Strategy comparison
Compare standard, own and licensed strategies with the same data and risk settings. Each has a separate model account, not a combined portfolio. Completed results remain visible when you stop a comparison.
Would your strategy have passed the test?
Test your rules against historical price data. Compare results, examine losses and try different settings.
Turn a trading idea into a useful test.
A backtest applies a defined rule version to available historical candles. A useful result tells you what happened, under which assumptions, and where your idea struggled.
Choose your data
Select an instrument, price source, timeframe and UTC date range. The archive's first and last dates do not guarantee uninterrupted coverage. The report identifies the data actually evaluated.
Fix your rules
Select an authorised standard, own or licensed strategy version. Depending on access, a single test can add up to two confirmation strategies and three indicator conditions. Confirmations must agree in direction within the chosen window of closed candles.
Set the simulation
Define starting capital, risk per trade, stop method, reward-to-risk ratio and maximum holding period. Enter commission and slippage deliberately: both start at zero.
Investigate the result
Read capital and drawdown alongside closed trades and open positions. Inspect individual trades in the chart, compare variants under the same assumptions and export closed trades as CSV.
You decide what the test should assume.
- Risk and position size
- Position size uses the current realised balance, risk percentage and distance from entry to stop. The model account has a normalised unit; broker account currency, leverage and margin are not simulated.
- Stop and target
- Choose a strategy anchor, ATR distance or percentage distance, with an optional ATR buffer. A missing strategy anchor is unavailable. Targets use a reward-to-risk ratio of 1:1, 1:2 or 1:3 from the actual simulated entry.
- Commission and slippage
- Commission in basis points is charged on the executed entry and exit value. Slippage moves both prices against the position. The costs metric shows commission; slippage is already included in prices and results.
- Holding time and extra filters
- Limit how many candles a position may stay open. Add supported indicator conditions or confirmation strategies without changing the original version. Missing indicator values do not pass a condition.
Read more than the final return.
Read metrics together with the number of observations, the period and the model assumptions.
A rising curve can still contain difficult phases.
Capital and drawdown share one time axis in the report. Inspect the path between peaks, deep declines and time below a previous high. The final point alone cannot explain that experience.
- Capital
- Previous high
- Drawdown
Time
Net return and net profit/loss
Net return includes the marked profit or loss of a position still open at the end. Realised net profit/loss includes closed trades and commission, including entry commission on an open position. The capital overview separates balance, open profit/loss and total equity.
Drawdown
Drawdown measures decline from the highest equity reached so far at observed candle closes. It helps explain difficult phases. Losses occurring only within a candle are not fully represented.
TP and SL frequency
These percentages use all closed trades as their denominator. Time exits, session exits and other reasons count separately: TP and SL need not total 100%. Historical frequencies are not future win probabilities.
Profit factor and expectancy in R
Profit factor compares positive and negative net trade results. Expectancy in R averages closed-trade results relative to each trade's initial risk. Undefined ratios are shown as unavailable.
Sample size and loss streaks
Inspect closed trades, wins, losses, neutral outcomes and consecutive losses. A result from few trades can be fragile. Long and short statistics reveal differences between directions.
Time in the market
Exposure and holding duration use elapsed calendar time. Duration distributions, exit reasons and monthly closed-trade profit/loss complement the full equity calculation.
How a signal becomes a simulated trade.
The execution model explains its assumptions; a historical candle does not guarantee a real fill.
Confirm at the close
Only closed, valid candles and confirmed rule events can trigger a trade. Entry is at the next available contiguous open; signals before the chosen start are not traded.
Follow the position
Each model account holds one position at a time. Further signals while it is open count as skipped. If one OHLC candle touches both stop and target, the stop takes priority.
Explain the exit
A stop gap uses the actual open; a favourable target gap is settled conservatively at the target. Holding limits and session boundaries can close a trade. Before a genuine data gap, the model exits at the last observed close.
Compare ideas. Keep validation independent.
Change one hypothesis at a time and keep the rest of the setup comparable. A better-looking result is a reason to investigate, not proof of a better strategy.
Parameter optimisation
Search permitted parameter ranges manually or automatically within a test budget, using up to three search axes. Ranking uses training only. The best tested setting is not a global optimum or a promise of future gains.
Training and validation
Choose where validation starts. Selection uses earlier training data; later validation tests selected settings without determining their rank. Missing coverage is reported. Applying a setting changes the configuration; you then start a new full backtest.
Validation starts
Training
Validation
Understand a trade instead of trusting a marker.
Open a closed trade from the list to inspect candles, confirmed signal, entry, stop, target and exit. The explanation shows the available rule summary, filters and execution evidence. Protected licensed rules stay private. CSV contains trade results and execution details without private rule graphs.
A practical experiment: change the exit.
Keep entry rules unchanged and compare two supported stop or target configurations. Record what you changed and which question the test should answer.
Use the same instrument, source, period and cost assumptions for both runs.
Compare return, drawdown, sample size and exit reasons together.
Check whether a few exceptional trades dominate the difference.
Investigate neighbouring parameters and a separate validation period before drawing conclusions.
Know what the simulation leaves open.
The archive sets the boundary
Missing session, event history, reference, volume or timeframe prerequisites can make a strategy unavailable. Data gaps are not filled with another source. Tick volume remains tick volume.
Open positions stay visible
A normal backtest leaves a position open at the data end and marks its result separately. Optimisation closes remaining positions at the last observed close of each training or validation range. That period-end exit is not a TP or SL hit.
The model is not a broker fill
OHLC cannot reveal the precise price order within a candle. Contract specifications, currency conversion, leverage, margin, financing and swap costs are not simulated. Commission and slippage do not reproduce every execution cost.
Historical evidence has limits
Access depends on your plan, data rights and authorised versions. Unverified stock adjustments and missing total-return data matter. Historical simulation can reveal weaknesses; it cannot guarantee future profits.
Frequently asked questions
Why can two tests give different results?
Different sources, periods, timeframes, rules or execution assumptions can change the outcome. Compare the settings and available history before attributing the difference to an improvement in the strategy.
Are spreads, fees and slippage always included?
Commission and adverse slippage are configurable and default to zero. The costs metric shows commission; slippage is already reflected in executed prices. A separate spread model, financing and swap costs are not included. Check the assumptions before comparing results.
Does a successful backtest predict future profits?
No. A backtest describes a historical simulation under specified assumptions. Future market conditions and real execution can differ. Use the result to challenge the idea and understand its weaknesses, not as a promise.