4. Hours, sessions and liquidity: why not every moment of the day is the same

Theory 2: How trading actually works

The same order, of the same size and on the same share, does not cost the same in mid-morning as it does at midday. The market is not the same place throughout the day.

So far we have talked about it as if it were always in the same state: the same spread, the same depth, the same appetite to trade. In practice, liquidity varies enormously with the time of day, the day of the week and even the time of year.

Trading without allowing for these variations is one of the quietest ways to lose money to execution costs without noticing.

The three stretches of a regular session

The regular session of an exchange such as the NYSE or Nasdaq usually lasts six and a half hours: from 9:30 to 16:00 New York time. But within that window, activity is far from uniform.

The first thirty minutes after the open account for a disproportionate share of the day's total volume. This is when all the information accumulated since the previous close gets resolved:

• News published overnight.
• Results from other exchanges that have already closed.
• Orders piled up during the night.

The price tends to be especially volatile while the market 'discovers' where it should sit.

For the opposite reason, the last half hour before the close also concentrates a lot of volume. Partly from index funds and institutional managers adjusting their portfolios right before the official closing price is calculated. That is the reference most indices and funds use to value their positions.

The middle stretch of the day runs roughly between 12:00 and 14:00 New York time. It is usually the quietest: less volume, somewhat wider spreads relative to volume, and generally slower price moves.

Pre-market and after-hours: fewer people, more noise

Outside the regular session there is the pre-market (before the official open) and the after-hours (after the official close). Some brokers allow trading in them, but with noticeably lower liquidity.

Fewer active participants means wider spreads and a much thinner order book. So a relatively small market order can move the price disproportionately.

Hence a very common phenomenon. You see apparently dramatic moves in pre-market or after-hours, following results published outside trading hours, that moderate considerably once the regular session opens and real liquidity arrives.

The same effect at weekly and annual scale

This intraday variation has a parallel on weekly and annual scales.

Mondays and Fridays tend to show volume patterns different from the rest of the week: Mondays digesting the weekend, Fridays closing positions before it.

And there are periods of the year when liquidity drops noticeably, because a good share of institutional managers are on holiday:

• The month of August in Europe.
• The week between Christmas and New Year, almost everywhere.

That leaves the market more prone to exaggerated moves caused by relatively small orders.

Why this decides whether a strategy works

All of this has a direct practical consequence for any strategy, discretionary or algorithmic: the same order size can have a radically different price impact depending on when it is sent.

Think of a strategy that works well trading in the most liquid stretch of the day. It can degrade noticeably if its signals, by accident of design, tend to fire in the least liquid stretch, or outside the regular session.

That is why any serious evaluation of a strategy has to take into account not only what is traded, but when it is traded. And we are not talking only about backtesting on prices, but about the design of when and how orders are sent.

The underlying reason is simple: the market is not a homogeneous entity throughout the day. In practice it is several different markets going by different session names.

When the market stops itself

There is also a mechanism designed explicitly for when volatility spikes too fast, even within the regular session: circuit breakers, or trading halts.

They are automatic pauses in trading, triggered when the price of an individual stock, or of the market as a whole, moves more than a given percentage within a short time window.

Their purpose is to give participants a forced pause so that information can be digested more calmly. That prevents an extreme, and possibly disproportionate, move from feeding on itself through panic or technical failures within seconds.

For any algorithmic strategy this has a direct practical implication. An order pending during a circuit-breaker halt simply does not get filled until trading resumes, and the reopening price can differ noticeably from the price just before the halt.

It is one more reason why no real strategy can assume its orders will always be filled exactly when and how it expected.