Paper Trading: How to Test a Strategy First

Why is paper trading worth the effort?

Paper trading is the stage between an idea and real money. It lets a trader find out whether a set of rules can be applied consistently, whether the signals appear as often as expected, and whether the results look anything like the backtest. It costs nothing but time, and it removes the financial consequence of a mistake while the rules are still being learned.

It is also easy to do badly. A demo account can be used to trade whenever a feeling suggests a trade, to change the rules when the results disappoint and to ignore the trades that went wrong. In that form it produces flattering numbers that do not describe any strategy, and it can give a false sense of readiness. The value of paper trading depends entirely on the discipline applied to it.

This guide sets out ten practices for paper trading that keep the test honest. They cover the written rules, realistic costs, the sample size, the record of skipped signals, the criteria for going live and the transition to small live trades. The strategies in this guide are examples of method, not recommendations, and the practices apply to any approach.

Why must the rules be written before the first trade?

A rule that exists only in the trader’s head can change without anyone noticing. A stop that is moved slightly to avoid a loss, an entry that is taken a little earlier because the setup looks promising, or a filter that is added after a poor week all alter the strategy being tested. The result is no longer the result of the original rules, and the record cannot tell the two apart.

Write the rules in full before the first trade. State the market, the timeframe, the entry condition, the stop, the target, the exit, the position size, the maximum number of trades per day and the news rule. Keep the rules on one page, and date them. The page becomes the reference for every trade, and any change to it starts a new sample.

A useful test is to give the rules to someone else and ask them to apply them to a chart from last month. If they reach a different decision from yours, the rules are not precise enough. Tighten them until the answer is the same for both of you, then start the paper test.

How should costs be set in the paper account?

A paper account should reflect the costs of the live account as closely as possible. Use the spread that your broker actually charges on the pair, not the tightest spread you can see. Include the commission if the account charges one, and include the swap for any trade held overnight. A paper test that ignores these costs will overstate the result, sometimes by a large margin.

Slippage is harder to model on a demo. A demo fill is often exactly at the quoted price, while a live fill can be worse, particularly in fast markets or around news. Record the fill price against the price shown at the signal, and note any difference. If the paper account gives no slippage at all, add an allowance in the analysis and say so in the report.

Check the costs at the time of the signal, not only the average. Spreads widen at rollover and during some sessions, and a strategy that trades those times will pay more than the average suggests.

What sample size is enough?

Thirty closed trades is a sensible minimum for a first look, and a larger sample is always better. A small sample can look excellent or terrible by chance, and a decision made on a few trades is usually a decision made on noise. The sample size should be stated in advance, and the strategy should not be judged until it is reached.

The sample should also cover more than one market condition. A strategy tested only during a strong trend has not been tested in a range, and the result says nothing about its performance when the trend ends. Record the condition at each entry, and check that the sample contains a reasonable mix.

Count the trades carefully. A trade that was signalled but not taken should be recorded as a skipped signal, not as a missing trade, and it should be counted in the analysis of the rule’s frequency.

Why must skipped signals be recorded?

A paper test that records only the trades taken hides the cost of the filter. If the strategy generates twenty signals and the trader takes eight, the other twelve may contain most of the losses or most of the wins, and the record without them is misleading. Every signal the rules produce should be logged, with a note if it was not taken and the reason.

Skipped signals also reveal execution habits. A trader who skips many valid signals may be hesitating for reasons that a rule could remove, such as an unclear stop or a fear of a particular setup. A trader who takes signals that the rules do not allow is showing the opposite problem. Both patterns are visible only if the skipped signals are recorded.

How do you set the criteria for going live?

Set the criteria before the paper test starts. A typical set includes a minimum number of closed trades, a minimum rule compliance rate, such as one hundred percent of trades entered under the written rules, a maximum drawdown in R-multiples, and a minimum average R-multiple after costs. The numbers are a choice, but they must be fixed in advance.

Write the criteria in the journal beside the rules. When the sample is complete, check the results against the criteria, and record the decision as pass or fail. A pass allows the move to live trading at a small size. A fail means the strategy needs review, a new sample or abandonment. Do not adjust the criteria after seeing the results, because that turns the test into a search for a passing number.

How do you measure the result honestly?

Measure the result in R-multiples after costs, not only in pips or money. An R-multiple divides each trade’s result by its planned risk, so trades of different sizes can be compared. The average R-multiple, the win rate, the profit factor and the largest losing run are the core measures.

Calculate the figures for the whole sample, and then for the subgroups that matter, such as each market condition and each session. A result that is positive overall but negative in every subgroup is a warning. A result that is positive in one subgroup only should be tested again before it is trusted.

Use a simple spreadsheet to calculate the figures from the journal. Check the formulas once against a manual calculation of a few trades, so that a formula error does not survive into the report.

