What systematic trading means

Rules decided in advance

A systematic trader decides, before a trade happens, exactly what will cause it: which conditions open a position, which close it, and how large it is. Once those rules are set, they are followed as written. The alternative, discretionary trading, leaves those decisions to the trader’s judgement at the time.

Neither approach is automatically better. A discretionary trader can adapt quickly to something new. A systematic trader gives up some of that flexibility for consistency: the same situation produces the same decision every time, whatever mood the trader is in.

Systematic is not the same as automated

The two words are often used together, but they describe different things. Systematic describes where decisions come from: written rules. Automated describes what carries them out: software. A person can follow a rulebook by hand, and software can place trades on a whim.

When a strategy is both, the rules are written as code, and the software watches the market, applies the rules and sends orders without anyone pressing a button. Automation removes delay and hesitation. It also removes the chance to catch a mistake in the rules before it is acted on, which is why the work done before automation matters so much.

What a rule has to say

A complete rule leaves nothing to decide later. At a minimum it answers four questions:

• Entry: what has to be true for a position to open.
• Exit: what closes it, whether that is a target, a stop or a time limit.
• Size: how many contracts to trade.
• Conditions: when the rule does not apply at all, such as around scheduled news or outside certain hours.

Strategies are often grouped by how long positions last. Short-term strategies open and close positions within hours or a single session. Swing strategies hold positions for longer, often across several days, to capture larger moves. The two can run side by side, because they respond to different behaviour in the market.

Testing before trading

The usual first test is a backtest: applying the rules to historical market data to see how they would have behaved. A backtest is good at rejecting bad ideas quickly. It is much weaker at proving good ones, for several reasons:

• Overfitting. With enough adjustments, almost any rule can be made to look excellent on past data. The more closely a rule was tuned to fit history, the less likely it is to describe anything that repeats.
• Look-ahead bias. A test can accidentally use information that would not have been available at the moment of the decision, such as a day’s closing price used to decide a trade during that same day.
• Costs and fills. Commissions, exchange fees and slippage, the gap between the expected price and the price actually traded, are small per trade and large in total, especially for short-term strategies.
• Changing markets. Behaviour that held for years can fade as conditions and participants change.

To allow for these, careful researchers keep some data aside that plays no part in building the rule, and test on it only once. They often then run a strategy on live prices without real orders, or at a very small size, before relying on it.

Improving without tinkering

Markets change, so rules that are never revisited tend to decay. But changing a rule after every bad week is discretionary trading with extra steps. The discipline is to treat a change like a new rule: state it, test it on data it was not built from, and only then let it replace what is running.

Keeping versions matters for the same reason. When each change is recorded, it is possible to tell whether a later result came from the market or from the edit.

In short

Systematic trading means decisions come from written rules rather than from judgement in the moment, and automation carries those rules out. The quality of the outcome depends far less on the software than on the research, testing and restraint that went into the rules.