For example, imagine you watched a video claiming that buying a stock when its short-term average price rises above its longer-term average is a dependable approach. You like the idea and want to start trading it with real money next week. A test can show whether the idea is clear enough to follow and how it behaved on past prices. It cannot tell you what prices will do next.

This article is general education only. It is not financial advice, and it does not recommend any strategy, product, or market. Trading carries a real risk of loss, and you can lose some or all of the money you put in. Testing lowers the chance of unpleasant surprises, but it does not remove that risk.

Written rules are the starting point of a systematic approach, and systematic trading in Singapore at Collin Seow’s site is a helpful place to see that approach explained. The site presents a method built around what to buy, when to buy and how much to buy, and it has a blog with articles on trading and investing.

Before you start, you need a trading idea you can state as rules rather than a feeling, historical price data for one market with clear dates, and a spreadsheet or notebook where you can log every test and every simulated trade. Work through the steps in order, without skipping ahead.

What a test can and cannot show

A test can show whether your rules are complete and followable, and how they behaved on data you already have. It can expose vague rules, hidden assumptions, and results that depend on a few lucky trades.

A test cannot prove that a strategy will work in the future. Markets change, and past results are not a promise. If anyone tells you a test guarantees results, treat that claim as a warning sign.

Write the rules down first

Before you look at any historical chart, write the rules in plain words. A complete set covers five things:

• Entry: the exact condition that makes you open a trade.
• Exit: the condition that closes a winning trade, and the condition that closes a losing one.
• Position size: how much of your money goes into each trade.
• Risk per trade: the most you would lose on one trade if it moves against you.
• Market and timeframe: one market and one timeframe, such as daily price bars. Testing several at once makes the results hard to read.

If someone else could not follow your rules without asking you questions, the idea is not ready to test.

Backtest on past prices

Backtesting means running your written rules over historical prices and recording every trade the rules would have taken. A backtest is only as useful as its data, so check the data first. Look for missing days, obvious price errors, and jumps that do not match the market’s history. Check whether prices have been adjusted for events such as share splits, and use one method consistently.

Next, choose a period that includes different market conditions, such as rising, falling, and sideways phases. A period that covers only one mood of the market tells you very little about how the rules handle the others. How long the test must run depends on how often the strategy trades: a strategy that trades rarely needs a longer history to produce enough trades to read. Either way, the period must include varied conditions.

Warning: do not stop the backtest when the results look good, or keep tweaking the rules until they do. That habit leads to over-fitting, described below.

Keep out-of-sample data untouched

Out-of-sample testing means checking your rules on a period of data you did not use to design or tune them. Split your history before you start. Use one part to build and adjust the rules, and keep the rest sealed until the rules are fixed.

Over-fitting happens when rules are tuned so closely to past data that they capture noise, the random ups and downs that will not repeat. An over-fitted strategy looks excellent in the backtest and then disappoints on new data. Warning signs include many tuned settings, rules that only work with one exact setting, and results that fall apart once you apply them to the data you held back.

Each time you adjust a setting after seeing a result, you are fitting the rules to the past again, and the held-back data becomes less useful. If you must change something, record the change and test it on fresh data.

Read the results

Record the results of every test in the same log. Then read these measures together, because no single one settles the question:

• Win rate: the share of trades that closed with a profit. A high win rate means little if the losing trades are far larger than the winning ones.
• Average win and loss: the typical size of winning trades compared with the typical size of losing trades.
• Profit factor: the total money made on winning trades divided by the total money lost on losing trades.
• Expectancy: the average result per trade in the test, combining how often you win with the size of your wins and losses.
• Maximum drawdown: the largest fall from a peak in account value to a later low during the test. It shows the pain along the way, which matters if you would have quit halfway through.

Warning: do not pick whichever measure looks best and ignore the rest. A strategy can look strong on one measure and weak on another.

Paper trade before real money

Paper trading means following your rules in live markets and recording the trades as if they were real, without risking money. It checks whether you can execute the rules in real time. Many rules look clean on a chart but are hard to follow when prices move quickly, when a signal appears at an awkward hour, or when the urge to skip a losing trade arrives.

Run paper trading after the out-of-sample check, so the rules are already fixed. Log each simulated trade with the time, the price you would have used, and the reason for the entry and exit. Then compare the paper log with the backtest. Some differences are expected, but large gaps are a signal to find out why before you go any further. As with the backtest, the length of this stage depends on how often the strategy trades and whether the period includes different market conditions.

Warning: paper results can look better than real results because simulated losses do not feel like losses. Judge the rules by the log, not by how calm you felt.

Move to live trading in stages

If the rules survive every earlier step, the next move still deserves care. Start with a small position size, one where a loss would sting but would not do lasting damage.

Before the first live trade, write down your stop conditions. For example, imagine you decide in advance that if live results fall well outside what the backtest and paper log suggested, you pause and review the rules instead of trading through the problem. Stop conditions might include a maximum loss you set for the period, or a rule that you do not trade while the market is in an unfamiliar condition.

Compare live trades with the paper log and the backtest as you go. If live results differ, return to the log to find the reason. Do not change the rules on the fly. Increase position size slowly, and only if the comparison makes sense and no stop condition has been triggered.

Warning: never put in money you cannot afford to lose. Losses can be larger than expected, and capital can be lost.

Checklist and next step

Before you go live, confirm each item:

• The rules are written in full: entry, exit, position size, and risk per trade.
• The rules cover one market and one timeframe.
• The historical data was checked for errors and gaps.
• The backtest covers different market conditions.
• Out-of-sample data was held back and tested only after the rules were fixed.
• Paper trading was run and compared with the backtest.
• Stop conditions and a starting position size are written down before the first trade.

Testing reduces the chance of surprises, but it does not remove risk and it does not guarantee that a strategy will work. Readers who want to keep building on a rules-based approach can return to the systematic trading resources mentioned at the start.