What should you do when the paper result is poor?

A poor paper result is information, and it is much cheaper to receive on paper than in a live account. The first step is to check whether the rules were followed. If they were not, the result reflects the execution, and the fix is discipline. If they were, the result reflects the rules, and the strategy needs review.

A review of a poor result should not change the rules during the same sample. If a change seems necessary, end the sample, record the reason, write the new rules with a new date and start a new paper test. Keep the failed sample in the journal, because it shows what the strategy did under the original rules.

How do you move from paper to a small live account?

Move to live trading only after the criteria are met. Start with the smallest lot the broker allows, and keep the same rules, costs and journal format. Compare the first live trades with the paper results, and expect a difference, because live fills, emotions and slippage are not present in the paper account.

Keep the live size small until the live record has enough trades to be compared with the paper sample, and check the margin figures against the leverage guide before the first live order. A reasonable step is to trade at the reduced size for a period, then increase only if the live results remain within the expected range of the paper results. A large difference is a reason to pause and investigate, not to increase the size.

What does a worked paper test look like?

Imagine a trader who writes a trend-following rule on the one-hour chart of EURUSD. The rule enters on a close above a 50-period moving average, places a 20-pip stop below the last swing low, targets 40 pips and exits at the target or the stop. The position size is 0.10 lots, and no more than two trades are allowed per day. The trader sets the sample at forty closed trades and the criteria at a positive average R-multiple after costs, a profit factor above 1.2 and no single loss larger than the planned risk.

The paper account uses the broker’s typical spread of 1.5 pips and no commission. The trader adds a slippage allowance of 0.5 pips to each entry in the analysis, because the demo fills are at the quoted price. Over the first month, the rule produces forty-three signals. Thirty-eight are taken, and five are skipped because the trader was away from the screen. Each skipped signal is recorded with the reason and the outcome it would have had, so the cost of missing them is visible.

At the end of the sample, the trades show an average R-multiple of 0.18 after costs and a profit factor of 1.3. The criteria are met, but narrowly, and the largest losing run is four trades. The trader records the pass, notes that the sample covers only one month and one trending period, and moves to a small live account with the same rule, rather than increasing the sample or changing the rule.

The example shows the discipline that gives a paper test its value: the rule is fixed, the costs are real, the skipped signals are counted and the criteria decide the outcome. A test that passes narrowly is still a pass, but it is a reason for caution at the next step.

How do you test across different market conditions?

A paper test that covers only one kind of market can mislead. A trend-following rule may look excellent during a trending month and poor during a range, and the trader who tests only the trending month will be surprised later. Plan the sample to cover trending, ranging and news-heavy periods, even if that takes longer.

Identify the condition at each entry in a simple way. A moving average slope and the recent range are enough for most traders. Record the condition in the journal, and check at the end of the sample that the mix is reasonable. A sample that contains almost no range days cannot tell you how the rule behaves in a range, and the criteria should reflect that gap.

If the sample cannot cover a particular condition, say so in the report. A paper test that is honest about its limits is more useful than one that claims to cover conditions it did not see.

Why do paper results need a discount when moving to live trading?

Paper results usually flatter a strategy. Fills are often better than live fills, spreads may be narrower than the live average, and the absence of real money removes the emotional pressure that affects decisions. A strategy that wins on paper can therefore lose a little on a live account even when the rules are the same. The gap is normal, and it should be expected rather than treated as a failure.

Plan for the discount in the criteria. A common approach is to require a paper result comfortably above the level needed to justify live trading, so that a normal discount still leaves a positive result. Another is to measure the live result against the paper result over the first live sample, and to pause if the difference is larger than the expected range.

Record the discount in the journal. Over several strategies, the discount tells you how realistic your paper testing is, and it helps you set better criteria for the next test.

What do you do with a result that is close to the criteria?

A result close to the criteria is the most difficult to judge. It may be a genuine pass, a genuine fail or simply noise. The temptation is to extend the sample until the result passes, or to adjust the criteria so that it does. Both responses corrupt the test.

The honest response is to decide in advance what happens with a borderline result. A sensible rule is to treat any result within a fixed margin of the criteria as a fail, and to require a new sample with the same rules before going live. Write that rule in the journal beside the criteria, so the decision does not depend on how the trader feels when the result arrives.

How do you keep a paper journal that can be reviewed later?

A paper journal is useful only if a stranger could review it. Keep one row per signal, with the date, the time, the pair, the timeframe, the condition, the reason, the entry, the stop, the target, the exit, the spread, the slippage allowance, the result in pips and R-multiples, and a yes or no for rule compliance. Add a column for skipped signals and the reason.

Keep the rules, the criteria and the decision in the same file, at the top, with dates. A reviewer should be able to read the file from top to bottom and see the strategy, the test and the outcome without any other document. A journal that is scattered across several places is easy to lose and hard to audit.

