What is a good downside capture ratio?

Below 100% means a smaller aggregate loss than the benchmark under the specified method; a negative value means an aggregate gain during benchmark-down periods. Neither is a universal investment target. Check the benchmark fit, upside capture, absolute losses, exposure and sample length together.

Why do two services show different capture ratios?

They may use different benchmarks, frequencies, time windows, net or gross returns, arithmetic or geometric aggregation, or annualization. Compare the selected observations and both return measures before comparing the final ratios.

Does no downside capture mean the strategy never lost money?

No. An unavailable downside capture ratio may mean there were no benchmark-negative periods. A numeric zero means the selected strategy return measure was zero. The strategy can also lose during benchmark-positive or flat periods, which do not enter downside capture.

TradingView Backtesting

Upside and Downside Capture Ratios for Strategy Benchmarks

The downside capture ratio compares a strategy with its benchmark using only periods when the benchmark lost money. The upside capture ratio uses benchmark-positive periods. Divide the strategy return measure by the benchmark return measure for the same selected observations, then multiply by 100. Under the geometric-mean convention used here, downside capture of 50% means the strategy had half the benchmark's geometric average loss in that sample. It does not promise half the loss in the next decline.

See the decision guide

Align strategy and benchmark returns before filtering

Start with equally spaced whole-account returns and benchmark total returns covering identical start and end timestamps. Monthly observations are one common convention. Match currency and treatment of dividends, then state strategy trading costs, financing and fees. Remove the performance effect of deposits and withdrawals before using account returns.

Define U as all valid paired periods with benchmark return b[t] > 0, and D as those with b[t] < 0. Exactly zero benchmark returns belong to neither set. Keep their strategy returns in full-period performance reporting.
Keep a positive strategy return in D if the benchmark fell that period. Filtering on strategy losses instead answers a different question.
A missing observation is not a zero return. Repair the source or use common valid intervals; a two-month strategy return cannot be matched to a one-month benchmark return.
Use equity marked to market at each valuation boundary, including idle capital and open positions. Individual closed-trade percentages do not form a synchronized account-return series.
Choose a benchmark that fits the mandate before reviewing the ratios. Record its identifier and price-versus-total-return variant so the comparison can be repeated.

Specify the formula instead of relying on the label

For either selected set S, let n be its observation count. Use decimals for returns and require positive growth factors for this geometric calculation. Define g_strategy = [product over S of (1 + r[t])]^(1/n) - 1, and g_benchmark the same way using b[t]. Capture_geo = 100 × g_strategy / g_benchmark. The selected periods need not be consecutive; this is a conditional average, not the return of a continuous investment through the full window.

Averaging the individual ratios r[t] / b[t] gives another answer. It can give extreme weight to a period when the benchmark barely moved; it is not the ratio of average returns.
Geometric annualization does not cancel in a ratio. A report using annualized conditional returns can disagree with a report using monthly geometric means even with identical input data.
Both outputs are usually displayed on a 100-based scale: 50% capture is a ratio of 0.50. It is not a 50-percentage-point return difference.

Convention	Return measure in each series	Comparison rule
Geometric mean	product(1 + return)^(1/n) - 1	Divide the two geometric means for the same subset.
Arithmetic mean	sum(return) / n	Divide the two arithmetic means for the same subset.
Cumulative subset return	product(1 + return) - 1	A different aggregation; do not label it as the mean-based ratio.
Annualized geometric return	(1 + geometric mean)^m - 1	Apply the same m to both series, then divide. For monthly inputs m = 12.

Recalculate a five-month example

The following returns are hypothetical. They assume matching month-end valuations in one currency, reinvested income, strategy returns after assumed costs, and no external cash flows. There are two up months, two down months and one flat benchmark month. These five observations illustrate the formulas, not a strategy evaluation.

Upside benchmark mean = sqrt(1.10 × 1.20) - 1 = 14.8913%. Strategy mean = sqrt(1.08 × 1.25) - 1 = 16.1895%. Upside capture = 100 × 16.1895 / 14.8913 = 108.72%, using unrounded means.
Downside benchmark mean = sqrt(0.90 × 0.80) - 1 = -15.1472%. Strategy mean = sqrt(0.95 × 0.90) - 1 = -7.5338%. Downside capture = 100 × (-7.5338) / (-15.1472) = 49.74%, using unrounded means.
Arithmetic upside capture = 100 × ((8 + 25) / 2) / ((10 + 20) / 2) = 110%. Arithmetic downside capture = 100 × ((-5 - 10) / 2) / ((-10 - 20) / 2) = 50%.
The ratio of cumulative subset returns gives 100 × 35 / 32 = 109.375% upside and 100 × (-14.5) / (-28) = 51.7857% downside. These are not the geometric-mean results.
Annualizing each conditional geometric mean before division gives 117.80% upside and 70.80% downside. This extrapolation does not turn two selected months into twelve observed months.
In a spreadsheet, the downside calculation for this exact example is =100*(SQRT(0.95*0.90)-1)/(SQRT(0.90*0.80)-1). The month-5 strategy loss remains relevant to total performance even though neither capture ratio includes it.

