with a wide range and a known catalyst, are the same conditions that make the interval between trigger and fill widest. A stop is not less useful for this, but it is not the boundary it is often described as. The rows below read one event twice.

Some firms offer a guaranteed variant of the stop, which fills at the level named regardless of where the market reopens, and prices that certainty as an explicit charge. Where such an order exists, its availability is commonly restricted or its charge raised around scheduled releases, which is the same fact stated from the other side: the guarantee is most expensive precisely where the ordinary instruction is least dependable.

Margin requirements can change around a known event 

A margin requirement is a percentage of the notional value of a contract, held as collateral for as long as the contract is open, and it is set by the firm rather than fixed by the instrument. Because profit and loss are calculated on the full notional value while only a percentage of it has been posted, an adverse move is measured against the whole contract and not against the collateral, so a loss can exhaust the margin entirely and is not limited to the amount deposited. A favourable move is measured on exactly the same basis and to exactly the same degree.

Because the requirement is a firm's setting rather than a constant, it can be changed, and the occasions on which firms conventionally change it are the ones on the calendar: a central bank decision, a national election or referendum, a scheduled index rebalance, and the weekly close, where a market that reopens after a non trading period can reopen at a distance from where it closed. A requirement raised ahead of one of those windows applies to positions that are already open. Nothing about the position changes and the price need not move at all; what changes is how much collateral the same contract ties up.

The second order effect is the one worth holding onto. Collateral that has been reallocated to an existing position is no longer free, so the same account can move closer to its close out threshold without a single tick going against it. That threshold is published rather than decided case by case: at YAL, positions become liable to be closed once equity falls to 50% of the margin required to hold them. Firms conventionally announce a scheduled change in requirements in advance, and an announcement is a notification rather than a negotiation, so the holder of an open position carries the change either way.

The calendar reaches positions that were never placed for it 

A position that is open when a release prints is exposed to that release. The exposure is a fact about the clock and the contract, not about the reasoning that produced the position, and nothing in the mechanism distinguishes a position taken for the event from one taken a fortnight earlier for entirely unrelated reasons. A contract held across a week is exposed to every scheduled item inside that week, and a contract held over a weekend is exposed to whatever accumulates while quoting is suspended.

The exposure also travels further than the obvious instrument. An index contract carries the rate decision of the currency it is denominated in. A dollar priced commodity or metal responds to releases about the dollar as well as to anything specific to the commodity itself. A currency whose exchange rate is administered against another transmits the anchor economy's calendar rather than insulating anyone from it, so the release that moves a position may belong to an economy that is nowhere in its name. Reading the calendar is therefore a description of what an open contract is exposed to while it stays open. It is position information rather than a technique, and this page puts forward no view on what anyone should do with it.

Where practitioners disagree 

Two arguments about scheduled releases are genuinely unsettled. The first concerns whether exposure across a known release should be reduced as a matter of routine. One tradition holds that the distribution of outcomes around a release is unusually wide and unusually unknowable, so carrying less across it reduces the size of the unknown. The counterargument is that closing and reopening pays the cost of crossing the spread twice, often at its widest, which converts an uncertain risk into a certain cost incurred repeatedly. Both are internally consistent, and which dominates depends on how often an account trades and how large the spread is relative to the position, neither of which this page knows.

The second concerns whether the release itself is a tradeable subject at all. One school treats the largest single period moves of the month as the ones worth participating in. The opposing case is arithmetic rather than temperamental: the time is public, the consensus is public, and the execution conditions in that window are the worst of the day by every measurable dimension, so the price paid for participating is highest exactly when the price of being wrong is highest too. This page does not put forward either practice. It describes the second because a reader will meet the first stated as advice somewhere else, and because the conditions described above apply identically to both.