TradingView Sharpe and Sortino ratios: compare the calculation first

What the official Strategy Tester documentation defines

TradingView's Sharpe help page defines SR = (MR − RFR) / SD using monthly account-equity returns. Its Sortino page replaces SD with downside deviation. The annual benchmark is set by the risk_free_rate parameter in the strategy declaration. The official RiskMetrics library's update notes confirm monthly periods are used consistently; older descriptions of switching to daily returns on short histories should not be applied to the current calculation.

A reproducible monthly-return example

This synthetic arithmetic example uses four monthly account returns: 2%, −1%, 3%, −2%. Set both the risk-free rate and the downside target to zero. Use population standard deviation and divide downside squared shortfalls by all four months. These are illustrative inputs, not a measured TradingView strategy result. Using a zero benchmark removes rate-conversion ambiguity from the example.

• The annualized row is a separate convention, not an assertion that TradingView displays those numbers. Square-root scaling assumes a return process for which variance scales with time; serial dependence can invalidate it.
• For an independent spreadsheet check, use AVERAGE and STDEV.P for Sharpe. For Sortino, square min(return − target, 0) for every row, average those values, then take the square root.
• For a nonzero benchmark, document the conversion. Simple monthly allocation of 2% is 2% / 12; an effective monthly rate is (1.02)^(1/12) − 1. These are different conventions.

Why Pineify can report a different Sharpe or Sortino

The current Backtest Deep Report implementation reads the exported P&L percentage for each parsed trade. It uses population standard deviation and scales the trade ratio by sqrt(trades per year), measured from the first entry to the latest exit using 365.25 days per year. Its Sortino denominator uses the root mean square of negative trade returns only. This is a trade-distribution analysis, not a reproduction of TradingView's monthly account-equity statistics.

• With the same four numerical returns treated as four trades spanning exactly one 365.25-day year, and risk-free input 0, Pineify's current formulas give annualized Sharpe 0.485071 and Sortino 0.632456. This is a calculation example, not an uploaded or live-tested report.
• A rolling window, a direction filter or inclusion of open trades changes the sample. Record those choices before comparing results.

Reconcile a trade CSV without losing the data boundaries

Export the Strategy Tester's List of Trades and open it in Backtest Deep Report. The parser expects trade identifiers and recognizable entry/exit types, pairs rows by trade ID, and reads P&L from the exit row. Keep timestamps and P&L percentage columns intact. A performance-summary export or a table of monthly equity values is a different input and should not be presented as a supported trade-list format.

• Check parsed trade count, first entry, last exit, direction and open-position treatment against the original report. The current parser can default missing numeric fields to zero, so a successful import alone does not prove complete data.
• For an independent monthly calculation, first obtain account equity at the same monthly boundaries, including the same treatment of unrealized P&L. Closed trade records alone do not reconstruct every equity observation.
• Align initial capital, currency, sizing, commission, slippage and test dates. Keep a copy of the original export and record the formula convention beside the result.
• Use the monthly spreadsheet calculation to reconcile monthly metrics. Use Pineify's report to inspect the parsed trade distribution; do not expect either calculation to substitute for the other.

Interpret the sample before interpreting the score

Four months are enough for the arithmetic above, but they provide little evidence about future performance. Hundreds of trades within those months do not create hundreds of independent monthly observations. Review the number of return periods, exposure, concentration of profits and losing periods before comparing strategies. There is no universal Sharpe or Sortino threshold that proves a strategy is ready for live trading.

• One monthly observation has no observed dispersion. Constant returns give a zero Sharpe denominator; no returns below the target give a zero downside denominator. An undefined or extreme result needs investigation.
• The linked Strategy Tester help pages do not specify a universal minimum trade count or all short-history display rules. Do not diagnose a blank ratio using an invented minimum such as 30 or 100 trades.
• A strategy selected from many parameter trials can have an inflated historical ratio. Preserve untouched evaluation data and review drawdown, costs and concentration alongside the ratio.
• Resampling past trades explores assumptions about that sample. It does not establish a genuine edge or guarantee that future market conditions will resemble it.