Month	Benchmark	Strategy	Selected set
1	+10%	+8%	Upside
2	-10%	-5%	Downside
3	+20%	+25%	Upside
4	-20%	-10%	Downside
5	0%	-2%	Neither

Read the sign and check the denominator

With valid down periods and positive growth factors, the benchmark geometric mean is negative. Dividing two losses produces a positive capture ratio; dividing a strategy gain by that negative benchmark produces a negative ratio. The sign must be interpreted with the selected return measures.

Single-period sign check: benchmark -4% and strategy +1% gives 100 × 1 / (-4) = -25% downside capture. Benchmark +4% and strategy -1% gives -25% upside capture.
If D is empty, report N/A with n_down = 0. If U is empty, upside capture is N/A. One selected observation permits arithmetic but offers little evidence of repeatability.
A zero denominator is undefined. A tiny denominator is unstable: benchmark -0.01% and strategy -0.10% gives 1,000% downside capture despite a small absolute loss. Show the underlying returns.
This page requires every included return to exceed -100%. At exactly -100%, a finite product can reach zero, but later account returns after capital is exhausted need separate treatment. Below -100%, the usual positive-equity geometric interpretation fails; do not force a result.

Downside capture	Meaning for the selected sample	Limitation
Between 0% and 100%	Strategy lost less on the stated aggregate measure.	It may still have suffered a large loss in one period.
100%	Strategy and benchmark aggregate return measures match.	Their individual period returns need not match.
Above 100%	Strategy lost more on the stated aggregate measure.	Check leverage, costs and whether the benchmark fits.
0%	Strategy aggregate return is zero while benchmark return is negative.	It does not mean the sample contains no losing strategy periods.
Below 0%	Strategy aggregate return is positive during benchmark-down periods.	It is not evidence of guaranteed protection.

Separate capture from beta, correlation and drawdown

Capture asks how the strategy performed in observations selected by the benchmark's direction. Beta measures a regression slope, and correlation measures the strength and direction of linear association. Their usual full-sample forms do not apply the same conditional filter. Even a downside beta is a slope within a subset, not a ratio of subset average returns.

For a concrete counterexample, suppose strategy returns equal benchmark returns plus one percentage point each period. With a varying benchmark, raw-return regression beta is 1 and correlation is 1. For benchmark-down returns of -10% and -20%, strategy returns are -9% and -19%, producing arithmetic downside capture of 14 / 15 × 100 = 93.33%.
Maximum drawdown depends on the chronological equity path from peaks to troughs. Capture discards flat-benchmark observations and aggregates a subset; it cannot identify the deepest account loss.
High upside capture and low downside capture are favorable descriptions relative to the chosen benchmark and method. Compare exposure, leverage, full-window return and drawdown before drawing a broader conclusion.
Changing from daily to monthly data changes which returns enter each subset. A positive month can contain many negative days. Report sample counts and dates rather than comparing ratios across frequencies as if they measured the same sample.

Use TradingView trade exports alongside a separate benchmark series

Pineify Backtest Deep Report accepts TradingView List of Trades CSV files and provides supported trade-based return and risk results. Use it to inspect the imported trades and account assumptions. A trade ledger alone does not contain the benchmark history or every marked-to-market account valuation required for capture ratios.

Keep the selected benchmark series and a synchronized strategy equity series separately. Derive interval returns, apply the benchmark filter, then calculate capture in a spreadsheet or research script.
Do not treat trade closing dates as the only observation dates or fill periods without closed trades with zero account returns. Open positions may have changed value.
This page supplies an independent calculation; it does not claim that Backtest Deep Report currently includes upside or downside capture.
For a custom Pine implementation, specify the benchmark, valuation frequency, equity definition, zero and missing-period policies, and aggregation convention to Pine Script AI Coding Agent. Review the generated code and reconcile each selected observation before relying on the statistic.

Where Pineify fits

Backtest Deep Report

Import a TradingView trade CSV to inspect supported return and risk results. Calculate capture separately from aligned equity and benchmark returns.

Also useful: Pine Script AI Coding Agent. Describe custom benchmark statistics with explicit return and sampling rules, then review and validate the generated Pine Script.

Explore Backtest Deep Report

Related resources

TradingView Backtesting Guides Backtest Accuracy Checks Set a Backtest Date Range Export TradingView Strategy Results

Frequently asked questions

Educational information, not investment advice. Numerical examples are hypothetical. Historical capture ratios depend on the selected benchmark and calculation method and do not guarantee future returns or protection from losses.

Sources and verification

Last verified: 2026-10-17

Published by Pineify. Pineify is an independent product and is not affiliated with TradingView. Sources support the technical explanations; examples do not establish trading performance.