TradingView Backtest & Robustness Lab

About TradingView Backtest & Robustness Lab

A backtest should not stop at the question “did it make money in the past?” Data quality, trading costs, the depth and duration of losses, trade ordering, out-of-sample degradation and stability around neighboring parameter values are separate questions. The Lab re-aggregates imported backtest results in one workspace so you can identify which assumptions a result depends on and inspect weaknesses hidden by one attractive headline number.

Intended users

Traders who use TradingView Strategy Tester, Pine Script strategy developers, and individuals or organizations that repeatedly save, compare, validate and report on backtests.

Intended data

Strategy Tester trade lists or summaries and generic CSV/XLSX files with equivalent fields. The available analysis depends on the columns and granularity in the file.

What it does

Mechanically, statistically and descriptively aggregates past trade results, then visualizes, diagnoses, compares assumptions and produces reports.

What it does not do

Investment advice, trade timing, instrument, size or parameter recommendations, price forecasts, real-time market data, auto-trading, order execution or guarantees of future results.

“Robustness” is not an absolute pass certificate. It is a way to describe how a result changes as different validation layers challenge it, within the limits of the imported data and selected settings.

What you can examine with the Lab

The analyses that can be produced depend on the imported columns, period, trade count, timestamp granularity and cost settings. An aggregate-only file cannot recreate trade order, intratrade MFE/MAE, sessions or a common time axis across several strategies.

Files and data you can import

The standard input formats are CSV and XLSX. If Strategy Tester column names differ by language or version, generic column mapping can match source columns to analysis fields. The Lab also supports multi-file import.

Record these assumptions before import

• Backtest start and end dates, instrument, timeframe and time zone
• Initial capital, base currency and position-sizing method, such as fixed, percentage or compounding
• Whether commission, spread, slippage, financing and other costs are included
• Open trades, duplicate rows, missing values, encoding, delimiter and date format
• Look-ahead, repainting, synthetic bars and fill assumptions in the source backtest

Data source, as-of dates and refresh timing

The Lab’s data source is the backtest file you import; it is not a service that periodically distributes live market data. Record the backtest start and end dates, trade timestamps and time zone, file-export date, analysis date and report-generation date separately. Re-export and re-import whenever the strategy version, parameters, instrument, timeframe, initial capital, cost assumptions or source data change. There is therefore no fixed refresh frequency: the as-of point of every result depends on the imported file and analysis settings.

A chart-data CSV contains OHLC price history, not a trade history. The Lab does not import Pine Script source, screenshots, PDFs or a real-time price feed and automatically decide whether a strategy is valid. See the Strategy Tester CSV export guide for the practical export workflow.

Analysis results and outputs

Under the current plan description, analyses, file imports, available diagnostic runs and report generation are unlimited within the feature scope of each plan. Save capacity, available diagnostics and report formats still differ by plan.

How to use the Lab

1. Lock the assumptions.Record the instrument, timeframe, period, initial capital, sizing, commission, slippage and time zone.
2. Export the trade list.Open List of Trades in TradingView Strategy Tester and save it as CSV or XLSX. Labels and export behavior can change, so confirm the current interface and official help.
3. Import the file.Drag and drop or select the file, then map date, direction, prices, quantity, P&L and other columns when needed.
4. Review quality warnings first.Check missing or duplicate rows, timestamps, units, open trades, sample size, outliers and cost fields; return to the source when a repair is required.
5. Read KPIs together.Do not stop at net profit or win rate; combine expectancy, Profit Factor, trade count, maximum drawdown, Sharpe and the return distribution.
6. Challenge the assumptions.Run Monte Carlo, risk-of-ruin, cost-sensitivity and stress checks to move from one path to a distribution of possible paths under the model.
7. Inspect OOS and neighboring values.When needed, add OOS or walk-forward analysis, parameter stability, time or regime checks and multi-strategy analysis.
8. Keep a comparable record.Save or export the settings, strategy version, data period and run date with the result so the validation can be reviewed later.

A practical order is: data quality → KPIs → drawdown → Monte Carlo and stress → costs → OOS → parameter neighborhood → multiple strategies. Lock each earlier assumption before relying on a later layer.

Metrics and diagnostic logic

Metrics are computed mechanically from the imported trade fields and selected settings. Diagnostic comments are rule-based. No single number or color is a universal pass or fail; compare results only when definitions, units, periods and cost assumptions are aligned.

Display precision, annualization, missing-value handling, the risk-free rate, ruin threshold and simulation method can change a result. Attach settings, period, units and data version to every report and reconcile differences between tools by checking definitions first.

Sample analysis

The following values are fictional educational data used only to demonstrate how to read several metrics together. They are not a real strategy, actual performance, recommended parameters or user averages.

