A trading plan is useful when it can prevent you from quietly changing the question after seeing a chart. It does not need to predict what a coin will do. It needs to record what you are investigating, what decision rule you will use, what risks that model includes and which result would make you stop.
Plan a research exercise, not a promised return. Write down a flat or “no trade” outcome as a legitimate decision. Nothing about a blank form or a filled example should imply that one particular instrument or risk amount is suitable for the reader.
Part 1: define the question
Write one sentence describing an observable condition and a possible mechanism. Replace “this setup looks strong” with something you can identify from dated data, such as “I want to investigate whether this completed-bar range break changes subsequent trade outcomes after stated costs.” Note alternative explanations: the signal might reflect the larger market trend, one exceptional event, a data-selection decision, or an assumed fill unavailable to a real order.
Name the asset and exact instrument. BTC spot and a BTC perpetual are not interchangeable exposures. Record the venue, price source, timeframe, UTC date range and data resolution. If the data source contains only hourly candles, do not claim that it validates second-level execution.
Part 2: make the rule executable on paper
List the entry condition, the precise time it becomes known, a simulated entry convention and a maximum holding window. Define invalidation and exit rules before looking at a large results table. Save a strategy version so its previous results can be reproduced. Add position-sizing math and a maximum notional. State the fees, spread, slippage, borrow or funding inputs; keep unknown values unknown. Include a baseline for comparison and a later evaluation period that you reserve before tuning.
A fictional research question can be precise without becoming a trading recommendation: “After a completed one-hour close above the prior 20-bar high, does the next-bar-open model show different four-bar outcomes from the named baseline after the same costs?” The plan still needs to identify the contract, data source and dates; define what counts as a valid signal, missing bar and open position; and choose how the baseline and costs are calculated. Those example lookbacks and horizons are arbitrary teaching choices, not a tested setup. A precisely written hypothesis can still fail.
For a linear example, size depends on the selected amount at risk and the entry-to-stop price distance. This gives a planned-loss estimate; it does not bound the actual result if the market gaps or the order executes worse. A venue’s initial margin, maintenance margin and liquidation use separate rules, which you must verify for the chosen contract and account mode.
Write risk limits before sizing. State the maximum loss you are prepared to model on one position, how simultaneous or closely related positions share that limit, and what happens when market data, fee rates or contract rules cannot be verified. A blank or stale cost input is unknown, not zero. If a required check fails, the plan should lead to a pause or a flat decision rather than an improvised number.
Margin is not a substitute for a loss budget. CME’s individual-futures education separates the number of contracts selected from the maximum allowed by initial-margin requirements; it recommends thinking in risk scenarios for the chosen size. That material describes CME futures, so use it as a planning principle and check the actual crypto contract’s own multiplier, margin tiers and liquidation rules. The CFTC warns that leverage in virtual-currency futures can amplify losses as well as gains. Name the full notional exposure, collateral and possible loss separately.
Order instructions also matter to the plan. Coinbase’s current Advanced Trade documentation says a stop-limit order posts a limit after its trigger, while its bracket and take-profit/stop-loss downside protection may fail during extreme volatility. Those mechanics are specific to that product and can change. Record the chosen order type, trigger source, limit if any, partial-fill handling and the response if an exit remains open. A stop level is not evidence that an order filled there.
Part 3: define what could change your mind
The failure condition should be observable, time-bounded and tied to the hypothesis. Examples include no persistence after the completed breakout bar, cost assumptions consuming the measured movement, failure in a later regime, or performance driven by very few trades. A plan should say whether that result means reject the idea, gather different data or design a new test. A new rule is a new version, not a correction silently applied to the old record.
A reusable pre-test card
| Record before testing | Why it helps |
|---|---|
| Hypothesis and alternative explanation | Makes the intended mechanism falsifiable |
| Instrument, timeframe, data source and dates | Defines what the history represents |
| Signal timing, entry, exit and hold limit | Prevents assuming unavailable future prices |
| Risk budget, size, costs and model gaps | Exposes actual assumptions |
| Benchmark, parameter search and reserved period | Shows how selection affects evaluation |
| Failure condition and next review date | Keeps rejection as an available conclusion |
After the test, add results, drawdown, actual historical trade count, costs, ledger provenance and important exceptions without overwriting the original. Later, record paper observations with fill-model assumptions visibly separate from live fills.
Make the review decision just as explicit as the entry. A missing or corrupted segment makes the affected result unassessable; it does not count as a zero-cost win or loss. A rule that misses its prewritten test condition has answered the question as defined. If you revise the entry, costs, stop or evaluation window after seeing results, save a new version and describe the earlier test as used during selection.
Use the plan to challenge, not defend, the idea
Try additional costs and small parameter changes. Inspect whether one asset, one date range or the best trades account for most of the outcome. If you looked at a later segment several times and changed a rule in response, record that it is no longer a clean holdout. Consider a “stay flat” observation when the conditions do not meet your own written rule.
SuperQuant’s workspace supports editable research rules, reproducible local backtests and manual paper practice. The included histories are synthetic. Its results cannot establish real venue execution or guarantee future performance. The plan template is a record of your own hypotheses; filling it in will not make them valid.
Start with the blank crypto trading plan template, and keep decisions and outcomes in distinct fields in the trading journal guide. Before testing a chosen quantity, review position sizing and the instrument’s current venue rules.
This plan-writing guide is educational and not personal investment advice. Crypto is volatile and may result in loss; neither this guide nor any example advises the reader to trade.