A plan, a journal and a backtest
The word plan and the sentence it holds
A plan is a sentence with a condition in it. It names the instrument, the direction, the size, the point where you would accept being wrong, and the point where you would take the result. On AUD/USD around 0.6640, a plan might read: buy 0.10 lots if the price closes above 0.6650, stop at 0.6635, target at 0.6680. That sentence can be checked before any money moves, because every number in it is a decision you made in advance.
The platform shows the same sentence in its own words. The order ticket has a field for volume, a field for stop loss and a field for take profit. The chart shows the price line and the two levels as horizontal marks. Reading the plan on the platform means matching each written number to a field or a line, so nothing is left to memory when the price starts moving.
A plan also states what would make you cancel it. If the price reaches the stop level before the entry condition is met, the plan is void. If the spread widens past a stated amount, the plan is void. Writing the cancellation rule down is what separates a plan from a hope, because a hope has no exit that you chose in advance.
The word journal and the columns it needs
A journal is a table, not a story. Each row is one trade, and each column is a fact that existed at a known moment. The columns that make a row useful are: date and time, instrument, direction, entry price, exit price, size in lots, stop level, target level, the reason for entry, and the outcome in account currency. A row missing the reason cannot be compared with any other row, because the rule that produced it is unknown.
On AUD/USD at 0.6640 with 0.10 lots, one pip is 1 unit of the quote currency, so the outcome column is arithmetic. If the entry was 0.6640 and the exit was 0.6655, the move is 15 pips and the gross result is 15 x 1 = A$15. If the exit was 0.6625, the move is 15 pips against and the gross result is -A$15. The journal stores the gross figure, and a separate column stores the costs the broker charged.
Costs vary between brokers, so the journal should record what your own account was charged rather than a rate copied from elsewhere. The spread at entry, any commission, and any swap for a position held overnight all belong in their own columns. Without them, a row that looks like a win may be a loss after costs, and fifty such rows cannot answer whether the rule worked.
The word backtest and the rule it tests
A backtest is the same journal run on prices that have already happened. You take one written rule, apply it to a past series of AUD/USD prices, and record the entry, exit, size and reason for each signal the rule would have produced. The point is not to predict the future. The point is to see whether the rule has a shape, such as many small losses and fewer larger wins, or the reverse.
The arithmetic is identical to the live journal. If the rule signals 50 trades on AUD/USD at 0.6640 with 0.10 lots and a 15 pip stop, each stopped trade is 15 x 1 = A$15 before costs. If 18 trades reached a 40 pip target, each is 40 x 1 = A$40. The backtest table then shows 18 x A$40 = A$720 and 32 x A$15 = A$480, a gross difference of A$240 before any spread or commission is subtracted.
A backtest is only as good as its assumptions. If you assume every signal filled at the exact price on the chart, the result is optimistic, because real fills can differ. If you assume a fixed spread, the result is a model, not a record. State the assumptions beside the table, and treat the output as a question about the rule, not as a forecast of your account.
Fifty trades under one rule
| Step | Amount | Note |
|---|---|---|
| Trades recorded | 50 | one rule, one instrument, one period |
| Wins | 18 | counted from the journal entries |
| Losses | 32 | counted from the journal entries |
| Average win | A$40 | 20 pips on 0.20 lots at 10 units per pip per standard lot, so 20 x 2 = A$40 |
| Average loss | A$20 | 10 pips on 0.20 lots, so 10 x 2 = A$20 |
| Gross from wins | A$720 | 18 x A$40 |
| Gross from losses | A$640 | 32 x A$20 |
| Net before costs | A$80 | A$720 minus A$640 |
The broker may round the pip value, charge a spread or commission on top, and quote a different conversion rate for the quote currency. Costs vary between brokers, so the net figure here is not the amount that would reach an account.
Doing it once, step by step
You will write one plan, keep one journal row, and run one backtest on the same rule, in that order, so the three artefacts refer to each other.
- 01
Write the rule as one sentence
Open a document and write a single sentence with an entry condition, a stop level, a target level and a size. Use AUD/USD around 0.6640 and 0.10 lots so the pip arithmetic is 1 unit per pip. Do not add a second condition until the first sentence is complete.
- 02
Set the levels on the platform
Find the order ticket and locate the volume, stop loss and take profit fields. Enter the numbers from your sentence and read them back against the chart. If a field cannot hold your number, the plan is not yet expressed in a way the platform can execute.
- 03
Open the journal table
Create columns for date, instrument, direction, entry, exit, size, stop, target, reason and gross outcome. Add two more columns for spread and commission. Leave the exit and outcome cells empty until a trade closes.
- 04
Record one row before the trade
Fill every column except exit and outcome before you place the order. The reason column must name the condition from your sentence, not a feeling. This row is now a prediction you can check.
- 05
Run the rule over past prices
Scroll the AUD/USD chart back and mark every point where your entry condition would have been met. For each signal, record the entry, the stop and the target in the same table format. Stop after 50 signals or when the chart runs out, whichever comes first.
