System Quality Number (SQN)

By Jarosław Wasiński • Published: Mar 3, 2026 • Updated: Oct 11, 2026

System Quality Number: commonly the mean trade result in R divided by its sample standard deviation, multiplied by the square root of the number of trades. Record the chosen convention. More observations can raise the statistic without proving a robust edge; dependence, selection bias and changing market conditions limit its interpretation.

Example.

For outcomes −1R, 0R, +1R and +2R, the mean is 0.5R. Using sample standard deviation s = √(5/3) ≈ 1.291R and n = 4 gives SQN = 2 × 0.5 / 1.291 ≈ 0.775. Four observations illustrate arithmetic rather than durable strategy quality.

Data convention

State whether outcomes are net, how initial R is defined and whether dispersion uses n−1. Implementations may use cash outcomes, a different dispersion convention or a trade-count cap. Reconcile conventions before comparing outputs.

Sample-size influence

The square-root factor increases the statistic for unchanged mean and dispersion. Duplicating a dependent sequence does not add independent information. SQN is not automatically a significance test or success probability.

Report alongside it

Include expectancy, R distribution, costs, drawdown, period and protected out-of-sample results. A mean driven by an extreme outcome needs sensitivity analysis. Zero dispersion does not produce an ordinary finite result; do not label it infinite quality.

Primary source checked 11 October 2026: www.backtrader.com.

FAQ

Does a high SQN justify greater leverage?

Not by itself. Sizing depends on capital, loss scenarios, limits and model uncertainty.

Is SQN the same as Sharpe?

No. Observation units, initial R and periods differ; Sharpe generally concerns excess-return series.

← Back to the full glossary