Create a complete paper-trading plan

Choose a learning question, not a profit target

A paper-trading plan is a specification for practice without committing real capital. A useful objective is concrete: learn to calculate currency exposure, understand how a limit order behaves, or test whether a clearly defined condition has consistent historical characteristics. “Make a large return this month” encourages scorekeeping but does not identify which skill is being learned.

Choose one product type and learn its mechanics before combining several. A cash share, a CFD, a currency contract and a crypto perpetual can reference similar price movements while having different ownership, financing, liquidation and counterparty arrangements. Write the full product name and the source of its specifications. A symbol by itself is not enough.

Specify the exercise completely

Set a hypothetical starting balance, observation schedule, instrument universe and decision rules. State how entries and exits will be modeled, what conditions prohibit an entry, and how quantities are calculated. Define whether positions can remain open overnight or during scheduled announcements. These are practice assumptions, not recommended settings for a live account.

List estimated commissions, spreads, slippage, financing and currency-conversion costs where relevant. Record whether results are before tax; tax treatment depends on local rules and personal circumstances. A platform showing zero commission does not imply that every other cost is zero. If a required cost is unknown, mark the uncertainty rather than silently omitting it.

Define the simulator’s fill convention, not just the strategy’s entry rule. Does a market purchase use the observed ask, the next available quote or an artificial midpoint? Does a limit order fill whenever a chart touches its price, or only when the data support a plausible quantity? If the platform cannot model queue position or partial fills, note the limitation and test less favorable assumptions separately. A paper balance is only as informative as the transaction rules behind it.

Practice operational failures too

Include exercises that are inconvenient rather than exciting: an order rejected for size, a partial fill, a canceled entry, a price gap and a connection interruption. Document how the simulated position would be reconciled in each case. The practice is incomplete if it assumes every order is accepted immediately and every desired exit is available.

Do not automatically reset the account after a loss. Keep the loss in the record and continue according to the written exercise rules, or end the exercise with its actual result. Resetting erases the very sequence that tests whether assumptions are workable. If a rule changes, start a clearly labeled version rather than editing the historical record.

Evaluate what the simulation can and cannot establish

At the review date, summarize the number of observations, adherence to rules, net modeled result, largest drawdown, cost sensitivity and unresolved questions. Inspect whether a few exceptional outcomes dominate the result. Compare modeled fills with available market data and explain any optimistic assumptions. A short transparent report is more useful than a screenshot of a favorable balance.

Paper trading cannot reproduce the full emotional, liquidity or operational experience of real execution. It can establish familiarity and expose arithmetic errors, but it cannot establish that a person will achieve the same returns with money at risk. Finishing the exercise does not create an obligation to trade live. A valid conclusion is that the method is unclear, unprofitable under plausible costs, or not worth further attention.

Before calling the exercise complete, make sure every open simulated position has a recorded valuation or a specified closing procedure. Otherwise a report may count realized winners while leaving losing positions outside the result. Include the value of open positions, their costs and the pricing timestamp. That simple reconciliation keeps the end-of-exercise balance connected to the full record rather than to a selected set of completed trades.

Write a small experiment that another learner could repeat

A complete specification should fit together like a set of instructions, not a collection of intentions. State what is observed, when the observation is made, what triggers a decision, how much simulated exposure is permitted and what ends the position. Also state what happens when no signal occurs. A week with no qualifying trades is a valid observation, not a reason to relax the entry rule until the exercise becomes more entertaining.

For a hypothetical order-mechanics exercise, the learning objective could be to reconcile every partial fill and cancellation over twenty recorded order decisions. Success means the quantity ledger balances and every remaining order has a known state; profit is not the pass criterion. For a strategy-research exercise, the objective would be different and would require a larger evidence framework. Keeping those purposes separate prevents operational familiarity from being mistaken for proof of a trading edge.

Set the review and stopping rules before starting. A missing price feed, an unreconciled position or an unknown contract specification can justify pausing new simulated decisions while the defect is investigated. A losing result, by itself, should not trigger a retroactive rewrite of the original rule. If the exercise ends early, preserve the actual ending holdings, costs and reason. Do not quietly remove a failed period from the record and restart the performance chart at a more convenient date.

At the end, produce an evidence packet: the dated specification, complete decision log, cost assumptions, account reconciliation and a short assessment of what was learned. Separate three conclusions: the mechanics were understood, the procedure was followed, and the evidence supports an economic hypothesis. The first two can be established even when the third remains unsupported. That is a productive learning result, and it does not imply that the next step has to involve real money.