Back up the journal at the end of each week. A corrupted file that loses a sample of thirty trades destroys the test, and a backup takes only a minute.

What are the common mistakes in paper testing?

The first mistake is changing the rules during the sample. The second is trading the paper account more often than the rules allow, perhaps out of boredom or to make the sample reach its target faster. The third is ignoring the skipped signals, which hides the cost of the filter.

The fourth mistake is judging the strategy on too few trades, or on one market condition only. The fifth is treating a single good week as proof. The sixth is moving to live trading with a result that barely meets the criteria, because the wish to start is strong and the criteria seem like a formality.

How do you know the paper test has finished?

A paper test is finished when the sample is complete, the criteria have been checked and the decision has been written. The end is set by the sample size and the criteria, not by the calendar or the mood of the trader. A test that is extended without a written reason has become a different test, and the record should say so.

Once the decision is written, close the paper test. Keep the file, and start a new journal for any live trading. The two records should be linked by the strategy name and the date of the transition, so that the comparison between paper and live can be made later, and so that a reader can see at a glance which strategy was tested under which conditions. A short paragraph at the top of each file that states the period, the broker and the account type will save the reader from reconstructing them later, and it costs a minute to write. Include the date the rules were fixed, the spread assumption, the slippage allowance and the name of the person, if any, who checked the rules before the first trade, since each of those facts changes how the result should be read. Keep this header up to date if any of those facts change during the test, and note the date of the change. A header that is never updated is a header that no longer describes the test, and that is worse than having none, because it looks reliable while being wrong. Check the header against the journal at each weekly review, as part of the same routine that checks the trades, and initial and date each correction so that the history of the test remains readable and every change can be traced to a date and a reason, and so that the person who reads the test next year can trust what it says.

How do you keep motivation steady during a long paper test?

A paper test can take months, and it is easy to lose interest once the novelty has passed. Boredom is a practical risk, because a trader who stops recording trades or starts trading outside the rules to add excitement has broken the test. A schedule helps: set fixed times for the session, the journal and the weekly review, and treat them as appointments rather than optional tasks.

Some traders find it useful to measure something other than profit during the test, such as the percentage of signals entered exactly as the rules describe. That measure improves with practice, rewards discipline, and is less affected by the random results of a few trades. It keeps attention on the process, which is what the paper test is meant to train.

Set a point at which the test will be reviewed, even if the sample is incomplete. A fixed review date gives the test a clear shape, and it prevents the sense of drifting that makes many paper tests end without a conclusion.

How do you separate the strategy from your own execution?

The paper test measures two things at once: the strategy and the trader’s ability to follow it. A good strategy can produce poor results if it is executed badly, and a weak strategy can look better than it is if the trader skips its losing signals. Separating the two is one of the main purposes of the journal.

Record the rule compliance for every trade. If the rule says to enter at the close and the trader entered two bars later, that is an execution deviation, and the trade should be marked as such. Analyse the deviating trades separately. If they perform worse, the lesson is about discipline. If they perform the same, the lesson is that the rule itself is robust to small timing errors.

This separation also helps when moving to live trading. A trader who knows that their execution slips during fast markets can build that into the plan, rather than discovering it with real money. It also makes the review fairer, because a poor result can be traced to a cause that the trader can actually change, instead of being written off as bad luck or a bad strategy.

What does this guide not promise?

This guide does not promise that any strategy will work when moved from paper to live trading. Paper trading reduces the risk of testing a rule, but it does not remove the risk of trading it. Markets change, costs change and human behaviour under real money differs from behaviour on a demo.

The guide also does not describe a single correct testing method for every trader. The practices here are a framework, and each trader should adapt the sample size, the criteria and the costs to their own circumstances and the instrument they trade.

What is the offer in this guide?

The offer is the EA demo testing workbook, a printable workbook for planning, recording and reviewing a demo test. It covers the rules, the costs, the sample and the criteria, and it has space for the skipped signals and the review. It is designed to be used with the paper trading practices in this guide.

Keep the workbook with your demo journal, and complete it for each test you run. The completed workbook is the record that shows whether the criteria were met before any live money was used.

How do you start this week?

Start by writing the rules for one strategy on one page, with the date. Include the market, the timeframe, the entry, stop, target, exit, position size, daily trade limit and news rule. Then write the sample size and the criteria for going live.

Next, set up a demo account with the spread and commission you expect live, and begin the journal with the columns described in this guide. Record every signal, taken or skipped. When the sample is complete, check the result against the criteria and write the decision in the journal.

For the arithmetic of stops and targets, use the risk-reward guide, and for the statistics behind a sample of trades, the Monte Carlo testing guide.

For more free tools and checklists, visit the resource library.