What systematic trading is, how it differs from discretionary and algorithmic trading, and the components every system needs to work.
Systematic trading means deciding your rules in advance and following them, rather than judging each situation as it arrives. The rules can be executed by code or by hand. What makes the approach systematic is that the decision was made before the moment arrived.
This guide covers what that involves, how it differs from the adjacent terms it gets confused with, and the components a working system needs. It is the hub for a set of deeper pieces, linked at the end.
A systematic approach specifies the conditions for entering, exiting and sizing a position before any of those situations occur. Given the same market data, it produces the same decision.
That property is the entire point. It means the approach can be described precisely, tested against history, and examined afterwards to see what it actually did rather than what you remember it doing.
Automation is common but not required. A trader working from a written checklist, entering orders by hand, is trading systematically as long as the checklist determines the decision. A trader running code who overrides it when the chart looks wrong is not.
The dividing line is whether the rule decides or the person does.
Discretionary trading uses judgment at the moment of decision. The trader weighs what they see and acts, drawing on experience that may not be fully articulable.
Each approach gives up something real.
A systematic approach cannot use information it was not designed to see. Something unusual happens, the rules have no view on it, and the system proceeds as though the situation were ordinary. A discretionary trader notices.
A discretionary approach cannot be tested, cannot be checked against what it claimed it would do, and varies with the state of the person running it. The same setup gets traded differently on a difficult morning.
Neither is better in the abstract. The systematic route suits anyone who wants to know whether their approach works rather than believe it, and who would rather be consistently mediocre than occasionally brilliant and occasionally terrible.
Three terms describe different axes and get used as though they were synonyms.
Systematic describes how the decision is made: from predetermined rules rather than in-the-moment judgment.
Algorithmic describes how the decision is executed: by code placing orders rather than a person clicking. A systematic approach can be executed by hand, and is then systematic but not algorithmic.
Quantitative describes where the rules came from: statistical analysis of data rather than reasoning or experience. A rule derived from an observed market mechanism can be systematic without being quantitative in origin.
Most working setups are all three, which is why the terms blur. Keeping them separate is useful when something fails, because it tells you which part to examine.
However simple the approach, the same pieces have to exist somewhere, even if some of them are one line.
A universe. What you trade, and how that set is decided.
An entry rule. The conditions under which a position is opened.
An exit rule. The conditions under which it is closed, including when the idea has simply not worked. Exits determine results more than entries in most systems, and they are where unspecified systems usually turn out to be unspecified.
A sizing rule. How much. This fails independently of everything else and causes more damage than most logic errors, because size can end up correlated with the conditions that perform worst.
A stopping rule. What would make you turn the system off. Decided in advance, because deciding while losing money is not deciding.
Anything left unspecified becomes a discretionary decision made under pressure, which is exactly what the approach was meant to avoid.
The most common failure in a system that once worked is not a broken rule. It is a rule still running correctly in conditions it was never built for.
Rules designed to fade extremes behave differently when a move keeps extending. Rules designed to follow breaks behave differently when price keeps returning to the middle of a range. Neither rule has changed. The market has.
Market regime is the state a market is in, described rather than predicted: trending, ranging, or bearish. As a system input it sits before the entry rule, as a gate that can block evaluation but never trigger it.
It is measurable. ADX supplies trend strength, with readings above 25 marking a defined trend. Moving average structure supplies direction. The part that needs care is confirmation, because a state read from a single candle flips repeatedly near the threshold. RegimeLab accepts a change only after three consecutive matching reads on the 4-hour timeframe, which removes that instability at the cost of accepting the change later than it happened.
Recording began on 20 June 2026, with a nine-day gap in late June and early July where collection stopped. That record is published for every tracked pair, which makes it possible to segment your own results by the conditions they occurred in.
The approach has characteristic failure modes, and most of them are human rather than technical.
Overriding the system. The most common failure. Once you intervene, you no longer know what the system would have done, and every future result is a mixture you cannot separate.
Fitting the rules to the data. Adjust parameters until the history looks good and you have described the past rather than built a rule. This is covered in walk forward optimization.
Ignoring costs. Fees, spread and slippage sit outside the rules and often decide whether a frequently trading system is viable at all.
Assuming conditions persist. A system tested entirely in one kind of market has been tested against one sample of one condition.
Confusing consistency with correctness. A system follows its rules reliably whether or not those rules have any edge. Reliability is not evidence.
The pieces below go deeper on each part of what is described above.
Building the system:
Choosing what the system does:
Testing before trusting:
Market state and regime:
Plumbing:
An approach where the rules for entering, exiting and sizing positions are decided in advance rather than judged in the moment. Given the same market data it produces the same decision, which is what makes it testable.
Discretionary trading applies judgment at the moment of decision. Systematic trading applies a rule decided earlier. The systematic route can be tested and checked afterwards; the discretionary route can use information the rules were never designed to see.
No. Systematic describes how the decision is made, from predetermined rules. Algorithmic describes how it is executed, by code. A trader following a written checklist by hand is systematic without being algorithmic.
Yes. Automation is common but not required. What matters is that the rule determines the decision rather than the person. Code removes execution errors and makes testing easier, which is why most systematic traders eventually adopt it.
Specify a complete set of rules including entry, exit, sizing and a stopping condition, then test them against history before running anything. Most of the work is making vague ideas specific enough to be tested at all.
A system cannot use information it was not designed to see, so unusual situations are handled as though they were ordinary. It also runs its rules reliably whether or not those rules have any edge, and reliability is easily mistaken for evidence.
Often nothing has broken. The market has moved into conditions the rules were not built for, and nothing in the system reports that. Recording the market state alongside results is what separates a conditional system from a failing one.