SQN

What is the System Quality Number (SQN)?

The System Quality Number, usually shortened to SQN, is a single score for trading-system quality introduced by trading psychologist Van K. Tharp. It combines the average size of trade results, their consistency, and the number of trades into one figure: the mean R-multiple of the trades divided by the standard deviation of those R-multiples, scaled by the square root of the trade count. In plain terms it asks whether the average trade result is large and steady enough, over enough trades, to be distinguishable from noise.

The metric lives inside Tharp's R-multiple framework, where every trade's outcome is expressed as a multiple of its initial risk. That normalization is what makes SQN comparable across instruments and account sizes: a trade that risked one unit and made two counts the same whether the unit was fifty dollars or fifty thousand. Structurally the statistic resembles a t-score applied to trade outcomes, which is why more trades raise the SQN even when the per-trade edge is unchanged.

Tharp used the score both as a quality grade and as an input to position-sizing decisions, arguing that higher-quality systems can support more aggressive sizing. His rough grading bands treated scores below about 2 as below average, roughly 2 to 2.5 as average, 2.5 to 3 as good, 3 to 5 as excellent, and higher still as superb, though those labels describe backtest arithmetic, not a promise about the future.

How it's calculated

SQN is computed on the R-multiples of a set of closed trades.

Because sqrt(n) multiplies the score, some practitioners cap n at 100 when comparing systems with very different trade counts, a variant Tharp himself discussed.

R-multiples require a defined initial risk per trade; systems without explicit initial stops need a proxy risk definition before SQN is meaningful.

How traders use it

• As a one-number screen when comparing candidate systems in backtesting, before digging into equity curves, drawdown statistics, and trade-level detail.
• As a position-sizing input in Tharp's methodology, where a higher SQN is treated as license for somewhat more aggressive sizing, always bounded by ruin considerations.
• To monitor live performance drift: recomputing SQN on rolling windows of recent trades and comparing against the backtest band can flag degradation earlier than raw profit does.
• With awareness of its sample-size lever: doubling trade frequency mechanically inflates SQN via sqrt(n), so a high-frequency mediocre edge can outscore a low-frequency strong one. Reading it alongside expectancy keeps that honest.
• Not as a standalone verdict; SQN ignores drawdown sequencing, costs already have to be embedded in the R-multiples, and it inherits every bias present in the underlying backtest.

SQN vs related metrics