How to read this result

1. Win rate does not reject it by itself.Although 47% is below half, the fictional average win is larger than the average loss and expectancy is positive.
2. Profit Factor does not approve it by itself.A PF of 1.38 means gross profit exceeded gross loss in this past sample, but it must be read beside the −15.3% maximum drawdown and recovery time.
3. Trade count is only a starting point.Check whether 214 trades are concentrated in one regime or depend on a few exceptional winners.
4. Challenge costs and sequence.Raise commission and slippage assumptions, then use Monte Carlo to examine losing-streak and drawdown distributions.
5. Finish with unseen data.Do not call the strategy robust until OOS, walk-forward and neighboring-parameter checks have been reviewed.

The sample conclusion is not “good” or “bad.” It is: the in-sample profitability looks positive, but loss behavior, costs, distribution, OOS and neighboring parameters still need to be challenged. Use the Lab’s diagnostics to identify the next assumption to test, not as a final trading decision.

Free and paid plans compared

Free is for correctly aggregating the file and seeing the core picture. Pro is for saving the work, running deeper practical diagnostics and exporting it. Premium is for extending the workflow into out-of-sample periods, neighboring parameters, time, multiple strategies and recurring reports. The table is a summary reviewed on July 15, 2026. The current plans page takes precedence.

No plan includes investment advice, trade signals, forecasts, guaranteed win rates or returns, auto-trading, order execution or real-time market data. Annual billing is priced at ten monthly payments, equivalent to two months off.

Paid features: turn a one-off result into a reproducible validation workflow

The value of a paid plan is not only a longer metric list. Pro turns a check that disappears after each session into a saved, comparable and exportable record. Premium extends the workflow across out-of-sample data, neighboring parameters, time and relationships among several strategies.

How your data is handled

Transmission of analysis data

CSV and XLSX analysis data is not transmitted to SG Group. Analysis follows a local-first design.

Vault and storage location

Pro and above can manage saved analyses and report templates in the encrypted Vault. When server Vault is used, data is stored in the customer’s own WordPress environment.

User responsibility

You are responsible for rights to the source file, removing confidential data, backups, passphrases and safe storage of local workspace and Vault files.

What calculation verification means

Cross-checking against independent reference implementations verifies calculation processing and output values; it does not validate the strategy or guarantee future profitability or live fills.

Data-safety checklist

• Remove names, account numbers, email addresses, order IDs and other personal or confidential fields that the analysis does not need.
• Keep the original file separately and import a cleaned copy. Export important work so browser-data removal, device failure or a lost passphrase does not leave you without a recoverable copy.
• Before using third-party or organizational data, confirm contracts, internal rules, confidentiality obligations and licenses.
• Attach period, currency, initial capital, costs, time zone, data version, missing-value treatment and analysis settings to shared reports.

“Not transmitted to SG Group” describes the standard CSV/XLSX analysis data. Stripe billing, contact forms and a server Vault in the customer’s own WordPress environment are separate functions with their own communications and terms.

Update history

• July 15, 2026 — Added the static guide and organized inputs, outputs, workflow, metric logic, the sample, plans, data handling, FAQs, related reading and the disclaimer.
• July 13, 2026 — Published or updated the ten English and ten Japanese articles in the Backtest Validation Series.

No dedicated public product changelog was confirmed at the time of review, so this section records changes to the static copy. The live tool and current plans page take precedence for features, prices, permissions and verification figures.

Disclaimer

TradingView Backtest & Robustness Lab is software for educational, informational and validation-support purposes. It mechanically, statistically and descriptively aggregates, diagnoses, compares and visualizes historical backtest results imported by the user. It does not provide investment advice, trade signals, solicitation, instrument, size or parameter recommendations, entries, exits, price forecasts, personalized portfolio advice, managed trading or guaranteed returns.

• Backtests are results under past data and assumptions. They do not indicate or guarantee future prices, win rates, returns, maximum loss, fills or live performance.
• Results change with data quality, period, instrument, timeframe, sizing, initial capital, currency, commissions, spread, slippage, taxes or financing, fill models, look-ahead, repainting and market regime.
• Monte Carlo, risk of ruin, VaR, CVaR, robustness and OOS scores, parameter stability, correlation and allocations are conditional reference values under the imported data, settings and model. They do not cover every unknown future event.
• You are responsible for source-file accuracy and rights, removal of personal or confidential data, backups, passphrases and safe management of saved files.
• TradingView is a trademark of TradingView, Inc. The Lab is independently provided by SG Group and is not provided, approved or sponsored by TradingView, Inc.
• Prices, plans, features, calculation specifications and output formats may change without notice. Confirm the current interface and plans page.
• Trading and investing can result in the loss of some or all capital, and leveraged products can produce losses beyond the amount deposited. Final decisions and all resulting profit, loss, costs and tax consequences remain the user’s responsibility.