What does +1R mean in one trade?
R is one unit of the price risk planned before costs. If your chosen position loses $50 when its stop 10 pips away is hit, 1R is $50. A $75 gain before costs is +1.5R; a $50 price loss before costs is −1R. R lets you compare trades of different sizes without pretending their dollar results are equal.
Consider a hypothetical EUR/USD buy at 1.1000 after a completed signal candle. With a $5,000 account, a planned $50 price loss (1% before costs), a 10-pip stop at 1.0990 and a position worth $5 per pip, the target at 1.1015 is 15 pips away. Allow one pip ($5) for the combined round-trip spread, commission and slippage. The next-bar fill and every price in this example are assumptions, not observed quotes.
| Exit | Price move | Before costs | After $5 costs |
|---|---|---|---|
| Target at 1.1015 | +15 pips | +$75 / +1.5R | +$70 / +1.4R |
| Stop at 1.0990 | −10 pips | −$50 / −1R | −$55 / −1.1R |
A winner followed by a loser would total +$15, or +0.3R. The loss at the planned stop is $55 (−1.1R) after costs; a gap or worse fill could increase it. These are alternative example outcomes, not trades we claim happened.
Try your own numbers
Use pips for all three distance fields in Forex. For crypto you can use percentage points for all three; keep the unit consistent. Enter the signed price move in your favour: positive for a gain, negative for a loss. The calculator uses your planned cash risk before costs.
Crypto example: a 2% stop, a +3% favourable price move and 0.2% in round-trip costs give (+3 − 0.2) ÷ 2 = +1.4R.
Enter valid numbers: account, risk and stop above zero; risk no greater than the account; cost zero or more.
This is an arithmetic worksheet, not a broker position-size quote. Verify pip value or percentage exposure, spread, fees and slippage with your broker. A stop can fill beyond its level; the actual loss may exceed your planned R.
Begin with a question you could disprove
“Buy when the chart looks strong” cannot be replicated. Name the market, hours, data frequency, signal, entry, exit, sizing and rule for no trade. For example: at the end of a completed 15-minute candle, enter only if a defined London range was crossed; place the order on the next candle. The rule must say what happens when the market gaps past the intended price.
Record the idea and its economic rationale before opening the performance report. A credible hypothesis might concern a recurring session change, but it still needs to survive different years and execution assumptions. Changing the stop after seeing losing trades produces a new strategy. Keep the previous version and the reason for the change.
Freeze the protocol before you count wins
Write a short research contract: instruments, exact dates, timezone and daylight-saving handling, missing data, position size, maximum simultaneous positions, stop and target priority, and treatment of weekends or exchange outages. Define a baseline such as holding BTC or staying in cash over precisely the same dates. If ten parameter settings were tried, say so; the best setting is not the only experiment.
Store the raw input and rule version with a timestamp. Never repair a poor result by quietly excluding an inconvenient year, asset or trade. A predeclared reason to exclude bad candles is legitimate; a reason invented after seeing the profit is selection bias. The goal is a test another researcher can reproduce, including the unattractive trades.
Put every fact on its true timeline
A completed daily close can guide tomorrow’s order, not an order filled earlier in that same candle. News values have publication times; revised values may be different from what traders first saw. A signal calculated from a 15-minute closing candle must not capture a fill at its earlier high or low as though the close were already known.
For crypto, decide whether the clock is UTC and how to handle continuous weekends. For Forex, convert sessions using the relevant timezone, including daylight saving changes. Missing quotes, changing symbol conventions and assets that stopped trading matter too. Selecting only today’s surviving coins silently imports knowledge from the future.
Model an order, not an idealised chart mark
Choose order type, order submission time and fill rule. A stop order can slip through its trigger; a limit order touched by the candle is not necessarily filled. If stop and target both appear inside one OHLC bar, you do not know which happened first without finer data. Treat uncertainty from coarse candle data conservatively, or flag the ambiguous trade instead of awarding yourself the profitable outcome.
A bid-only Forex series cannot reveal the historical ask or the exact spread you paid. An exchange candle does not prove that your full size traded at its open. State the assumptions, compare with actual broker or exchange records if available, and allow missed fills. A simulation that gets perfect execution during thin or volatile periods is answering the wrong question.
