What it means
A trading system is a complete recipe. It specifies the currency pairs traded, the signals that trigger a trade, the stop-loss and profit target, and the position size, with nothing left to mood.
Systems can be mechanical, where a trader follows the rules manually, or fully automated, where software places the orders. Automated systems, sometimes called expert advisers or trading robots, remove emotion and can work around the clock, but they also need monitoring in case of faults.
The first step is back-testing, which means applying the rules to past price data to see how the system would have performed. A good test includes costs such as spreads and commissions, covers different market conditions, and checks that the results are not simply a result of fitting the rules too closely to history, which is known as over-fitting.
After back-testing, the system is normally run on a demo account (forward testing) and then with small real positions. Only when its live results match the tests should the trader scale up.
Key measures of performance include the profit factor, the maximum drawdown (the biggest peak-to-trough fall in the account) and the number of trades. A system with a high return but frequent deep drawdowns may be too painful to follow in practice.
The main risk is that markets change, so a system that worked in the past may stop working. Traders should set a limit on how much loss will trigger a review and keep a record of every trade.
A regular review every few months helps to spot a decline before it becomes costly.
In practice
Real-world examples.
Example
A part-time trader writes down a rule: buy EUR/USD when the 20-day average crosses above the 50-day average, risk 1% of the account and exit if the cross reverses. He follows it every time, even after a run of losses. His records show that most of his past mistakes came from breaking his own rules.
Example
A small fund develops an automated system that trades four currency pairs. Before going live, it runs the system on a demo account for three months and compares the results with the back-test. It sets a threshold, so that if live results diverge too far from the test the system is paused.
Example
A business owner who trades currencies in her spare time decides to stop discretionary trading after a series of emotional mistakes. She adopts a simple rules-based system and limits herself to checking it once a day. Her trading time falls, and her results become easier to measure.
Formula
Calculation
Profit factor = gross profit / gross loss
Net profit = gross profit - gross loss
Suppose a system takes 100 trades in a back-test. The winning trades add up to a gross profit of $18,000 and the losing trades add up to a gross loss of $12,000.
The profit factor is 18,000 / 12,000 = 1.5, and the net profit is 18,000 - 12,000 = $6,000, or $60 per trade on average. If costs of $10 a trade were not included, the real net result would be 6,000 - (100 x 10) = $5,000.Case study
Seen in the real world.
Ironbridge Capital is an illustrative, fictional trading firm that tested a new currency system. The back-test over ten years showed a profit factor of 2.2, which looked excellent.
On closer inspection, the analyst found that the rules had been adjusted repeatedly until the past results looked good. When the system was tested on data it had never seen, the profit factor fell to 1.05, which would vanish after costs.
In this illustrative case, the firm discarded the system and adopted a rule that every test must reserve a period of data for out-of-sample testing. The lesson was that a beautiful back-test can be a warning rather than a comfort. Ironbridge now requires a second analyst to try to break every system before any capital is committed.
Watch out
Common mistakes.
- Tuning a system until it fits past data perfectly, when this over-fitting tends to fail with new data.
- Overriding the rules during a losing streak, when consistency is the whole point of a system.
- Ignoring costs and slippage in back-tests, when they can turn a profitable system into a losing one.
Questions
People also ask.
Is a trading robot better than a human?
Not automatically, because a robot is consistent and fast but only as good as its rules, and it cannot adapt to unusual events. Many traders therefore keep a manual override for extreme market conditions.
How long should I test a system?
Long enough to cover different market conditions, typically several years of data and a meaningful number of trades, followed by a live trial with small positions.
What is drawdown?
It is the fall from a peak in account value to the following low, and it shows how much pain you must tolerate to follow a system. A system whose worst drawdown is larger than your account can bear is not suitable, however good the average return looks.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%Related