Rollover, holidays and thin markets

The cut off, and what happens at it 

The cut off comes from the settlement calendar, not from any firm's preference. A spot currency transaction agreed today settles two business days later, and that date is part of what was agreed. A position still open at the end of the trading day would fall due on a date that has arrived, so the market rolls it: the position is closed and reopened at the same price for value purposes only, the settlement date moves forward a day, and the interest owed on the two currencies over that day is settled in cash. Size and opening price are untouched. What moves is the date, and an interest line appears on the account.

Convention fixes that moment at five in the afternoon in New York, the boundary the interbank market treats as the end of one value date. Anchored to New York, it shifts in Gulf local time when North America changes its clocks, and again for a fortnight when Europe changes its own on a different date.

On a contract written on something other than a currency pair the adjustment has a different origin and the same shape. Because the full notional value of a CFD is never funded, a contract held past the cut off carries a financing charge on the portion not posted, conventionally a reference rate for the currency of denomination, adjusted by the counterparty. Contracts on shares and indices also carry a dividend adjustment on the day the underlying goes ex dividend, debited from one side and credited to the other.

The adjustment is applied separately from the trade's own profit and loss and appears as its own line, which is why a position closed at exactly its opening price can still leave the balance somewhere else. The rate for each instrument is published in its contract specifications, alongside its size convention and trading hours, across the markets pages. The arithmetic is set out step by step in the rollover guide.

Why the line is interest 

A currency pair is two interest rates carried in one price. Holding one currency and owing the other across a day produces an amount receivable on the one and payable on the other, and the financing adjustment is the difference, modified by the counterparty's markup. The side holding the higher yielding currency is in principle credited and the other debited. The qualifier does real work: the markup is applied to each side separately rather than shared between them, so when two policy rates sit close together the arithmetic commonly produces a debit on both sides of the same pair.

The two results are deliberately not mirror images. The markup is applied to each side independently, so the credit on one side is the smaller figure, and a financing line read as symmetrical produces the wrong number every time.

The weekend, charged on one night 

Settlement runs on business days and a weekend contains none. A position held past the cut off on a Wednesday rolls its value date from Friday to the following Monday, three calendar days, so three days of interest are applied on that single night. Nothing about the market changes that evening. The charge follows the calendar rather than the trading.

The night the triple charge lands on is not universal. Currency pairs conventionally take it on Wednesday. Instruments settling on an exchange calendar commonly take it on Friday, and a national holiday in either currency of a pair can move the day or add another to the count. The convention is published per instrument and firms differ, so the specification is the only reliable statement of it.

Why spreads commonly widen around the cut off 

The cut off falls at the end of the New York afternoon, after London has closed and before Tokyo has opened, the point of the cycle at which the fewest institutions are at their desks. The market makers that quote continuously use those minutes to square and roll their own books, and several stop quoting while they do it. A quoted spread is only ever the distance between the best order on each side, so fewer quoting participants widen it as an arithmetic consequence. It is the same market with less of the market in it.

Two costs taught separately earlier arrive together in that window: the spread is wider, and the financing line lands the same evening. The block below shows the proportion rather than any quoted figure.

The days liquidity is absent 

A thin cut off lasts minutes. A holiday thins the book the same way and lasts a day or a fortnight. The periods below are the ones participants conventionally treat as reliably thin, and the dates move every year.

• The end of the year, from the last full week of December into the first days of January, when desks in every centre run reduced at the same time.
• National bank holidays in the country of either currency in a pair. A closed banking system cannot settle, so value dates shift and the days of interest applied shift with them.
• Exchange holidays, which close the underlying market for an index or a share CFD outright. A contract cannot reference a price nobody is quoting.
• Good Friday and Easter Monday, which close most of Europe and part of the United States within days of each other.
• Japan's holiday sequence in late April and early May, the August lull across continental Europe, and Thanksgiving in the United States with the shortened session after it.
• The fortnight in spring and again in autumn when Europe and North America change their clocks on different dates, moving session boundaries and the cut off relative to Gulf Standard Time.

The working week mismatch described earlier in this module compounds during these periods. A holiday that closes New York removes the deepest hours of a Gulf working day, and the Gulf weekend already falls when the largest centres are shut. A calendar of the underlying exchanges is a more useful object here than a clock.

What a thin book changes 

Depth is the quantity available at each price rather than the price itself, and depth is what a holiday removes. In a deep book an order of a given size fills at or close to the best quoted price, because enough is resting there to absorb it. In a thin book the same order exhausts the best price and takes the next, so the average price it achieves sits further from the quote it was sent against. The order did not change. The book it arrived into did.

That has a direct consequence for orders that are triggered rather than placed. A stop is an instruction to close a position once a specified level trades, and once triggered it becomes an order at the market. It is not an instruction to close at the level and it is not a guarantee of the closing price. In a thin or fast market the next available price can sit some distance beyond the level named, and across a weekend or holiday closure a market can reopen past it entirely.

Margin requirements move for the same underlying reason. A requirement is a percentage of a contract's notional value, set by the counterparty, and it is not fixed for all time. Requirements are commonly raised on some instruments ahead of a weekend or a scheduled policy decision, because what is collateralised in those windows is the risk of a gap rather than of a continuous move. Since profit and loss are calculated on the full notional value while only a percentage of it is posted, an adverse gap is measured against the whole contract and the resulting loss is not limited to the amount deposited. A favourable gap is measured on the same basis and to the same degree.

A raised requirement also has an effect involving no price movement at all. More margin is held against a position that has not changed size, so free equity falls on its own and an account can move closer to the level at which open positions become liable to be closed, in a period when the market is shut. Cost, margin and execution are all functions of the same book, and a holiday changes the book.

Where practitioners disagree 

Two arguments here are genuinely unsettled. The first is whether financing is a reason to avoid holding through the cut off. One tradition treats the adjustment as a running cost accruing every night regardless of what the market does. Another observes that closing before the cut off and reopening after it crosses the spread twice, in the window where the spread is widest, and that the two crossings can exceed several nights of financing. Which is larger depends on the instrument, the rate and the width at the moment of re entry, so neither argument generalises into a rule.

The second is whether a holiday market is quiet or dangerous. One description holds that ranges compress when participants are away. The other holds that a thinner book lets a modest order consume more levels, so ordinary flow prints a larger move, which often retraces when normal liquidity returns. Both describe the same book: fewer participants produce fewer events and larger prints per event. The balance differs by instrument, by holiday and by year.