Let costs challenge the apparent edge
Here is an original arithmetic check, not a claim about any real market: among 100 hypothetical trades, 52 win +1R and 48 lose −1R before expenses. Gross result is 52R − 48R = +4R, or +0.04R per trade. Subtract a combined round-trip spread, commission and slippage of 0.05R on each trade and the result becomes 4R − 100 × 0.05R = −1R.
| Round-trip cost per trade | Net result after costs |
|---|---|
| 0R | +4R |
| 0.02R | +2R |
| 0.05R | −1R |
| 0.10R | −6R |
Arithmetic example, not historical performance. 52 trades gain +1R each, 48 lose −1R each; R is the planned risk per trade before transaction costs.
The win rate stays 52%, even though the net expectancy turns negative. At 0.04R per trade the result is exactly zero; the table tests other assumptions. Express costs relative to each trade’s planned risk, then vary them for calm and stressed sessions. A one-pip allowance or a flat exchange fee is an assumption until actual execution records support it.
Protect data the strategy has never seen
Split development and evaluation chronologically. Build and revise the rule on the first period; freeze it before measuring the later, untouched period. Repeatedly peeking at the holdout and adjusting the rule converts that holdout into training data. Walk-forward testing can reselect parameters on a trailing window, but the schedule and permitted choices must be fixed in advance.
A favourable holdout is evidence, not a certificate. Check whether several market regimes, session conditions and instruments contribute; one extraordinary month can dominate a long chart. Keep a dated paper-trading record after the historical test and compare expected versus actual fills and costs. Live monitoring is a new observation, not a replay of the backtest.
Read the whole scorecard, not just total return
Report the sample dates, number of trades, win rate, average win and loss in R, gross and net expectancy, worst drawdown, exposure, turnover and a relevant benchmark. Also show results by year or regime and a cost sensitivity range. A return without position size or drawdown hides how hard it was to stay invested. A high win rate can coexist with large occasional losses.
Small samples need especially modest claims. Thirty trades across news events may expose a promising failure mode without estimating a stable future edge. Check whether a handful of trades account for most of the gains, whether the worst period clusters, and whether nearby parameter choices behave similarly. If a single precise setting produces all the profits, investigate fitting.
Compare three published cases, including their limits
Our EUR/USD London breakout study reports 550 modelled trades on 15-minute bid data and a one-pip round-trip allowance; its next-candle rule and London clock make the test inspectable, while real fills remain unobserved. Our NFP study used 32 release dates: after an assumed 1.2-pip allowance, EUR/USD had 16 entries and −1.763R, while USD/JPY had 22 entries and −2.236R. Neither assumed cost is a record of execution.
The BTC/ETH monthly rotation study uses the prior completed close to compare 90-day returns, then switches at the next open with modelled fees. It reports +2,098.14% over its sample but also a −71.48% maximum drawdown; much of its growth occurred before 2022. These are results of those specific historical rules, not independent confirmation that the approach will pay in the future. Follow the three internal reports for dates, inputs and caveats.
Decide what would make you reject the idea
Before committing capital, write a failure condition: net expectancy no longer positive under a plausible cost increase, poor performance on unseen data, unacceptable drawdown, reliance on one asset or year, or a gap between simulated and observed fills. Keep an experiment log with rejected variants; otherwise only the winner survives in the story.
A useful backtest can save you from a bad strategy. If the evidence is mixed, lower the claim instead of making the chart louder: identify exactly which markets and dates were tested, which assumptions drive the answer, and what new data would change your mind. The best research result is a decision you can explain and revisit.
A six-line plan to repeat the test
Use this as a record for each signal, including the trades you would rather forget:
- 01Freeze one rule and write down the asset, time zone, test dates, signal candle and no-trade conditions.
- 02Record the signal time and the later order time; never give an order a price from before the signal was known.
- 03Log order type, entry, planned stop and exit, including orders that never filled.
- 04Record spread, commission and slippage separately; translate their combined round-trip cost into the same unit as the stop.
- 05Calculate gross R = price move ÷ stop distance and net R = (price move − cost) ÷ stop distance for every trade.
- 06Freeze the strategy before opening the later holdout, then compare net results and drawdowns with a simple benchmark on the same dates.
The decision rule
Treat an impressive backtest as a hypothesis. Publish the exact rule, reserve data you have not used to choose it, apply costs that fit the market and compare it with a simple benchmark. If the edge disappears under a small, plausible change, the result has not earned a claim of robustness.
Risk note
Trading and investing can lead to losses. Historical and hypothetical outcomes cannot guarantee future returns. Slippage, gaps, leverage, liquidity and taxes can make realised results materially different from a simulation. This article is educational, not individual financial advice.
Two references for the method
Our worked ledger is calculated here. These two primary documentation pages explain out-of-sample and survivorship concerns, and how an execution simulator treats fees, slippage and fills. Neither is evidence that the illustrative strategy works.