- 06
Compare the two tables
Put the live row and the backtest rows side by side. Check whether the reason column describes the same rule in both. If it does not, the backtest tested a different rule from the one you traded, and neither table can answer whether the rule worked.
The mistake most people make here
The common mistake is to keep a journal that records only the result, such as a profit or a loss, and nothing about the rule that produced it. Then fifty entries cannot answer the only question that matters: did the rule work, or did one large win hide many small losses. Write down the entry, the exit, the size and the reason before the trade, and the outcome after it. If the journal cannot be turned into the table above, it is a diary, not a record.Check yourself
A journal shows 50 trades: 18 wins at an average of A$40 and 32 losses at an average of A$20. What is the net before costs?
18 x A$40 = A$720. 32 x A$20 = A$640. A$720 - A$640 = A$80 net before costs.
The same rule is tested on AUD/USD at 0.6640 with 0.10 lots and a 15 pip stop. What is the loss in A$ if the stop is hit?
One pip on one standard lot is 10 units of the quote currency, so 0.10 lots gives 1 unit per pip. 15 pips x 1 = A$15, before any spread or commission.
Questions that come up next
What does a backtest actually test, and what does it leave out?
A backtest applies one written rule to past prices and records what the rule would have produced. It tests the rule, not the person, because no order was sent and no emotion was present. It leaves out the spread at the moment you would have traded, any commission, and any slippage between the price on the chart and the price you would have received. It also leaves out the effect of your own hesitation, which is why a backtest result is a ceiling rather than a forecast. The figure it produces is evidence about the rule, and nothing more.
How many trades does a backtest need before it says anything useful?
A small number of trades can be produced by chance, so a rule tested on ten trades tells you very little. Fifty trades is a common starting point because it is enough to show whether a few large results are carrying the whole total. Even fifty is not proof, because the market conditions in that period may not repeat. The honest use of a backtest is to compare two rules over the same period, not to predict what the next fifty trades will do. Treat the number as a way to reject a bad rule, not to confirm a good one.
Where does the plan end and the journal begin?
The plan is written before the trade and contains only decisions you have already made. The journal is written after the trade and contains only facts that occurred, including the outcome and the costs. The reason for entry appears in both, which is what allows you to line the two documents up. If the reason in the journal does not match the reason in the plan, the trade was not the one you planned. That mismatch is the most useful thing a journal can reveal.
What if the price moves before the entry condition is met?
The plan should state this in advance. If the price reaches the stop level before the entry condition is met, the plan is void and there is no trade. If the spread widens past an amount you wrote down, the plan is void as well. Writing the cancellation rule down removes the decision from the moment, when the price is moving and the decision is hardest. A plan without a cancellation rule is not a plan, because it has no condition under which you would stand aside.
Do costs really change whether a rule worked?
Yes, and they change it more than most beginners expect. A rule with many small wins can be turned into a net loss by the spread and commission on each trade. Because costs vary between brokers, the only figure that matters is the one charged on your own account, which is why the journal keeps its own cost columns. A backtest that ignores costs will overstate the rule. The net figure after costs is the one to compare against the same rule with no costs, so you can see how much of the result the costs consumed.
Can the same journal serve two different rules?
Only if the reason column identifies which rule produced each trade. Two rules mixed in one table cannot be compared, because the rows do not belong to the same test. The clean approach is one table per rule, or one table with a rule column that lets you filter. Then fifty entries can answer whether one rule worked, which is the only question the journal exists to answer. A table that cannot be split by rule is a diary.
The words used on this page
- plan
- A written sentence that names the instrument, direction, size, stop level and target level before any money moves. It also states the conditions under which the trade is cancelled. Every number in it is a decision made in advance.
- journal
- A table in which each row is one trade and each column is a fact recorded at a known moment. It holds the entry, exit, size, reason and outcome, with costs in their own columns. A row without a reason cannot be compared with any other row.
- backtest
- The application of one written rule to past prices to see what the rule would have produced. It tests the rule rather than the person, because no order was sent. It leaves out costs and slippage, so its result is a ceiling rather than a forecast.
- pip value
- The money value of a one pip move for the size you are trading. On AUD/USD, one pip on one standard lot is 10 units of the quote currency, so 0.10 lots gives 1 unit per pip. Brokers may round this figure or convert it at a different rate.
- costs
- The spread at entry, any commission, and any swap charged for holding a position overnight. They vary between brokers, so the journal records what your own account was charged. A row that looks like a win may be a loss after costs.
- cancellation rule
- The condition written into the plan under which you would not take the trade at all. It might be the price reaching the stop before the entry condition is met, or the spread widening past a stated amount. It removes the decision from the moment when the price is moving.
What to take away
- A plan is written before the trade and holds only decisions you made in advance, including the condition that cancels it.
- A journal is a table of facts recorded after the trade, with the reason and the costs in their own columns, so rows can be compared.
- A backtest applies one rule to past prices and produces a ceiling, not a forecast, because it leaves out costs and slippage.
- Fifty entries can answer whether a rule worked only if the reason column identifies the rule that produced each trade.