3. The calendar: which days change the rules of the game

Macro 6: From macro to strategy

There is a handful of dates on which the market stops behaving like the rest of the month. Not because they are predictable (they are not) but because they change the conditions in which you execute: the liquidity available, the width of the bid-offer spread, the depth of the order book and the usual relationship between a signal and the price you actually get.

Why the calendar matters even if you predict nothing

Every strategy has two parts: the idea and the execution. The idea is tested on closing prices and comes out well; the execution happens in a real market where somebody has to be on the other side. In normal conditions that difference is small and can be modeled with an average cost. On certain days the difference explodes, and a strategy with a narrow margin can go from profitable to ruinous without the signal having failed even once.

Central bank days

The Federal Reserve holds eight FOMC meetings a year. The statement comes out at 2pm New York time and the press conference starts half an hour later. That long hour concentrates a good part of the month's movement, and it is common for the initial reaction to the statement to reverse during the press conference, as the message gets qualified. The European Central Bank follows an equivalent pattern: decision in the early afternoon and a press conference shortly afterwards. Add the Jackson Hole symposium in August, which without being a formal meeting has moved markets regularly.

What changes on those days is not just volatility. Liquidity providers know there is a binary event at 2pm, so they pull their orders or post them a good deal further away. The result is a thin book in the hours beforehand and a very wide spread in the first seconds afterwards.

The data: inflation and payrolls

The two that move the US market most are released at 8.30am New York time, before the equity market opens:

• CPI, monthly. Remember from level 3 that what moves the price is the gap with what was priced in, not the level of the number.
• The non-farm payrolls report, on the first Friday of each month. Its later revisions are frequently large: the market reacts violently to a figure that may be corrected months afterwards.

In the seconds around a release like that, the order book empties out. A market order sent at that instant can fill a long way from the last price you saw on screen, and that difference shows up in no backtest that uses closing prices.

Expiries, month ends and August

• The third Friday of March, June, September and December sees index futures and options and single-stock options all expire at once: the so-called triple witching. It usually coincides with the quarterly rebalance of index composition and concentrates a volume in the closing auction that dwarfs a normal day. Those moves reflect mechanical repositioning, not new information.
• Month end and quarter end drive rebalancing by index funds and pension schemes, and generate buying or selling flows that owe nothing to any view on prices.
• Year end adds a further strain: institutions shrink their balance sheets for regulatory reasons, which pushes up the cost of very short-term funding on exactly those days.
• August and the public holidays work the other way round: there is no event, there is an absence of participants. With half the usual liquidity, any piece of news produces an outsized move.

"Don't trade that day" is not advice, it is a consequence

It is worth saying clearly, because it circulates as a universal rule and it is nothing of the sort. The fact that conditions change on a given day does not mean you should stay out. It depends entirely on what you do:

• If your horizon is years, those days are all but irrelevant to you: you will pay for one worse fill and it makes no difference.
• If your strategy makes money precisely when things move, those days may be the best of the month.
• If you trade intraday on thin margins with market orders, then yes, you may be giving away over a few days what you earn across all the rest.

The right attitude is not to memorize prohibitions but to measure: split your trades into event days and normal days, compare the gap between the price you expected and the one you got, and decide with that figure in front of you. It is a concrete version of the previous module's idea: do not predict the day, know your exposure to that day's conditions.

A real case: August 5, 2024

In late July 2024 the Bank of Japan raised rates and the yen strengthened sharply. That hit the yen carry trade from level 4: years of positions funded in the cheap currency which, as that currency became expensive, all had to be unwound at once. It began to come apart in the last week of July and blew up on Monday August 5.

The Nikkei fell 12.4% in a single session, its largest daily drop since 1987. The VIX index went above 60 points at the US open, a level seen only in 2008 and in March 2020, only to close that same day below 40. The S&P 500 ended down around 3%: far less than that volatility spike suggested.

That mismatch is the lesson of the module. A good part of the size of the move came not from the content of the news but from when it arrived: the first days of August, with half the trading desk away and an order book that could not take the size thrown at it. In October, with normal liquidity, the same news would have moved things a good deal less. The following weeks recovered most of the fall, which confirms that much of it was market mechanics and not a reassessment of the future.